In this study, we propose an end-to-end emotion recognition system using an ear electroencephalogram (EEG)-based on-chip device that is enabled using the machine-learning model. The system has an integrated device that gathers EEG signals from electrodes positioned behind the ear; it is more practical than the conventional scalp-EEG method. The relative power spectral density (PSD), which is the feature used in this study, is derived using the fast Fourier transform over five frequency bands. Directly on the embedded device, data preprocessing and feature extraction were carried out. Three standard machine learning models, namely, support vector machine (SVM), multilayer perceptron (MLP), and Random Forest, were trained on these rich emotion classification features. The traditional approach, which integrates a model into the application software on a personal computer (PC), is cumbersome and lacks mobility, which makes it challenging to use in real-life applications. Besides, the PC-based system is not sufficiently real-time because of the connection latency from the EEG data acquisition device. To overcome these limitations, we propose a wearable device capable of performing on-chip machine learning and signal processing on the EEG data immediately after the acquisition task for the real-time result. In order to perform on-chip machine learning for the real-time prediction of emotions, Random Forest was chosen as a pre-trained model using the relative PSD characteristics as input based on the evaluation of the set results.

One of the forces driving science and industry is machine learning, but the proliferation of Big Data necessitates paradigm shifts from conventional approaches in applying machine learning techniques to this massive amount of data with varying velocity. Computers are now capable of accurately diagnosing a variety of medical conditions thanks to the availability of immense healthcare datasets and advancements in machine learning techniques. The study’s primary aim is to identify the most compelling questions on anxiety and depression in pregnant women by extracting features through performance-optimized algorithms. In this way, it is aimed to reach the result in a shorter time with fewer questions. The next goal of this work is to create an instant remote health status prediction system for depression and anxiety in pregnant women using machine learning models. In this scalable system, the application receives data from pregnant women to forecast the patient’s health condition. It then applies the Random Forest Classifier machine learning algorithm that produces the best results for this dataset with accuracy and precision 99% and 99% respectively. With this machine learning technique, the time-consuming anxiety and depression detection procedure in a pregnant woman can be replaced with a computer-based technique that works in an instant with a respectable amount of accuracy.

Near real-time ship monitoring is crucial for ensuring safety and security at sea. Established ship monitoring systems are the automatic identification system (AIS) and marine radars. However, not all ships are committed to carry an AIS transponder and the marine radars suffer from limited visibility. For these reasons, airborne radars can be used as an additional and supportive sensor for ship monitoring, especially on the open sea. State-of-the-art algorithms for ship detection in radar imagery are based on constant false alarm rate (CFAR). Such algorithms are pixel-based and therefore it can be challenging in practice to achieve near real-time detection. This letter presents two object-oriented ship detectors based on the faster region-based convolutional neural network (R-CNN). The first detector operates in time domain and the second detector operates in Doppler domain of airborne Range-Compressed (RC) radar data patches. The Faster R-CNN models are trained on thousands of real X-band airborne RC radar data patches containing several ship signals. The robustness of the proposed object-oriented ship detectors is tested on multiple scenarios, showing high recall performance of the models even in very dense multitarget scenarios in the complex inshore environment of the North Sea.

Fingerprint authentication techniques have been employed in various Internet of Things (IoT) applications for access control to protect private data, but raw fingerprint template leakage in unprotected IoT applications may render the authentication system insecure. Cancelable fingerprint templates can effectively prevent privacy breaches and provide strong protection to the original templates. However, to suit resource-constrained IoT devices, oversimplified templates would compromise authentication performance significantly. In addition, the length of existing cancelable fingerprint templates is usually fixed, making them difficult to be deployed in various memory-limited IoT devices. To address these issues, we propose a novel length-flexible lightweight cancelable fingerprint template for privacy-preserving authentication systems in various resource-constrained IoT applications. The proposed cancelable template design primarily consists of two components: 1) length-flexible partial-cancelable feature generation based on the designed reindexing scheme 2) lightweight cancelable feature generation based on the designed encoding nested difference XOR scheme. Comprehensive experimental results on public databases FVC2002 DB1–DB4 and FVC2004 DB1–DB4 demonstrate that the proposed cancelable fingerprint template achieves equivalent authentication performance to state-of-theart methods in IoT environments, but our design substantially reduces template storage space and computational cost. More importantly, the proposed length-flexible lightweight cancelable template is suitable for a variety of commercial smart cards (e.g., C5-M.O.S.T. Card Contact Microprocessor Smart Cards CLXSU064KC5). To the best of our knowledge, the proposed method is the first length-flexible lightweight, high-performing cancelable fingerprint template design for resource-constrained IoT applications.

Yoga has been a great form of physical activity and one of the promising applications in personal health care. Several studies prove that yoga is used as one of the physical treatments for cancer, musculoskeletal disorder, depression, Parkinson’s disease, and respiratory heart diseases. In yoga, the body should be mechanically aligned with some effort on the muscles, ligaments, and joints for optimal posture. Postural-based yoga increases flexibility, energy, overall brain activity and reduces stress, blood pressure, and back pain. Body Postural Alignment is a very important aspect while performing yogic asanas. Many yogic asanas including uttanasana, kurmasana, ustrasana, and dhanurasana, require bending forward or backward, and if the asanas are performed incorrectly, strain in the joints, ligaments, and backbone can result, which can cause problems with the hip joints. Hence it is vital to monitor the correct yoga poses while performing different asanas. Yoga posture prediction and automatic movement analysis are now possible because of advancements in computer vision algorithms and sensors. This research investigates a thorough analysis of yoga posture identification systems using computer vision, machine learning, and deep learning techniques.

Osteoporosis is a skeletal disease that is difficult to identify in advance of symptoms. Existing skeletal disease screening methods, such as dual-energy X-ray absorptiometry, are only used for specific purpose due to cost and safety reasons once symptoms develop. Early detection of osteopenia and osteoporosis using other modalities for relatively frequent examinations is helpful in terms of early treatment and cost. Recently, many studies have proposed deep learning-based osteoporosis diagnosis methods for various modalities and achieved outstanding results. However, these studies have limitations in clinical use because they require tedious processes, such as manually cropping a region of interest or diagnosing osteoporosis rather than osteopenia. In this study, we present a classification task for diagnosing osteopenia and osteoporosis using computed tomography (CT). Additionally, we propose a multi-view CT network (MVCTNet) that automatically classifies osteopenia and osteoporosis using two images from the original CT image. Unlike previous methods that use a single CT image as input, the MVCTNet captures various features from the images generated by our multi-view settings. The MVCTNet comprises two feature extractors and three task layers. Two feature extractors use the images as separate inputs and learn different features through dissimilarity loss. The target layers learn the target task through the features of the two feature extractors and then aggregate them. For the experiments, we use a dataset containing 2,883 patients’ CT images labeled as normal, osteopenia, and osteoporosis. Additionally, we observe that the proposed method improves the performance of all experiments based on the quantitative and qualitative evaluations

The speech and hearing-impaired community use sign language as the primary means of communication. It is quite challenging for the general population to interpret or learn sign language completely. A sign language recognition system must be designed and developed to address this communication barrier. Most current sign language recognition systems rely on wearable sensors, keeping the recognition system unaffordable for most individuals. Moreover, the existing vision-based sign recognition frameworks do not consider all of the spatial and temporal information required for accurate recognition. A novel vision-based hybrid deep neural net methodology is proposed in this study for recognizing American Sign Language (ASL) and custom sign gestures. The proposed framework aims to establish a single framework for tracking and extracting multi-semantic properties, such as non-manual components and manual co-articulations. Furthermore, spatial feature extraction from the sign gestures is deployed using a Hybrid Deep Neural Network (HDNN) with atrous convolutions. The temporal and sequential feature extraction is carried out by employing attention-based HDNN. In addition, the distinguished abstract feature extraction is done using modified autoencoders. The discriminative feature extraction for differentiating the sign gestures from unwanted transition gestures is done by leveraging the hybrid attention module. The experimentation of the proposed model has been carried out on the novel multi-signer ASL and custom sign language dataset. The proposed sign language recognition framework with hybrid neural nets, specifically using HDNN, yields better results than other state-of-the-art frameworks. Additionally, a detection module is incorporated using Flask web, allowing manual input of images/signs for real-time recognition.

The field of image classification and identification tasks has seen tremendous progress driven by the quick development of Convolutional Neural Network (CNN) methods. Unlike the widely used Vision Transformer (ViT) methods, this study presents an improved CNN-based model for pest recognition, segmentation, and classification. According to recent research, ViT is better at classifying images than standard machine learning and CNN methods. Taking this into consideration, we investigate how CNN models can incorporate two branch segment representations by using a double-layer CNN encoder. This new CNN-based method handles token chunks with different sizes and levels of computational complexity with ease. These elements are then combined with various attention processes to improve the overall aspects of the image. We use publicly accessible pest databases that impact peanut & other crops in our experiments. When compared to cutting-edge algorithms, the suggested CNN model shows unique characteristics and performs better in pest picture prediction, obtaining a high accuracy rate of 99.25%.

Monitoring nutritional values in food can help an individual in planning a healthy diet. In addition, regular dietary assessment can improve and maintain the physical and mental health of individuals. Recent advancement in computer vision using Deep Learning has enabled researchers to develop various techniques for automatic food nutrition estimation frameworks. Researchers have also contributed to prepare large food image datasets consisting of various food classes for this purpose. However, automatic estimation of nutritional values from food images still remains a challenging task. This review paper critically analyzes and summarizes existing methodologies and datasets used for automated estimation of nutritional value from food images. We first define the taxonomies in order to categorize the existing research works. Then, we study different methods to detect the food value estimation from food images in those categories. We have critically analyzed existing methods and compared the performance of various approaches for estimating food value using conventional performance metrics such as Accuracy, Error Rate, Intersection over Union (IoU), Sensitivity, Specificity, Precision, etc. In particular, we emphasize the current trends and techniques of Deep Learning-based approaches for food value estimation from images. Moreover, we have identified the ongoing challenges associated with automated food estimation systems and outlined the potential future directions. This review can immensely benefit researchers and practitioners, including computer scientists, health practitioners, and nutritionists.

Large blocks of data must be analyzed and explored by utilizing the data mining procedures in order to uncover significant patterns and trends. Medical databases are one area where the data mining procedures can be utilized. Many people all over the world are struggling with their health and medical diagnoses. Massive amounts of data are produced by hospital information systems (HIS), yet it might be difficult to extract knowledge from diagnosis case data. By just giving the symptoms they are experiencing, patients can quickly learn about the sickness they are experiencing and the medication that can assist, treat it using the approaches utilized in this paper. In this paper, we give drug recommendations relied on ratings and conditions to customers. To analyze the reviews and finally, probabilistic and weighted average methodologies are utilized to recommend the medications. Each model and strategy utilized in this paper is described in detail. The experimental findings presented in this work can be utilized in future studies and for a variety of different medicinal applications.

E-Learning Ecosystems (ELE) offer excellent opportunities to manage teaching activities by incorporating state-of-the-art technologies, practices, and professional support, as well as learning and assessment resources that can be adaptive. Therefore, it can help people with disabilities or conditions such as Autism Spectrum Disorder (ASD) to develop skills. However, some technological factors prevent this population’s implementation of support scenarios and hinder the proper learning process. This paper systematically reviews relevant studies on E-Learning Ecosystems for people with ASD, identifying the influence of Information and Communication Technologies (ICT) on forming ELE and the technological barriers that affect their development and appropriate use on people with ASD. This work conducted a systematic review using the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta- Analyses) methodology, including a search of five scientific literature databases from 2017 to 2022. The main aspects identified were 1) a shortage in design guides for the implementation of e-learning ecosystems adapted for people with ASD, 2) technological barriers that prevent the development of ELE, and 3) recommendations that help to mitigate the limitations of this field. In addition, the authors identified that the skills with the most significant focus of interest were social, communicative, and cognitive. The most implemented technologies include virtual and augmented reality or mobile applications. Most studies involved children with ASD between 8 and 15 years, followed by works with children between 5 to 8 years. Very few researches linked adults with ASD. Very few studies mention the ASD level of the participants, but most highlight the positive results of implementing ICT in training processes.

This study focuses on improving the accuracy of automated diagnoses in medical imaging, specifically endoscopic images, using advanced deep learning models. We aim to enhance the understanding of these models' decision-making processes by employing Explainable AI (XAI) techniques. Our research uses a subset of the Kvasir-capsule dataset, a Wireless Capsule Endoscopy imaging dataset, concentrating on the top 9 classes for training and testing. MobileNetv3Large is a convolutional neural network (CNN) architecture specifically designed for mobile and edge devices. It is an evolution of the MobileNet architecture, optimized for efficient and lightweight model inference. ResNet152v2 is part of the ResNet (Residual Network) family, known for its deep architecture with residual connections. ResNet152v2 is an extension with 152 layers. These scores surpass previous studies on the same dataset, indicating improved diagnostic reliability. To unravel the decision-making of these models, we utilized XAI techniques, including presenting heatmaps that highlight important regions in the images. This effort aims to provide insights into the top-performing deep learning models and demystify their complex decision processes. n summary, our research not only achieves improved diagnostic reliability but also contributes to the interpretability of deep learning models in the medical imaging domain. The use of Explainable AI techniques allows us to peek into the decision processes of these models, making them more transparent and trustworthy for medical professionals.

Realization of situation-awareness for autonomous robotics applications in edge computing environment is challenging. First, computing capabilities of edge devices are limited, which must be considered in the execution of machine learning (ML)-based solutions. Second, many technologies are available for realizing situation-aware capabilities, but comparison and integration of solutions creates additional challenges. Third, existing ML-based models are often not directly applicable for realizing custom applications, and model(s) may need to be re-trained with new data. The contribution of this paper is efficiency and feasibility evaluation of human pose recognition and object detection technologies in edge computing environment. Several lessons learnt covering constraints are presented regarding feasibility of the experimented technologies and data sets. The efficiency evaluation results indicated that simultaneous human pose recognition (Google’s Movenet) and object detection (Yolov5) on Jetson AGX Xavier achieved ∼13-16 FPS, while GPU and CPU utilization remained at a medium level, and most of the memory remained unused (< 44 %). Object concept and human pose concept activation algorithms may be considered as an additional contribution. Realized architecture design of the prototyped system in multiple computing environments can be considered as a partial evaluation of a ML-based big data reference architecture.

To solve real-life problems for different smart city applications, using deep Neural Network, such as parking occupancy detection, requires fine-tuning of these networks. For large parking, it is desirable to use a cenital plane camera located at a high distance that allows the monitoring of the entire parking space or a large parking area with only one camera. Today’s most popular object detection models, such as YOLO, achieve good precision scores at real-time speed. However, if we use our own data different from that of the general-purpose datasets, such as COCO and ImageNet, we have a large margin for improvisation. In this paper, we propose a modified, yet lightweight, deep object detection model based on the YOLO-v5 architecture. The proposed model can detect large, small, and tiny objects. Specifically, we propose the use of a multi-scale mechanism to learn deep discriminative feature representations at different scales and automatically determine the most suitable scales for detecting objects in a scene (i.e., in our case vehicles). The proposed multi-scale module reduces the number of trainable parameters compared to the original YOLO-v5 architecture. The experimental results also demonstrate that precision is improved by a large margin. In fact, as shown in the experiments, the results show a small reduction from 7.28 million parameters of the YOLO-v5-S profile to 7.26 million parameters in our model. In addition, we reduced the detection speed by inferring 30 fps compared to the YOLO-v5-L/X profiles. In addition, the tiny vehicle detection performance was significantly improved by 33% compared to the YOLO-v5-X profile.

Recent advancements in transformers exploited computer vision problems which results in state-of-the-art models. Transformer-based models in various sequence prediction tasks such as language translation, sentiment classification, and caption generation have shown remarkable performance. Auto report generation scenarios in medical imaging through caption generation models is one of the applied scenarios for language models and have strong social impact. In these models, convolution neural networks have been used as encoder to gain spatial information and recurrent neural networks are used as decoder to generate caption or medical report. However, using transformer architecture as encoder and decoder in caption or report writing task is still unexplored. In this research, we explored the effect of losing spatial biasness information in encoder by using pre-trained vanilla image transformer architecture and combine it with different pre-trained language transformers as decoder. In order to evaluate the proposed methodology, the Indiana University Chest X-Rays dataset is used where ablation study is also conducted with respect to different evaluations. The comparative analysis shows that the proposed methodology has represented remarkable performance when compared with existing techniques in terms of different performance parameters

As an interdisciplinary course, Machine Vision combines AI and digital image processing methods. This paper develops a comprehensive experiment on forest wildfire detection that organically integrates digital image processing, machine learning and deep learning technologies. Although the research on wildfire detection has made great progress, many experiments are not suitable for students to operate. Also, the detection with high accuracy is still a big challenge. In this paper, we divide the task of forest wildfire detection into two modules, which are wildfire image classification and wildfire region detection. We propose a novel wildfire image classification algorithm based on Reduce-VGG net, and a wildfire region detection algorithm based on the optimized CNN with the combination of spatial and temporal features. The experimental results show that the proposed Reduce-VGG Net model can reach 91.20% in accuracy, and the optimized CNN model with the combination of spatial and temporal features can reach 97.35% in accuracy. Our framework is a novel way to combine research and teaching. It can achieve good detection performance and can be used as a comprehensive experiment for Machine Vision course, which can provide the support for talent cultivation in machine vision area.

Heart rate arrhythmia is an important manifestation of common clinical cardiovascular diseases, posing a serious threat to human life and health. Due to its suddenness, insidiousness and rapid changes, it often causes patients to miss the best treatment time. Therefore, real-time monitoring of heart rate changes is particularly important to monitor and prevent the onset of such diseases. Nonetheless, heart rate information in telemedicine systems is often presented in plaintext, rendering it vulnerable to interception or tampering, and posing a substantial threat to users’ privacy. To address this issue, we propose a privacy-preserving remote heart rate abnormality monitoring system, which utilizes a privacy comparison protocol. By implementing a two-server model, the privacy comparison protocol ensures privacy not only during the comparison process but also in the resulting outcomes. During the monitoring process, the monitor is only able to obtain the final number of abnormalities of the patient’s heart rate and cannot obtain information about the patient’s original heart rate. This allows for a more rational and effective use of medical experts, so that patients can enjoy a high level and quality of service from medical experts without having to leave home. Finally, a detailed security analysis demonstrates that our scheme can effectively protect the privacy and security of patient medical data and hospital health indicators. And our experimental results show that our scheme is computationally efficient and the scheme is effective and feasible.

Intelligent detection of road cracks is crucial for road maintenance and safety. because of the interference of illumination and totally different background factors, the road crack extraction results of existing deep learning ways square measure incomplete, and therefore the extraction accuracy is low. we tend to designed a brand new network model, referred to as AR-UNet, that introduces a convolutional block attention module (CBAM) within the encoder and decoder of U-Net to effectively extract global and local detail information. The input and output CBAM features of the model are connected to increase the transmission path of features. The BasicBlock is adopted to replace the convolutional layer of the original network to avoid network degradation caused by gradient disappearance and network layer growth. we tested our method on DeepCrack, Crack Forest Dataset, and our own tagged road image dataset (RID). The experimental results show that our method focuses additional on crack feature info and extracts cracks with higher integrity. The comparison with existing deep learning ways conjointly demonstrates the effectiveness of our projected technique. The code is out there at: https://github.com/18435398440/ARUnet .

Addressing the inherent challenges of manufacturing data silos and credibility issues in traditional aviation supplier management, this paper introduces a pioneering blockchain-based platform designed for the seamless sharing of process quality data in the aviation industry. The proposed solution explores the integration of blockchain technology into manufacturing supply chain quality management, offering a novel approach to enhance transparency and reliability. The platform's architecture is structured to cater to the unique production processes of various aviation suppliers, emphasizing the quality state and diversity among suppliers. A comprehensive method for implementing secure quality and data sharing is outlined, ensuring real-time and organized platform operation. Key technologies, including manufacturing quality data block packaging models, secure data storage and sharing mechanisms, and supplier assessment models, are crucial components of this platform. The paper culminates in a practical application scenario within a specific aircraft industrial park, showcasing the platform's efficacy in collecting supplier product production data and providing intelligent solutions for airlines seeking reliable product quality data. This blockchain-based initiative stands as a transformative step towards fostering trust and efficiency in the aviation supply chain.

The relentless growth of data in healthcare has fueled the prominence of Big Data Analytics as a transformative force. This paper presents a comprehensive review of frameworks, implications, applications, and impacts of harnessing Big Data Analytics in healthcare. By assimilating data from diverse sources and leveraging advanced analytical techniques, elevating diagnostic precision, facilitating personalized medicine, and ultimately enhancing patient outcomes. Through an exhaustive analysis of scientific studies spanning the years 2013 to 2023, encompassing a total of 180 studies, this review establishes a robust foundation for future research endeavors and identifies collaborative opportunities within the healthcare domain. Examining the ecosystem, applications, and data sources, the study explores various application areas, emphasizing successful implementations and their capacity to improve healthcare outcomes while curbing costs. In addition to the paper delves into modeling tools, techniques, and real-world use cases, showcasing deployed solutions. By identifying key open research challenges, the study seeks to propel advancements in Big Data Analytics in healthcare, contributing to heightened healthcare outcomes and more informed decision-making processes.

Blockchain technology, introduced in 2008, has revolutionized the creation of distributed networks by enabling peer-to-peer connections based on consensus. This approach eliminates the need for central authorities or controllers, forming chains from accepted blocks. One significant application of blockchain is its utilization in storage systems, where individuals can rent out unused hardware space to others, fostering a decentralized network. This network, facilitated by end-to-end encryption, mitigates the risks associated with centralized data control, ensuring secure file transmission for clients. While numerous studies have focused on storage capacity and efficiency, addressing security, integrity, and privacy concerns is paramount. This paper provides an overview of blockchain-based storage systems, including their functionality and a comparative analysis with cloud-based storage networks. Additionally, the survey delves into various decentralized storage networks such as SIA, Filecoin, and Storj. We discuss the merits and drawbacks of blockchain-based storage solutions, culminating in an exploration of security challenges within decentralized storage networks. The paper concludes with an examination of potential solutions and proposes future research directions in this dynamic field.

In the evolving landscape where personal data holds increasing value, businesses seek insights through data processing, but this poses potential risks to individuals' privacy. While many companies collect consent for direct handling of personal data, a need for transparency and accountability arises to align data processing with obtained consent. This paper presents a novel Consent-Based Privacy-Compliant Personal Data-Sharing System, designed to enhance data quality while adhering to privacy regulations. The system addresses personal data-sharing flows and enterprise requirements, ensuring alignment with privacy frameworks. Through a comprehensive analysis of data-sharing processes and roles within enterprise environments based on established privacy frameworks, this paper outlines system requirements, architecture, and a detailed procedure for a consent-based privacy-compliant processing method, encompassing both compliance and consent checks. To validate the system's feasibility, a prototype is demonstrated, and performance analysis is conducted in both laboratory and real-world environments. This proposal aims to establish a robust framework for privacy-aware personal data sharing, fostering ethical and responsible practices in the evolving data-driven landscape.

In the ever-evolving landscape of cloud computing security, this paper introduces an innovative solution to address persistent challenges associated with safeguarding sensitive information. The pervasive use of cloud technology has amplified concerns regarding data security, necessitating novel approaches to fortify defenses against potential threats. The crux of the issue lies in the vulnerability of cloud-stored data, emphasizing the need for advanced encryption and key management strategies. Conventional methods often fall short in mitigating risks linked to compromised encryption keys and centralized key storage. To overcome these challenges, our proposed solution employs a two-phase approach. The initial phase focuses on the implementation of dynamic encryption keys, generating unique and ever-changing keys for each file. This approach significantly bolsters file-level security, restricting an attacker's ability to decrypt multiple files even if a key is compromised. The subsequent phase introduces blockchain technology for secure key storage, accompanied by metadata, to reinforce security and data integrity. In conclusion, our comprehensive approach enhances cloud security by providing robust encryption, decentralized key management, and protection against unauthorized access. The scalability and adaptability of our solution position it as a valuable asset in contemporary cloud security paradigms, ensuring users of reliable data security in the cloud.

In the rapidly evolving landscape of healthcare technology, the intersection of cloud computing, and blockchain has paved the way for transformative advancements in Electronic Medical Records (EMR) sharing. While existing approaches focus on securing outsourced encrypted EMRs in the cloud, there remains a critical oversight regarding the security and privacy of data collected by devices during transmission. This paper introduces a novel access control scheme, LightMED, specifically designed for Cloud-based EMR sharing. Leveraging fog computing blockchain technology, our proposed framework ensures secure, fine-grained, and scalable EMR sharing. We present a robust methodology for secure IoT data transmission and aggregation, incorporating lightweight encryption and digital signing. Key innovations include an outsourced encryption model, a privacy-preserving access policy scheme, and a collaborative encryption and decryption algorithm involving fog nodes and blockchain. Additionally, we introduce a lightweight policy update algorithm, empowering EMR data owners to efficiently manage access policies. Comparative analysis and experimental evaluations attest to the superior performance of our scheme, emphasizing its efficiency and practicality in safeguarding healthcare data in Fog-assisted IoT Cloud environments.

Attribute-based encryption is a valuable technique for ensuring data privacy and confidentiality in the realm of cloud computing. Using this cryptographic primitive, the data owners can securely store and share data within the cloud environment. On the other hand, in recent years, extensive advances have been made in quantum processors, which have raised hopes of solving certain mathematical problems includes factoring integers and computing discrete logarithms of large numbers. The advent of quantum computers has posed a significant security threat to existing cryptographic protocols. The existing post-quantum attributebased encryption schemes have not satisfied the essential features such as verifiability, user privacy and user revocability, simultaneously. In this paper, we present the first secure, practical and post-quantum attributebased encryption scheme based on rank metric codes. Our scheme enjoys all mentioned features due to utilization of low rank parity check codes. The proposed scheme provides security against chosen plaintext attacks in the standard model, as well as resistance against reaction attacks as a kind of chosen ciphertext attacks.

As one fundamental data-mining problem, ANN (approximate nearest neighbor search) is widely used in many industries including computer vision, information retrieval, and recommendation system. LSH (Local sensitive hashing) is one of the most popular hash-based approaches to solve ANN problems. However, the efficiency of operating LSH needs to be improved, as the operations of LSH often involve resource-consuming matrix operations and high-dimensional large-scale datasets. Meanwhile, for resource-constrained devices, this problem becomes more serious. One way to handle this problem is to outsource the heavy computing of high-dimensional large-scale data to cloud servers. However, when a cloud server responsible for computing tasks is untrustworthy, some security issues may arise. In this study, we proposed a cloud server-aided LSH scheme and the application model. This scheme can perform the LSH efficiently with the help of a cloud server and guarantee the privacy of the client’s information. And, in order to identify the improper behavior of the cloud server, we also provide a verification method to check the results returned from the cloud server. Meanwhile, for the implementation of this scheme on resourceconstrained devices, we proposed a model for the real application of this scheme. To verify the efficiency and correctness of the proposed scheme, theoretical analysis and experiments are conducted. The results of experiments and theoretical analysis indicate that the proposed scheme is correct, verifiable, secure and efficient.

Verifiable Searchable Symmetric Encryption, as an important cloud security technique, allows users to retrieve the encrypted data from the cloud through keywords and verify the validity of the returned results. Dynamic update for cloud data is one of the most common and fundamental requirements for data owners in such schemes. To the best of our knowledge, the existing verifiable SSE schemes supporting data dynamic update are all based on asymmetric-key cryptography verification, which involves time-consuming operations. The overhead of verification may become a significant burden due to the sheer amount of cloud data. Therefore, how to achieve keyword search over dynamic encrypted cloud data with efficient verification is a critical unsolved problem. To address this problem, we explore achieving keyword search over dynamic encrypted cloud data with symmetric-key based verification and propose a practical scheme in this paper. In order to support the efficient verification of dynamic data, we design a novel Accumulative Authentication Tag (AAT) based on the symmetric-key cryptography to generate an authentication tag for each keyword. Benefiting from the accumulation property of our designed AAT, the authentication tag can be conveniently updated when dynamic operations on cloud data occur. In order to achieve efficient data update, we design a new secure index composed by a search table ST based on the orthogonal list and a verification list VL containing AATs. Owing to the connectivity and the flexibility of ST, the update efficiency can be significantly improved. The security analysis and the performance evaluation results show that the proposed scheme is secure and efficient.

The key exposure is a serious threat for the security of data integrity auditing. Once the user’s private key for auditing is exposed, most of the existing data integrity auditing schemes would inevitably become unable to work. To deal with this problem, we construct a novel and efficient identity-based data integrity auditing scheme with key-exposure resilience for cloud storage. This is achieved by designing a novel key update technique, which is fully compatible with used in identity-based data integrity auditing. In our design, the Third Party Auditor (TPA) is responsible for generating update information. The user can update his private key based on the private key in one previous time period and the update information from the TPA. Furthermore, the proposed scheme supports real lazy update, which greatly improves the efficiency and the feasibility of key update. Meanwhile, the proposed scheme relies on identity-based cryptography, which makes certificate management easy. The security proof and the performance analysis demonstrate that the proposed scheme achieves desirable security and efficiency.

The rapid growth in technology and several devices make cyberspace unsecure and eventually lead to Significant Cyber Incidents (SCI). Cyber Security is a technique that protects systems over the internet from SCI. Data Mining and Machine Learning (DM-ML) play an important role in Cyber Security in the prediction, prevention, and detection of SCI. The dataset (SCI as per the report of the Center for Strategic and International Studies (CSIS)) is divided into two subsets (pre-pandemic SCI and post-pandemic SCI). Data Mining (DM) techniques are used for feature extraction and well know ML classifiers such as Naïve Bayes (NB), Support Vector Machine (SVM), Logistic Regression (LR) and Random Forest (RF) for classification. A centralized classifier approach is used to maintain a single centralized dataset by taking inputs from six continents of the world. The results of the pre-pandemic and post-pandemic datasets are compared and finally conclude this paper with better accuracy and the prediction of which type of SCI can occur in which part of the world. It is concluded that SVM and RF are much better classifiers than others and Asia is predicted to be the most affected continent by SCI.

Finding the right journal for a manuscript to be submitted is difficult and often time-consuming because authors take into account some criteria while searching for the appropriate journal for their manuscript. One of the most important criteria is the content similarity of the journals and manuscript. For this purpose, the subject of the manuscript should be in accordance with the scope of the journal. Also, the manuscript content should be closed to the journals’ trend for higher chance of acceptance. Second criterion is to take into account the impact-factor, acceptance-rate, review-time and publishing houses of the journal, which are suitable for the author’s past publication profile. In this study, a novel method is proposed in which both the content of the article and the author / authors profile are considered together to find the appropriate journal. To the best of our knowledge, this is the first effort in this direction. Experimental results conducted on real data sets have shown that the proposed method is applicable and performs high accuracy values.

The proposed study addresses the increasing mortality rate of breast cancer in females by introducing an automated disease diagnosis system. This system combines federated learning and deep learning to enhance efficiency and address the challenge of securely sharing sensitive medical images. The process involves image acquisition, encryption using an Image Encryption method, secure data storage using the Federated Learning Flower framework, and disease classification using the Deep Learning Neural Network model. The proposed system achieves high performance in terms of accuracy, recall, precision, F-measure, as demonstrated through simulation analysis using the BreakHis Database. The results show promising outcomes for automated breast cancer diagnosis with improved security and efficiency.

Traditional techniques for smart contract vulnerability detection rely on fixed expert criteria to discover vulnerabilities, which are less generalizable, scalable, and accurate. Deep learning algorithms help to address these issues, but most fail to encode true expert knowledge and remain interpretable. In this paper, we present a smart contract vulnerability detection mechanism that operates in phases with graph neural networks and expert patterns in deep learning to mutually address the deficiencies of the two detection approaches and improve smart contract vulnerability detection capabilities. Experiments show that our vulnerability detection mechanism outperforms the original deep learning model by an average of 6 points in detecting vulnerabilities and that the second stage of the checking mechanism can also block contract transactions containing dangerous actions at the Ethernet Virtual Machine (EVM) level and generate error reports for submission. This strategy helps to construct more stable smart contracts and to create a secure environment for smart contracts

Reversible data hiding in encrypted images (RDHEI) is an essential data security technique. Most RDHEI methods cannot perform well in embedding capacity and security. To address this issue, we propose a new RDHEI method using Chinese remainder theorem-based secret sharing (CRTSS) and hybrid coding. Specifically, a hybrid coding is first proposed for RDH to achieve high embedding capacity. At the content owner side, a novel iterative encryption is designed to conduct block based encryption for perfectly preserving the spatial correlation of original blocks in their encrypted blocks. Then, the CRTSS with the constraints is exploited to generate multiple encrypted image shares, in which spatial correlations of the encrypted blocks are also preserved. Meanwhile, the CRTSS provides good security properties for the proposed method. Since there are strong spatial correlations in the blocks of each share, the data-hider can exploit the proposed hybrid coding to perform data embedding for improving capacity. On the receiver side, even if some shares are corrupted/missing, the original image can be losslessly recovered as long as enough uncorrupted marked shares are obtained. Experiment results show that the proposed RDHEI method outperforms some state-of-the-art methods, including some secret sharing (SS) based methods in terms of embedding capacity.

Spelling correction is the process of finding the correct word for a misspelled word in a text. Any system aimed to fix this error cannot know the writer’s intent. But at the same time it should find the word that the user wanted to write. In this study, we trained a recurrent neural network with dictionary words and used as an oracle. For a misspelled word, this oracle returns a candidate dictionary word. Character level bigram model is used to generate new query words from a misspelled word. These new query words are also given to the trained network for getting more candidate dictionary words. For testing the method’s performance, randomly distorted dictionary words are used. Results showed that the trained network had an acceptable accuracy level. Also finding candidates using generated new query words have a positive impact on accuracy rather than using only misspelled word.

Enabling a flexible and natural human-robot interaction (HRI) for industrial robots is a critical yet challenging task that can be facilitated by the use of conversational artificial intelligence (AI). Prior research has concentrated on strengthening interactions through the deployment of social robots, while disregarding the capabilities required to boost the flexibility and user experience associated with human-robot collaboration (HRC) on manufacturing tasks. One of the main challenges is the lack of publicly available industrial-oriented dialogue datasets for the training of conversational AI. In this work, we present an Industrial Robot Wizard-of-Oz Dialoguing Dataset (IRWoZ) focused on enabling HRC in manufacturing tasks. The dataset covers four domains: assembly, transportation, position, and relocation. It is created using the Wizard-of-Oz technique to be less noisy. We manually constructed, annotated and validated dialogue segments (e.g., intentions, slots, annotations), as well as the responses. Building upon the proposed dataset, we benchmark it on the state-of-the-art (SoTA) language models, generative pretrained (GPT-2) models, on dialogue state tracking and response generation tasks.

Political speeches have played one of the most influential roles in shaping the world. Speeches of the written variety have been etched into history. These sorts of speeches have a great effect on the general people and their actions in the coming few days. Moreover, if left unchecked, political personnel or parties may cause major problems. In many cases, there may be a warning sign that the government needs to change its policies and also listen to the people. Understanding the emotion and context of a political speech is important, as they can be early indicators or warning signs for impending international crises, alignments, wars and future conflicts. In our research, we have focused on the presidents/prime ministers of China, Russia, the United Kingdom and the United States which are the permanent members of the United Nations Security Council and classified the speeches given by them based on the context and emotion of the speeches. The speeches were categorized into optimism, neutral, joy or upset in terms of emotion and five context categories, which are international affairs, nationalism, development, extremism and others. Here, optimism is a secondary emotion, whereas joy and upset are primary emotions. Apart from classifying the speeches based on context and emotion, one of the major works of our research is that we are introducing a dataset of political speeches that contains 2010 speeches labelled with emotion and context of the speech. The speeches we have worked on are large in word count. We propose EMPOLITICON-Context, a soft voting classifier ensemble learning model for context classification and EMPOLITICON-Emotion, a soft voting classifier ensemble learning model for emotion classification of political speeches. The proposed EMPOLITICON-Context model has achieved 73.13% accuracy in terms of context classification and the EMPOLITICON-Emotion model has achieved 53.07% accuracy in classifying the emotion of the political speeches

The popularity of social networks has only increased in recent years. In theory, the use of social media was proposed so we could share our views online, keep in contact with loved ones or share good moments of life. However, the reality is not so perfect, so you have people sharing hate speech-related messages, or using it to bully specific individuals, for instance, or even creating robots where their only goal is to target specific situations or people. Identifying what such text is not easy and there are several possible ways of doing it, such as using natural language processing or machine learning algorithms that can investigate and perform predictions using the metadata associated with it? In this work, we present an initial investigation of which is the best machine learning techniques to detect offensive language in tweets. After an analysis of the current trend in the literature about the recent text classification techniques, previously they have selected Linear SVM we have obtained 92% of accuracy and 95% and Naive Bayes algorithms, we have obtained 90% of accuracy and 93% for our initial tests. For the preprocessing of data, we have used different techniques for attribute selection that will be justified in the literature section. After our experiments, we have obtained 94% of accuracy and 95% of recall to detect offensive language with NLP with XGBoost (eXtreme Gradient Boosting).

Information and Communication Technologies have propelled social networking and communication, but cyber bullying poses significant challenges. Existing user-dependent mechanisms for reporting and blocking cyber bullying are manual and inefficient. Conventional Machine Learning and Transfer Learning approaches were explored for automatic cyber bullying detection. The study utilized a comprehensive dataset and structured annotation process. Textual, sentiment and emotional, static and contextual word embeddings, psycholinguistics, term lists, and toxicity features were employed in the Conventional Machine Learning approach. This research introduced the use of toxicity features for cyber bullying detection. Contextual embeddings of word Convolutional Neural Network (Word CNN) demonstrated comparable performance, with embeddings chosen for its higher F-measure. Textual features, embeddings, and toxicity features set new benchmarks when fed individually. The model achieved a boosted F-measure of 64.8% by combining textual, sentiment, embeddings, psycholinguistics, and toxicity features in a Logistic Regression model. This outperformed Linear SVC in terms of training time and handling high-dimensionality features. Transfer Learning utilized Word CNN for fine-tuning, achieving a faster training computation compared to the base models. Additionally, cyber bullying detection through Flask web was implemented, yielding an accuracy of 93%. The reference to the specific dataset name was omitted for privacy.

This project introduces a transformer-based model designed to generate music in a controllable manner, aiming for collaborative music composition with a computer. Leveraging a natural language processing model (GPT-2) as a foundation, the system utilizes the similarities between symbolic music representation and written language. The model can predict musical sequences based on conditional input, providing controllability without explicit programming or extensive retraining. A study involving 939 participants confirmed the effectiveness of this method for controlling music generation in its symbolic domain. The approach is adaptable to various controls, but the study particularly focuses on influencing the emotional aspects of the generated music. The goal was to confirm if the system is good at allowing people to control and influence the music it creates. The study aimed to understand if users could effectively use the system to produce the kind of music they wanted, especially focusing on the emotional feel of the music. The positive results from the study suggest that the proposed method is indeed effective and can be used to control the generation of music in its symbolic domain, which refers to the way music is represented in a computer system.

Emotion classification has become a valuable tool in analyzing text and emotions people express in response to events or crises, particularly on social media and other online platforms. The recent news about monkeypox highlighted various emotions individuals felt during the outbreak. People’s opinions and concerns have been very different based on their awareness and understanding of the disease. Although there have been studies on monkeypox, emotion classification related to this virus has not been considered. As a result, this study aims to analyze the emotions individual expressed on social media posts related to the monkeypox disease. Our goal is to provide real-time information and identify critical concerns about the disease. To conduct our analysis, first, we extract and preprocess 800,000 datasets and then use, a Python library, to predict and measure the emotional significance of each text. Secondly, we develop deep learning models based on Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), Bi-directional LSTM), and the combination of Convolutional Neural Networks and Long Short-Term Memory (CLSTM) for emotion classification. We use SMOTE (Synthetic Minority Oversampling Technique) and Random Under sampling techniques to address the class imbalance in our training dataset. The results of our study revealed that the CNN model achieved the highest performance with an accuracy of 96%. Overall, emotion classification on the monkeypox dataset can be a powerful tool for improving our understanding of the disease. The findings of this study will help develop effective interventions and improve public health.

Network attacks refer to malicious activities exploiting computer network vulnerabilities to compromise security, disrupt operations, or gain unauthorized access to sensitive information. Common network attacks include phishing, malware distribution, and brute-force attacks on network devices and user credentials. Such attacks can lead to financial losses due to downtime, recovery costs, and potential legal liabilities. To counter such threats, organizations use Intrusion Detection Systems (IDS) that leverage sophisticated algorithms and machine learning techniques to detect network attacks with enhanced accuracy and efficiency. Our proposed research aims to detect network attacks effectively and timely to prevent harmful losses. We built an advanced artificial intelligence-based machine learning methods. We propose a novel approach called Class Probability Random Forest (CPRF) for network attack detection performance enhancement. We created a novel feature set using the proposed CPRF approach. The CPRF approach predicts the class probabilities from the network attack dataset, which are then used as features for building applied machine learning methods. The comprehensive research results demonstrated that the random forest approach outperformed the state-of-the-art approach. The performance of each applied technique is validated using a k-fold approach and optimized with hyperparameter tuning. Our novel proposed research has revolutionized network attack detection, effectively preventing unauthorized access, service disruptions, sensitive information theft, and data integrity compromise.

In the current era of information explosion, users’ demand for data storage is increasing, and data on the cloud has become the first choice of users and enterprises. Cloud storage facilitates users to backup and share data, effectively reducing users’ storage expenses. As the duplicate data of different users are stored multiple times, leading to a sudden decrease in storage utilization of cloud servers. Data stored in plaintext form can directly remove duplicate data, while cloud servers are semi-trusted and usually need to store data after encryption to protect user privacy. In this paper, we focus on how to achieve secure de-duplication and recover data in cipher text for different users, and determine whether the indexes of public key searchable encryption and the matching relationship of trapdoor are equal in cipher text to achieve secure de-duplication. For the duplicate file, the data user’s re-encryption key about the file is appended to the cipher text chain table of the stored copy. The cloud server uses the re-encryption key to generate the specified transformed cipher text, and the data user decrypts the transformed cipher text by its private key to recover the file. The proposed scheme is secure and efficient through security analysis and experimental simulation analysis.

The safeguarding of digitized data against unwanted access and modification has become an issue of utmost importance as a direct result of the rapid development of network technology and internet applications. In response to this challenge, numerous secret image sharing (SIS) schemes have been developed. SIS is a method for protecting sensitive digital images from unauthorized access and alteration. The secret image is fragmented into a large number of arbitrary shares, each of which is designed to prevent the disclosure of any information to the trespassers. In this paper, we present a comprehensive survey of SIS schemes along with their pros and cons. We review various existing verifiable secret image sharing (VSIS) schemes that are immune to different types of cheating. We have identified various aspects of developing secure and efficient SIS schemes. In addition to that, a comparison and contrast of several SIS methodologies based on various properties is included in this survey work. We also highlight some of the applications based on SIS. Finally, we present open challenges and future directions in the field of SIS.

This paper proposes a lightweight image encryption approach for medical Internet of Things (MIoT) networks using compressive sensing and a modified seven-dimensional (MSD) hyper chaotic map. Initially, 7D hyper chaotic map is modified to generate more secure and complex secret keys. SHA-512 is used to create the initial conditions for MSD, which ensures its sensitivity towards input images. Using non sub sampled contourlet transform (NSCT), further improvements in the compressive sensing are achieved, and then the measurement matrices are generated using the secret keys obtained from MSD. Finally, to generate encrypted images, the diffusion and permutation are carried out row and column-wise on compressed images using secret keys obtained from MSD. The comparative analyses verify the performance of the proposed lightweight encryption approach in terms of robustness, security, and statistical analysis.

This paper introduces a novel approach to secure cloud-based identity management by proposing a Delegated and Verified Update Keys framework. The primary objective is to efficiently delegate the generation of update keys to a semi-trusted cloud server, ensuring the security of the overall system. We establish the security models for this framework and present a streamlined solution that integrates the widely adopted RSA algorithm. Our proposed methodology leverages the robustness of RSA, combined with a hierarchical structure, to enhance the security and efficiency of update key generation. We employ cryptographic primitives to facilitate secure delegation and verification processes, ensuring the integrity of the system. Additionally, a scheme for occasional base update key generation and periodic incremental update keys is introduced, aiming to optimize the update key size. The integration of the RSA algorithm into this framework not only strengthens the security foundations but also ensures compatibility with contemporary cryptographic standards. The presented approach offers a practical solution for secure and scalable identity management in cloud environments, emphasizing the delegation and verification of update keys as a crucial aspect of maintaining a robust and dynamic system.

In locations covered in debris, identifying catastrophe victims effectively requires making subtle decisions to manage the many obstacles that stand in the way of rescue operations. Innovative methods are required while working under pressure, sometimes with limited information and time limits. In order to improve decision-making when it comes to catastrophe victim detection, this study makes use of causal artificial intelligence. The goal is to provide interpretable and debatable knowledge by encoding causal assumptions similar to human intelligence. We provide a unique method that combines the advantages of boosting methods with decision trees with an AdaBoosted Random Forest model. This combination improves the precision and resilience of critical judgments on catastrophe victim identification. The suggested model produces dependable and understandable results by clarifying the causal links present in the data, fostering confidence in its suitability for high-pressure scenarios. The work shows how software agents may mimic human intelligence by using the causal model to capture cognitive comprehension through accurate and strong links. This invention has potential applications in disaster response situations, especially in the area of effective victim identification in areas with a lot of debris.

This research introduces an innovative approach to enhance the quality of nighttime roadside images, crucial for intelligent transportation systems. Current methods for improving low-light photos often result in color abnormalities and other issues. Our solution addresses these challenges by incorporating multiple sensors and techniques. Instead of conventional methods, we utilize a novel approach called bidirectional area segmentation-based inverse tone mapping to enhance photos. Additionally, we tackle the problem of moving objects appearing dim by employing a unique highlighting method based on precise identification of moving objects in the image data. Ultimately, we generate high-quality traffic photos using a pyramid-based fusion approach. Experiments with various images demonstrate that our technique outperforms existing methods in enhancing image details and creating more realistic colors for improved human observation.

In this research endeavor, we delve into the realm of fall risk prediction for the elderly, employing sophisticated machine learning techniques with a primary focus on the Logistic Regression algorithm. Our project, titled "Machine Learning Techniques Applied to the Development of a Fall Risk Index for Older Adults," seeks to address the critical issue of falls, a major cause of unintentional trauma-related deaths and a significant factor contributing to elderly dependency. The Logistic Regression algorithm takes center stage in our analysis due to its interpretability and efficacy in binary classification tasks. By training the model on our extensive dataset, we aim to predict and classify older adults into distinct categories – fallers and non-fallers. The algorithm's performance metrics, including accuracy, sensitivity, and specificity, undergo rigorous evaluation to ascertain its effectiveness in fall risk prediction. Our findings underscore the prowess of Logistic Regression in accurately predicting fall risk among older adults, with notable sensitivity and specificity. This research acts as a pivotal step towards establishing a nuanced Fall Risk Index, laying the groundwork for tailored fall prevention strategies. Beyond its immediate implications, this study paves the way for future initiatives, envisioning a dynamic and comprehensive Fall Risk Index. The overarching objective is to contribute to the development of proactive monitoring systems, enabling healthcare professionals to make informed decisions and implement targeted interventions, thereby enhancing the well-being of older adults. 

Date palm production is critical to oasis agriculture, owing to its economic importance and nutritional advantages. Numerous diseases endanger this precious tree, putting a strain on the economy and environment. White scale Parlatoria blanchardi is a damaging bug that degrades the quality of dates. When an infestation reaches a specific degree, it might result in the tree’s death. To counter this threat, precise detection of infected leaves and its infestation degree is important to decide if chemical treatment is necessary. This decision is crucial for farmers who wish to minimize yield losses while preserving production quality. For this purpose, we propose a feature extraction and VGG19 technique-based framework for classifying the stages of infestation by white scale disease (WSD) in date palm trees by investigating their leaflets images. To classify the WSD into its three classes (healthy, brown spots, and white scale), we use Convolutional Neural Network (CNN). The CNN model i.e., VGG19 was trained and evaluated using this dataset. The results indicate that the VGG19 achieve an accuracy of 99% during training and 94.8% under testing. The proposed framework could be beneficial to the oasis agricultural community in terms of early detection of date palm white scale disease (DPWSD) and assisting in the adoption of preventive measures to protect both date palm trees and crop yield.

In this research, we proposed a two-stage pipeline for segmenting urinary stones. The first stage U-Net generated the map localizing the urinary organs in full abdominal x-ray images. Then, this map was used for creating partitioned images input to the second stage U-Net to reduce class imbalance and was also used in stone-embedding augmentation to increase a number of training data. The U-Net model was trained with the combination of real stone-contained images and synthesized stone-embedded images to segment urinary stones on the partitioned input images. In addition, we proposed to use an inverse weighting method in the focal Tversky loss function in order to rebalance lesion size. The U-Net model using our proposed pipeline produced a 71.28% pixel-wise F2 score and a 69.82% region-wise F2 score, which were 2.88% and 7.63%, respectively, higher than those of a baseline method. Experimental results showed that the proposed method improved urinary stone segmentation results, especially for small stones and stones in uncommon locations.

Recent advancements in the fields of Machine Learning and Deep Learning made a huge transformation in other fields that are not related to Computer Science. In this work, a new framework is proposed to tackle the problem of translating the old Egyptian Hieroglyphic writings to English language through deploying both Image Processing combined with DL approaches. Our primary goal is to design an application that completely revolutionizes a tourist’s experience while navigating Egyptian Historical sites. This work utilizes different DL techniques to automatically convert the scanned photos of hieroglyphic language to understandable and readable English language, through two main sub-tasks: The automatic detection and recognizing of the scanned glyphs images and the translation of them into English language. Different data sources of this low-resource language were explored and augmented to train and test our models. Results of different models and algorithms are assessed and analyzed to evaluate our work. State-of-the-art results are achieved compared to literature in both automatic glyphs recognition, and glyphs-to-English translation.

Deep learning has become one of remote sensing scientists’ most efficient computer vision tools in recent years. However, the lack of training labels for the remote sensing datasets means that scientists need to solve the domain adaptation (DA) problem to narrow the discrepancy between satellite image datasets. As a result, image segmentation models that are then trained, could better generalize and use an existing set of labels instead of requiring new ones. This work proposes an unsupervised DA model that preserves semantic consistency and per-pixel quality for the images during the style-transferring phase. This article’s major contribution is proposing the improved architecture of the SemI2I model, which significantly boosts the proposed model’s performance and makes it competitive with the state-of-the-art CyCADA model. A second contribution is testing the CyCADA model on the remote sensing multiband datasets, such as WorldView-2 and SPOT-6. The proposed model preserves semantic consistency and per-pixel quality for the images during the style-transferring phase. Thus, the semantic segmentation model, trained on the adapted images, shows substantial performance gain compared to the SemI2I model and reaches similar results as the state-of-the-art CyCADA model. The future development of the proposed method could include ecological domain transfer, a priori evaluation of dataset quality in terms of data distribution, or exploration of the inner architecture of the DA model.

The paper describes implementation of a real-time visual tracking system equipped with an active camera. The system is intended for indoor human motion tracking. Real-time tracking is achieved using simple and fast motion detection procedures based on frame differencing and camera motion compensation. Results of on-line person tracking are presented. Based on preliminary results of object detection in each image which may have missing and/or false detection, the multiple object tracking method keeps a graph structure where it maintains multiply hypotheses about the number aid the trajectories of the object in the video. the image information drives the process of extending and pruning the graph, and determines the hest hypothesis to explain the video. While the image-hued object detection makes a local decision, the tracking process confirm and validates the detection through time, therefore, it can he regards as temporal detection which makes a global decision across time. The multiple object tracking method gives feedbacks which are predictions of object locations to the object detection module. Therefore, the method integrates object detection and trucking tightly. The most possible hypothesis provides the multiple object tracking result. The experimental results are presented.

Cancer is the second biggest cause of death worldwide, accounting for one of every six deaths. On the other hand, early detection of the disease significantly improves the chances of survival. The use of Artificial Intelligence (AI) to automate cancer detection might allow us to evaluate more cases in less time. In this research, AI-based deep learning models are proposed to classify the images of eight kinds of cancer, such as lung, brain, breast, and cervical cancer. This work evaluates the deep learning models, namely Convolutional Neural Networks (CNN), against classifying images with cancer traits. Pre-trained CNN variants such as MobileNet, VGGNet, and DenseNet are employed to transfer the knowledge they learned with the ImageNet dataset to detect different kinds of cancer cells. We use Bayesian Optimization to find the suitable values for the hyperparameters. However, transfer learning could make it so that model scan no longer classifies the datasets they were initially trained. So, we use Learning without Forgetting (LwF), which trains the network using only new task data while keeping the network’s original abilities. The results of the experiments show that the proposed models based on transfer learning are more accurate than the current state-of-the-art techniques. We also show that LwF can better classify both new datasets and datasets that have been trained before.

An interpretable deep learning framework for land use and land cover (LULC) classification in remote sensing using Shapley additive explanations (SHAPs) is introduced. It utilizes a compact convolutional neural network (CNN) model for the classification of satellite images and then feeds the results to a SHAP deep explainer so as to strengthen the classification results. The proposed framework is applied to Sentinel-2 satellite images containing 27 000 images of pixel size 64 × 64 and operates on three-band combinations, reducing the model’s input data by 77% considering that 13 channels are available, while at the same time investigating on how different spectrum bands affect predictions on the dataset’s classes. Experimental results on the EuroSAT dataset demonstrate the CNN’s accurate classification with an overall accuracy of 94.72%, whereas the classification accuracy on three-band combinations on each of the dataset’s classes highlights its improvement when compared to standard approaches with larger number of trainable parameters. The SHAP explainable results of the proposed framework shield the network’s predictions by showing correlation values that are relevant to the predicted class, thereby improving the classifications occurring in urban and rural areas with different land uses in the same scene.

Soil surface texture classification is a critical aspect of agriculture and soil science that affects various soil properties, such as water-holding capacity and soil nutrient retention. However, existing methods for soil texture classification rely on soil images taken under controlled conditions, which are not scalable for high spatiotemporal mapping of soil texture and fail to reflect real-world challenges and variations. To overcome these limitations, we propose a novel, scalable, and high spatial resolution soil surface texture classification process that employs image processing, texture-enhancing filters, and Convolutional Neural Network (CNN) to classify soil images captured under Uncontrolled Field Conditions (UFC). The proposed process involves a series of steps for improving soil image analysis. Initially, image segmentation is utilized to eliminate non-soil pixels and prepare the images for further processing. Next, the segmented output is divided into smaller tiles to isolate relevant soil pixels. Then, high-frequency filtering is introduced to enhance the texture of the images. Our research has shown that the Gabor filter is more effective than Local Binary Patterns (LBP) for this purpose. By creating four distinct Gabor filters, we can enhance specific, hidden patterns within the soil images. Finally, the split and enhanced images are used to train CNN classifiers for optimal analysis. We evaluate the performance of the proposed framework using different metrics and compare it to existing state-of-the-art soil texture classification frameworks. Our proposed soil texture classification process improves performance. We employed various CNN architectures in our proposed process for comparison purposes. Inception v3 produces the highest accuracy of 85.621%, an increase of 12% compared to other frameworks. With applications in precision agriculture, soil management, and environmental monitoring, the proposed novel methodology has the potential to offer a dependable and sustainable tool for classifying soil surface texture using low-cost ground imagery acquired under UFC.

The built year and structure of individual buildings are crucial factors for estimating and assessing potential earthquake and tsunami damage. Recent advances in sensing and analysis technologies allow the acquisition of high-resolution street view images (SVIs) that present new possibilities for research and development. In this study, we developed a model to estimate the built year and structure of a building using unidirectional SVIs captured using an onboard camera. We used geographic information system (GIS) building data and SVIs to generate an annotated built-year and structure dataset by developing a method to automatically combine the data with images of individual buildings cropped through object detection. Furthermore, we trained a deep learning model to classify the built year and structure of buildings using the annotated image dataset based on a deep convolutional neural network (DCNN) and a vision transformer (ViT). The results showed that SVI accurately predicts the built year and structure of individual buildings using ViT (overall accuracies for structure = 0.94 [three classes] and 0.96 [two classes] and for age = 0.68 [six classes] and 0.90 [three classes]). Compared with DCNN-based networks, the proposed Swim transformer based on ViT architectures effectively improves prediction accuracy. The results indicate that multiple high-resolution images can be obtained for individual buildings using SVI, and the proposed method is an effective approach for classifying structures and determining building age. The automatic, accurate, and large-scale mapping of the built year and structure of individual buildings can help develop specific disaster prevention measures.

Automated age estimation from face images is the process of assigning either an exact age or a specific age range to a facial image. In this paper a comparative study of the current techniques suitable for this task is performed, with an emphasis on lightweight models suitable for implementation. We investigate both the suitable modern deep learning architectures for feature extraction and the variants of framing the problem itself as either classification, regression or soft label classification. The models are evaluated on Audience dataset for age group classification and FG-NET dataset for exact age estimation. To gather in-depth insights into automated age estimation and in contrast to existing studies, we additionally compare the performance of both classification and regression on the same dataset. We propose a novel loss function that combines regression and classification approaches and show that it outperforms other considered approaches. At the same time, with a lightweight backbone, such architecture is suitable for implementation.

Acute lymphoblastic leukemia (ALL) is a type of leukemia cancer that arises due to the excessive growth of immature white blood cells (WBCs) in the bone marrow. The ALL rate for children and adults is nearly 80% and 40%, respectively. It affects the production of immature cells, leading to an abnormality of neurological cells and potential fatality. Therefore, a timely and accurate cancer diagnosis is important for effective treatment to improve survival rates. Since the image of acute lymphoblastic leukemia cells (cancer cells) under the microscope is complicated to recognize the difference between ALL cancer cells and normal cells. In order to reduce the severity of this disease, it is necessary to classify immature cells at an early stage. In recent years, different classification models have been introduced based on machine learning (ML) and deep learning (DL) algorithms, but they need to be improved to avoid issues related to poor generalization and slow convergence. This work enhances the diagnosis of ALL with a computer-aided system that yields accurate results by using DL techniques. This research study proposes a lightweight DL-assisted robust model based on EfficientNet-B3 using Depth wise separable convolutions for classifying acute lymphoblastic leukemia and normal cells in the white blood cell images dataset. The proposed lightweight EfficientNet-B3 uses less trainable parameters to enhance the performance and efficiency of the leukemia classification. Furthermore, two publicly available datasets are considered to evaluate the effectiveness and generalization of the proposed lightweight EfficientNet-B3. In addition, different measures are employed, such as accuracy, precision, recall, and f1-score, to evaluate the effectiveness of the proposed and baseline classifiers. In addition, a detailed analysis is given to evaluate and compare the performance and efficiency of the proposed with existing pre-trained and ensemble DL classifiers. Experimental results show that the proposed model for image classification achieves better performance and outperforms the existing benchmark DL and other ensemble classifiers. Moreover, our finding suggests that the proposed lightweight EfficientNet-B3 model is reliable and generalized to facilitate clinical research and practitioners for leukemia detection

Remote sensing images (RSIs) are characterized by complex spatial layouts and ground object structures. ViT can be a good choice for scene classification owing to the ability to capture long-range interactive information between patches of input images. However, due to the lack of some inductive biases inherent to CNNs, such as locality and translation equivariance, ViT cannot generalize well when trained on insufficient amounts of data. Compared with training ViT from scratch, transferring a large-scale pretrained one is more cost-efficient with better performance even when the target data are small scale. In addition, the cross-entropy (CE) loss is frequently utilized in scene classification yet has low robustness to noise labels and poor generalization performances for different scenes. In this article, a ViT-based model in combination with supervised contrastive learning (CL) is proposed, named ViT-CL. For CL, supervised contrastive (SupCon) loss, which is developed by extending the self-supervised contrastive approach to the fully supervised setting, can explore the label information of RSIs in embedding space and improve the robustness to common image corruption. In ViT-CL, a joint loss function that combines CE loss and SupCon loss is developed to prompt the model to learn more discriminative features. Also, a two-stage optimization framework is introduced to enhance the controllability of the optimization process of the ViT-CL model. Extensive experiments on the AID, NWPU-RESISC45, and UCM datasets verified the superior performance of ViT-CL, with the highest accuracies of 97.42%, 94.54%, and 99.76% among all competing methods, respectively.

Deep learning models need sufficient training samples to support them in the training process; otherwise, over fitting occurs, resulting in model failure. However, in the field of smart agriculture, there are common problems, such as difficulty in obtaining high-quality disease samples and high cost. To solve this problem, this paper proposed a high-quality image augmentation (HQIA) method for generating high-quality rice leaf disease images based on a dual generative adversarial network (GAN). First, the original samples were used to train Improved Training of Wasserstein GANs (WGAN-GP) to generate pseudo-data samples. The pseudo-data samples were put into the Optimized-Real-ESRGAN (Opt-Real-ESRGAN) to generate high-quality pseudo-data samples. Finally, the high-quality pseudo-data samples were put into the disease classification convolution neural network, and the effectiveness of the method was verified by indicators. Experimental results showed that this method can generate high-quality rice leaf disease images, and the recognition accuracy of high-quality rice disease image samples augmented by this method was 4.57% higher than that of using only the original training set on ResNet18 and 4.1% higher on VGG11. The results demonstrate the effectiveness of the proposed method with limited training datasets.

Accurate visual classification of bladder tissue during Trans-Urethral Resection of Bladder Tumor (TURBT) procedures is essential to improve early cancer diagnosis and treatment. During TURBT interventions, White Light Imaging (WLI) and Narrow Band Imaging (NBI) techniques are used for lesion detection. Each imaging technique provides diverse visual information that allows clinicians to identify and classify cancerous lesions. Computer vision methods that use both imaging techniques could improve endoscopic diagnosis. We address the challenge of tissue classification when annotations are available only in one domain, in our case WLI, and the endoscopic images correspond to an unpaired dataset, i.e. there is no exact equivalent for every image. Method: We propose a semisurprised Generative Adversarial Network (GAN)-based method composed of three main components: a teacher network trained on the labeled WLI data; a cycle-consistency GAN to perform unpaired image-to-image translation, and a multi-input student network. To ensure the quality of the synthetic images generated by the proposed GAN we perform a detailed quantitative, and qualitative analysis with the help of specialists. The overall average classification accuracy, precision, and recall obtainted with the proposed method for tissue classification are 0.90, 0.88, and 0.89 respectively, while the same metrics obtained in the unlabeled domain (NBI) are 0.92, 0.64, and 0.94 respectively. The quality of the generated images is reliable enough to deceive specialists. Significance: This study shows the potential of using semi-supervised GAN-based bladder tissue classification when annotations are limited in multi-domain data.

Space situational awareness (SSA) system requires recognition of space objects that are varied in sizes, shapes, and types. The space images are challenging because of several factors such as illumination and noise and thus make the recognition task complex. Image fusion is an important area in image processing for various applications including RGB-D sensor fusion, remote sensing, medical diagnostics, and infrared and visible image fusion. Recently, various image fusion algorithms have been developed and they showed a superior performance to explore more information that are not available in single images. In this paper, we compared various methods of RGB and Depth image fusion for space object classification task. The experiments were carried out, and the performance was evaluated using fusion performance metrics. It was found that the guided filter context enhancement (GFCE) outperformed other image fusion methods in terms of average gradient (8.2593), spatial frequency (28.4114), and entropy (6.9486). additionally, due to its ability to balance between good performance and inference speed (11.41 second), GFCE was selected for RGB and Depth image fusion stage before feature extraction and classification stage. The outcome of fusion method is fused images that were used to train a deep ensemble of CoAtNets to classify space objects into ten categories. The deep ensemble learning methods including bagging, boosting, and stacking were trained and evaluated for classification purposes. It was found that combination of fusion and stacking was able to improve classification accuracy.

In order to identify smoking behavior, this research presents a novel method that uses deep learning techniques more particularly, MobileNetV2 - to extract key characteristics from images. A conditional detection technique is used in the suggested approach to improve efficiency and simplify the model. Furthermore, Flask is used to provide a user interface that accepts manual input for picture analysis. In order to evaluate the method, 3,254 photographs of equal numbers of smokers and non-smokers in a variety of settings were used in the experiment. Both quantitative and qualitative evaluation indicators were used to gauge the technique's efficacy. On the dataset, the findings showed a classification accuracy of 97.12%. The user interface's integration of Flask offers a useful method for manual input, expanding the approach's practicality in real-world situations.

According to current data, sleepy drivers are responsible for an estimated 15.5% of fatal traffic accidents, which presents a serious threat to public safety. This research presents a unique solution to Driver Drowsiness Detection utilising Convolutional Neural Networks (CNN) combined with Flask for image/live input in order to handle this problem and take use of the extensive use of mobile devices. With this two-step technique, mobile devices in the car record and evaluate the driver's present state while protecting their privacy. As a decision-maker, the smart edge verifies drowsiness whether the data from the mobile client matches the real-time input that was observed. The suggested framework includes a data fusion technique and is centred on the distributed edge architecture, which guarantees effective administration of the area of interest. This method uses flask manual input or real-time input of the car environment and the CNN model to identify driver fatigue locally based on facial expressions. With an average accuracy of 97.7%, the framework's sleepiness detection performance is remarkable.

The offline signature verification system’s feature extraction stage is regarded as crucial and has a significant impact on how well these systems perform because the quantity and calibration of the features that are extracted determine how well these systems can distinguish between authentic and fake signatures. In this study, we introduced a hybrid method for extracting features from signature images, wherein a Convolutional Neural Network (CNN) and Histogram of Oriented Gradients (HOG) were used, to identify the key features. Finally, the CNN and HOG methods were combined. The experimental findings indicated that our suggested model executed satisfactorily in terms of efficiency and predictive ability, with accuracy of 99.98% on the ICDAR Dataset. This accuracy is deemed to be of high significance, particularly given that we checked skilled forged signatures that are more difficult to recognize than other forms of forged signatures like (simple or opposite).

Blood cancer, also known as leukemia, is a life-threatening disease that requires early and accurate diagnosis for effective treatment. In this project, we present an innovative approach for the detection and classification of blood cancer using deep learning techniques. Specifically, we employ the MobileNetV2 architecture implemented in Python to achieve remarkable results in terms of accuracy. The dataset utilized for this research is the "Blood Cells Cancer (ALL) dataset," comprising four distinct classes: Benign, [Malignant] early Pre-B, [Malignant] Pre-B, and [Malignant] Pro-B. This dataset contains a total of 3242 peripheral blood smear (PBS) images. Accurate diagnosis of Acute Lymphoblastic Leukemia (ALL) through PBS images is a critical step in the early screening of cancer cases. Our deep learning model, based on the MobileNetV2 architecture, has demonstrated exceptional performance. During training, it achieved an impressive accuracy of 98.00%, showcasing its ability to effectively distinguish between different types of blood cancer cells. Furthermore, the validation accuracy of 96.00% emphasizes the robustness and generalization capability of our model. This project not only highlights the potential of deep learning in the field of medical image analysis but also contributes to the early detection and classification of blood cancer, ultimately improving patient outcomes. The utilization of MobileNetV2 and Python makes our solution both efficient and accessible for healthcare professionals, paving the way for enhanced cancer screening and diagnosis.

Accurate identification of medications is crucial in order to prevent medical errors, as it directly impacts the well-being of patients and avoids significant consequences. The misuse of drugs poses a serious risk to patients, leading to potential harm and complications. This issue places a burden on healthcare professionals who must manually search through pill databases to identify medications when patients are unable to provide their prescription information. This situation arises frequently as patients often discard the containers containing their medication along with the prescription. To address these challenges, there is a pressing need to develop computerized medication systems that leverage information technology to accurately identify medications and detect potential interactions between them.This final year project introduces an innovative deep learning-based pill detection system with intelligent medicinal drug identification capabilities. The system is developed using Python programming language and utilizes the MobileNet architecture as the underlying model.The main objective of this project is to achieve accurate pill detection from images and provide intelligent identification of medicinal drugs. To accomplish this, the system is trained on a dataset consisting of 1,268 samples for both training and testing. The system's training process utilizes the MobileNet architecture, resulting in impressive performance metrics. The achieved training accuracy and validation accuracy are both reported at 98.00%. The high accuracy rates validate the system's ability to effectively detect pills and identify medicinal drugs with precision.In practice, this deep learning-based pill detection system offers significant advantages in the healthcare domain. It automates the pill identification process, reducing human error and saving valuable time for healthcare professionals. Patients can also benefit from the system, as it enables them to verify their prescribed medications and gain comprehensive information about their drugs.The system's evaluation includes rigorous testing on diverse pill images, ensuring its reliability, accuracy, and robustness. Through extensive experimentation and validation, the project demonstrates the effectiveness of the developed system in achieving accurate pill detection and intelligent medicinal drug identification.

In this project, we present an innovative approach to generate descriptive captions for images using deep learning techniques. The project is implemented in Python, utilizing the powerful combination of the ResNet50 architecture for image feature extraction and LSTM (Long Short-Term Memory) for caption generation. Our model achieved an impressive overall loss of 2.57, with an accuracy ranging between 67% to 70%.The project leverages the popular and widely used Flickr 8k Dataset, which consists of 8,000 images, each accompanied by human-annotated captions. This dataset provides a diverse and rich source of images and captions for training and evaluation purposes.To begin, we employ the ResNet50 architecture, a deep convolutional neural network, to extract high-level visual features from the input images. These features serve as a crucial foundation for understanding the content and context of the images.Next, we utilize LSTM, a type of recurrent neural network, to generate captions based on the extracted image features. LSTM models excel at capturing the sequential dependencies and linguistic structures necessary for generating coherent and contextually relevant captions.Throughout the project, we meticulously train and fine-tune the model using the Flickr 8k Dataset, ensuring that it learns to associate the visual features with the corresponding textual descriptions. We employ a combination of loss functions and optimization techniques to guide the learning process, aiming to minimize the overall loss and enhance the accuracy of caption generation.Through extensive experimentation and iterative refinement, our model achieves remarkable results, generating captions that accurately capture the essence of the images. The successful implementation of this project demonstrates the potential of deep learning in the domain of image caption generation. The generated captions have the potential to enhance accessibility, aid visually impaired individuals in perceiving visual content, and enrich the user experience in various applications, including content indexing and retrieval, social media platforms, and autonomous systems. By combining the ResNet50 architecture for image feature extraction with LSTM for caption generation, we achieve remarkable accuracy and overall loss metrics. This project contributes to the ever-evolving field of deep learning and opens avenues for further advancements in AI-driven image understanding and communication.

The advent of digital image manipulation tools has exacerbated the proliferation of image forgeries, necessitating robust solutions for their detection. This project presents a novel approach to address this challenge, utilizing Python and Convolutional Neural Network (CNN) model architecture.The CNN model, employed as the core of our forgery detection system, has exhibited remarkable performance. With a training accuracy of 98% and a validation accuracy of 92%, it showcases its efficacy in distinguishing authentic from tampered images. The dataset utilized in this study comprises 12,615 images, consisting of 7,492 authentic (real) images and 5,123 tampered (fake) images, providing a diverse and extensive testbed for evaluation.To enhance the precision of our approach, we incorporate Error Level Analysis (ELA) as a preprocessing step. Each image is resized to a standardized 256x256 resolution, after which ELA is applied. ELA aids in the identification of regions within an image that exhibit varying compression levels. In an untampered image, all regions should exhibit uniform compression. Deviations from this uniformity may indicate digital manipulation. The processed images are stored as numpy arrays for subsequent analysis.Our proposed system leverages the synergy between deep learning through CNNs and the subtleties uncovered by ELA. This combination empowers the model to not only achieve high accuracy but also to provide insights into the specific regions of potential manipulation within an image. By harnessing the capabilities of Python and a well-structured CNN architecture, this project represents a significant stride towards robust digital image forgery detection, with potential applications in various domains where image authenticity is paramount.

The identification of bird species plays a vital role in various domains, including wildlife conservation, ecological research, and biodiversity monitoring. However, manual identification of bird species from images can be a time-consuming and error-prone task, especially with the vast number of avian species present worldwide.The project "Image-Based Bird Species Identification Using Deep Learning" presents a novel and highly accurate approach for automatically classifying bird species from images using the powerful Xception architecture. Developed entirely in Python, this project aims to address the challenging task of recognizing a diverse range of bird species with high precision.The project addresses a critical need in the field of ornithology and computer vision.The core of the system lies in the utilization of the Xception deep learning model, renowned for its exceptional ability to extract intricate features from images, enabling it to capture fine-grained details that are crucial for accurate bird species identification. Through meticulous training and optimization, the model has achieved an impressive training accuracy of 99% and a validation accuracy of 97%, showcasing its efficacy in handling complex classification tasks.The project's success is further bolstered by the extensive dataset it employs, comprising a comprehensive collection of 60,388 bird images spanning 510 distinct species. This dataset diversity allows the model to learn from a vast array of avian features, ensuring robust performance even when faced with previously unseen species.The proposed image-based bird species identification system finds valuable applications in wildlife monitoring, ecological research, and birdwatching enthusiasts, among others. By leveraging the power of deep learning and the Xception architecture, it sets new benchmarks in the realm of bird species recognition, making it a valuable contribution to the field of computer vision and ornithology. This project demonstrates an exceptional solution to the challenge of bird species identification, outperforming conventional methods and opening avenues for further research and application. Its high accuracy, efficient implementation in Python, and use of the Xception architecture position it as a pioneering and unique endeavor in the realm of image-based bird species classification.

In recent years, there has been a growing interest in the identification and classification of medicinal plants due to their potential health benefits. This project presents an innovative AI-based approach for advancing medicinal plant identification using deep learning techniques, specifically employing the Xception architecture. Developed using Python, our model achieves remarkable training accuracy of 93.34% and validation accuracy of 96.79%.To train and evaluate the model, we utilized the VNPlant-200 dataset, consisting of a comprehensive collection of 17,973 images of medicinal plants distributed among 200 distinct categories. This dataset encompasses a wide variety of plant species with diverse visual characteristics, enabling robust and accurate plant identification.Through a meticulous training process, the Xception-based model learns intricate patterns and features within the images, enabling it to effectively distinguish between different medicinal plant species. Leveraging the power of deep learning, our approach significantly enhances the accuracy and efficiency of medicinal plant identification.Additionally, hyperparameter tuning and fine-tuning of the Xception architecture were performed to optimize the model's performance and achieve exceptional accuracy.The results obtained demonstrate the efficacy of our AI-based approach for medicinal plant identification. The high training and validation accuracies validate the model's capability to accurately recognize and categorize medicinal plant species. This project contributes to the advancement of automated identification systems in the field of herbal medicine, enabling researchers, botanists, and healthcare professionals to rapidly and reliably identify medicinal plants for various purposes. Overall, this project showcases the potential of AI and deep learning techniques, specifically the Xception architecture, in advancing medicinal plant identification. The successful application of our approach on the VNPlant-200 dataset opens up new possibilities for further research and development in this domain, fostering advancements in herbal medicine and botanical studies.

Monkeypox is an infectious disease caused by the monkeypox virus (MPXV) that primarily affects animals but can be transmitted to humans, resulting in serious health implications. Early detection and accurate diagnosis of the disease are crucial for effective containment and treatment. Controlling the rapid transmission of the disease necessitates timely and accurate diagnosis, but the availability of traditional confirmatory tests, such as Polymerase Chain Reaction (PCR), remains limited. In this challenging scenario, by leveraging advanced technology, this approach can present a valuable alternative to conventional testing methods. To address this issue, we present a novel web application developed using Python's Flask framework, employing deep learning techniques for monkeypox disease detection. In this project, we develop a novel approach for the detection of Monkeypox disease using Deep Learning, specifically leveraging the ResNet50V2 architecture.The core of our approach relies on the ResNet50V2 architecture, which has demonstrated impressive performance in medical image analysis tasks. The model was trained on the Monkeypox Skin Lesion Dataset (MSLD), consisting of 1428 images labeled as 'Monkeypox' and 1764 images labeled as 'Others,' encompassing various skin lesion types.Through rigorous training and validation, our deep learning model achieved remarkable results, with a training accuracy of 93.00% and a validation accuracy of 92.00%. The easy-to-use web application allows users to upload skin lesion images, which are then analyzed by our trained model to provide rapid identification of monkeypox.By providing a consumer-level software solution, our project aims to empower individuals and healthcare practitioners with a reliable tool for early detection of monkeypox, facilitating prompt action and containment efforts. This system can play a pivotal role in mitigating the impact of the ongoing monkeypox outbreak and enhancing global public health preparedness in the face of infectious diseases.

Pancreatic cancer is a highly lethal disease that demands early detection and accurate classification for improved patient outcomes. This project leverages the power of Python-based Machine Learning and Deep Learning techniques to address this critical medical challenge. In the Machine Learning phase, two powerful algorithms, Random Forest Classifier and Naive Bayes, were employed. The Random Forest Classifier achieved an impressive Accuracy Train score of 100% and a remarkable test score of 99.2%. Similarly, Naive Bayes demonstrated robust performance with an Accuracy Train score of 99.3% and a test score of 99.2%. The dataset utilized in this phase is "Urinary biomarkers for pancreatic cancer," comprising 590 records with three distinct classes: Control, Benign, and PDAC. Key features of this dataset include creatinine, LYVE1, REG1B, and TFF1. Creatinine serves as an indicator of kidney function, LYVE1's potential role in tumor metastasis is explored, REG1B's association with pancreas regeneration is investigated, and TFF1's relevance to urinary tract repair is studied. The Deep Learning component of this project utilized Convolutional Neural Network (CNN) architecture. The CNN model exhibited excellent results with a Training accuracy of 98.7% and a Validation accuracy of 100%. The dataset in this phase comprised 1411 images categorized into two classes: normal and pancreatic tumor. This deep learning approach contributes to the project's robustness and complements the machine learning results. The combination of Machine Learning and Deep Learning techniques with a focus on urinary biomarkers and imaging analysis provides a comprehensive solution for the detection and classification of pancreatic cancer. This project demonstrates exceptional accuracy, setting the stage for potentially transformative applications in the field of medical diagnosis and early cancer detection.

The project "Traffic Sign Classification using Deep Learning" represents a significant advancement in the field of computer vision, specifically focusing on the recognition and classification of traffic signs. Leveraging the power of Python, two distinctive models were employed to address the complex challenges associated with traffic sign classification: MobileNet Architecture and YOLOv5. With MobileNet Architecture, an impressive level of performance was achieved, with a Training Accuracy of 97.00% and Validation Accuracy of 98.00%. This achievement was realized through the utilization of a meticulously curated dataset comprising 4,170 images encompassing a diverse array of 58 traffic sign classes, including but not limited to speed limits, directional instructions, prohibitory signs, and hazard warnings. These classes span the entire spectrum of traffic regulation, ensuring comprehensive coverage of the subject matter. Moving forward, the implementation of YOLOv5 introduced real-time traffic sign recognition using image data and real time web camera data. This model was trained on a dataset comprising 39 unique traffic sign classes. These classes encompass a wide range of signs, such as pedestrian crossings, speed limits, warnings, and regulatory signs, contributing to the project's practical applicability in real-world scenarios. The project represents a notable contribution to the field of deep learning-based traffic sign classification. By employing two distinct architectures, it ensures both high accuracy and real-time capability, addressing the growing demand for intelligent traffic sign recognition systems. The results showcase the feasibility of employing deep learning techniques to enhance road safety and traffic management.

Phishing attacks are a type of cybercrime that has grown in recent years. It is part of social engineering attacks where an attacker deceives users by sending fake messages using social media platforms or emails. Phishing attacks steal users’ information or download and install malicious software. They are hard to detect because attackers can design a phishing message that looks legitimate to a user. This message may contain a phishing URL so that even an expert can be a victim. This URL leads the victim to a fake website that steals information, such as login information, payment information, etc. Researchers and engineers work to develop methods to detect phishing attacks without the need for the eyes of experts. Even though many papers discuss HTML and URL-based phishing detection methods, there is no comprehensive survey to discuss these methods. Therefore, this paper comprehensively surveys HTML and URL phishing attacks and detection methods. We review the current state-of-art machine learning models to detect URL-based and hybrid-based phishing attacks in detail. We compare each model based on its data preprocessing, feature extraction, model design, and performance.

Education is very important for students’ future success. The performance of students can be supported by the extra assignments and projects given by the instructors for students with low performance. However, a major problem is that students at-risk cannot be identified early. This situation is being investigated by various researchers using Machine Learning techniques. Machine learning is used in a variety of areas and has also begun to be used to identify students at-risk early and to provide support by instructors. This research paper discusses the performance results found using Machine learning algorithms to identify at-risk students and minimize student failure. The main purpose of this project is to create a hybrid model using the ensemble stacking method and to predict at-risk students using this model. We used machine learning algorithms such as Random Forest, Decision Tree and Logistic Regression in this project. The performance of each machine learning algorithm presented in the project was measured with various metrics. Thus, the hybrid model by combining algorithms that give the best prediction results is presented in this study. The data set containing the demographic and academic information of the students was used to train and test the model. In addition, a web application developed for the effective use of the hybrid model and for obtaining prediction results is presented in the report. In the proposed method, it has been realized that stratified k-fold cross validation and hyperparameter optimization techniques increased the performance of the models. The hybrid ensemble model was tested with a combination of two different datasets to understand the importance of the data features. The accuracy of the hybrid model was obtained as 94.8%. This study focuses on predicting the performance of at-risk students early. Thus, teachers will be able to provide extra assistance to students with low performance.

A high level of computation is required for edge detection in color images captured by unmanned aerial vehicles (UAVs) to address issues, such as noise, distortion, and information loss. Thus, an edge detection method for UAV-captured color images based on the improved whale optimization algorithm (WOA) is proposed in this study. In this method, the color image pixels are represented by quaternions, and the global random position variables and information exchange mechanism are introduced into the random walk foraging formula of the WOA. Further, a random disturbance factor is also introduced into the predator-prey formula of the spiral bubble net. The proposed improved WOA is then used to obtain the preliminary edge of the UAV-captured color image. An edge-point classification method using the radius of the shortest distance between the whale and the current global optimum in each iteration is presented to enhance a preliminary edge. The experimental results show that the proposed edge detection method has the advantages of strong denoising, fast speed, and good quality.

Travel mode choice prediction is critical for travel demand prediction, which influences transport resource allocation and transport policies. Travel modes are often characterized by severe class imbalance and inequality, which leads to the inferior predictive performance of minority modes and bias in travel demand prediction. In existing studies, the class imbalance in travel mode prediction has not been addressed with a general approach. Basic resembling methods were adopted without much investigation, and the performance was assessed by commonly used metrics (e.g., accuracy), which is not suitable for predicting highly imbalanced modes. To this end, this paper proposes an evaluation framework to systematically investigate the combination of six over/under sampling techniques and three prediction methods. In a case study using the London Passenger Mode Choice dataset, results show that applying over/under sampling techniques on travel mode substantially improves the F1 score (i.e., the harmonic mean of precision and recall) of minority classes, without considerably downgrading the overall prediction performance or model interpretation. These findings suggest that combining over/under sampling techniques and statistical/machine-learning methods is appropriate for predicting travel mode, which effectively mitigates the influence of class imbalance while achieving high predictive accuracy and model interpretation. In addition, the combination of over/under sampling techniques and prediction methods enriches the model options for predicting mode choice, which would better support transport planning.

Measurement of e-commerce usability based on static quantities variable is state-of-theart because of the adoption of sequential tracing of the next phase in the categorical data. The global COVID-19 outbreak has completely disrupted society and drastically altered daily life. The concept refers to an electronic commerce network that appears with thorough, understandable conviction, demand, and rapid confirmation as a replacement for the economical market’s ‘‘brick-and-mortar’’ model, which replaces how we do everything, including business strategy, and provides a better understanding with the interpretation of e-commerce features. This study was supervised to analyze usability assessments using statistical methods and security assessments using online e-commerce security scanner tools to investigate e-business standards that consider the caliber of e-services in e-commerce websites across Asian nations. The method was developed to optimize complex systems based on multiple criteria. The initial (supplied) weights are used to determine the compromise ranking list and compromise solution. This paper examines the usability of e-commerce in rural areas using a new data set from the Jharkhand region. On the e-commerce websites of Jharkhand, India, usability is commonly considered in conjunction with learn ability, memorability, effectiveness, engagement, efficiency, and completeness. Using a user-oriented questionnaire testing method, this survey attempts to close the gaps mentioned above. Then, across each column, divide each value by the column-wise sum that is created using their corresponding value, whichever produces a new matrix B. Finally, determine the row-wise sum of matrix B representing the (3 × 1) matrix. Using model trees and bagging, this study addresses classification-related issues.

The fast increase in electric vehicle (EV) usage in the last 10 years has raised the need to properly forecast their energy consumption during charge. Lithium-ion batteries have become the major storage component for electric vehicles, avoiding their overcharge can preserve their health and prolong their lifetime. This paper proposes a Machine Learning model based on the K-Nearest Neighbors classification algorithm for EV charging session duration forecast. The model forecasts the duration of the charge by assigning the event to its correct class. Each class contains the charging events whose duration is comprised of a certain interval. The only information used by the algorithm is the one available at the beginning of the charging event (arrival time, starting SOC, calendar data). The model is validated on a real-world dataset containing records of charging sessions from more than 100 users, a sensitivity analysis is performed to assess the impact of different information given as input. The effectiveness of the model with respect to the benchmark models is demonstrated with an increase in performance

Artificial Intelligence (AI) and Machine Learning stand as pivotal technological advancements, reshaping diverse domains such as computing, finance, healthcare, agriculture, music, space, and tourism. This study focuses on the intricate realm of audio analysis within AI, encompassing music information retrieval, generation, and classification. Extracting meaningful features from music data poses a significant challenge, leading to the exploration of various algorithms, including classical and hybrid neural networks. Our investigation revolves around music genre classification, emphasizing the effectiveness of a Multi-Layer Perceptron (MLP) compared to traditional methods like Support Vector Classifier (SVC), Logistic Regression, and others. The evaluation utilizes a subset of the Free Music Archive (FMA) dataset, with the proposed MLP achieving an accuracy of 99.55% and a validation accuracy of 92.21%. The baseline model, SVC, is replaced by MLP, showcasing its superior performance. The study observes that image-based features outperform conventional audio-extracted features in label classification. Integration of Flask web enables real-time music genre detection. This paper provides an in-depth exploration of music genre classification methods, emphasizing the efficacy of MLP and introducing a web-based detection approach.

Stroke, a critical medical condition stemming from disrupted blood flow to the brain, poses a significant global threat with substantial health and economic ramifications. Researchers are addressing this challenge by developing automated stroke prediction algorithms for timely intervention and potential life-saving measures. As the population at risk continues to rise with aging demographics, the demand for precise and effective prediction systems becomes increasingly imperative. In this study, we assess the efficacy of a proposed machine learning (ML) technique, specifically XGBoost and a fusion of XGBoost with Random Forest, through a comparative analysis with six established classifiers. Evaluation metrics pertaining to both generalization capability and prediction accuracy were considered. The experimental results demonstrate that more intricate models outperform simpler ones, with the top model achieving an accuracy of 96%, while other models range from 84-96% accuracy. The proposed framework, incorporating global and local explainable methodologies, offers a standardized approach for comprehending complex models, thereby enhancing insight into decision-making processes crucial for advancing stroke care and treatment. Furthermore, we propose extending the model for web-based stroke detection, broadening its potential impact on public health.

The increasing popularity of wearable devices like smart watches, smart phones, and wristbands has generated a need to analyze user patterns and activity relationships. This study utilizes the MHHBA Physical Activity Monitoring dataset, consisting of data from 12 different physical activities performed by 10 subjects wearing 3 inertial measurement units and a heart rate monitor. The dataset is invaluable for activity recognition, intensity estimation, and the development of algorithms for data processing, segmentation, feature extraction, and classification. To enhance accessibility, the original data has been transformed and merging all the datasets into a comprehensive CSV file. Leveraging this enriched dataset, we apply machine learning techniques to accurately predict the specific activity a user is engaged in. We employ the K-Nearest Neighbors (KNN) algorithm, Random Forest algorithm, Grid Search CV for best n_neighbours and Robust Scaler to standardize the features.

The project, "CO2 Emission Rating by Vehicles Using Data Science," is a data-driven initiative aimed at assessing and rating the carbon dioxide (CO2) emissions of new light-duty vehicles available for retail sale in Canada in 2022. Leveraging the power of Python programming and employing sophisticated machine learning models, namely the Random Forest Classifier and the Decision Tree Classifier, this project offers a comprehensive analysis of vehicle emissions.The dataset utilized for this project contains crucial information, including fuel consumption ratings, CO2 emissions in grams per kilometer, CO2 ratings on a scale from 1 (worst) to 10 (best), and smog ratings on a scale from 1 (worst) to 10 (best). These data elements provide a holistic perspective on the environmental performance of various vehicle models, allowing consumers and policymakers to make informed choices.The Random Forest Classifier, a powerful ensemble learning algorithm, and the Decision Tree Classifier were employed to build predictive models. These models achieved remarkable accuracy scores, with the Random Forest Classifier achieving a 100% accuracy on the training dataset and an impressive 99% accuracy on the test dataset. Similarly, the Decision Tree Classifier exhibited outstanding performance with a 100% training accuracy and a 98% test accuracy.By combining these advanced algorithms and a rich dataset, this project contributes to sustainable transportation solutions and empowers consumers to make environmentally conscious decisions when purchasing vehicles. The CO2 Emission Rating system developed here serves as a valuable tool for evaluating the environmental impact of different vehicle models, helping reduce carbon emissions and mitigate climate change.In summary, "CO2 Emission Rating by Vehicles Using Data Science" is a pioneering project that demonstrates the potential of data science and machine learning to address critical environmental challenges. It underscores the importance of transparency and informed decision-making in the automotive industry, ultimately promoting a greener and more sustainable future.

In an age where social media has become an integral part of our lives, the challenge of detecting fake accounts on platforms like Instagram has gained significant importance. This project, titled "Instagram Fake Account Detection using Machine Learning" employs Python as its primary tool to tackle this problem. It leverages two powerful machine learning algorithms, the Random Forest Classifier and the Decision Tree Classifier, to accomplish this task. The Random Forest Classifier demonstrates remarkable performance, achieving a 100% accuracy on the training dataset and an impressive 93% accuracy on the test dataset. Meanwhile, the Decision Tree Classifier exhibits its effectiveness with a training accuracy of 92% and a test accuracy of 92%. The dataset employed in this project is composed of 576 records, each characterized by 12 distinct features. These features encompass critical aspects of Instagram profiles, including the presence of a profile picture, the ratio of numerical characters in usernames, the breakdown of full names into word tokens, the ratio of numerical characters in full names, the equality between usernames and full names, the length of user bios, the existence of external URLs, the privacy status of accounts, the number of posts, the count of followers, the number of accounts followed, and the ultimate classification of an account as "Fake" or "Not." By harnessing the capabilities of Python and these advanced machine learning models, this project endeavors to provide a robust and efficient solution for the identification of fake Instagram accounts. In doing so, it contributes to the preservation of the platform's integrity and the security of its users.

"Crime Prediction Using Machine Learning" is a comprehensive project developed in Python that employs Machine Learning algorithms, specifically the Decision Tree Classifier and Bagging Classifier, to predict and classify various crime categories in Portland, Oregon, USA, from the years 2015 to 2023. The dataset utilized for this project consists of 505,063 data points, with a focus on 20 distinct crime classes, including 'Larceny Offenses,' 'Motor Vehicle Theft,' 'Assault Offenses,' 'Drug/Narcotic Offenses,' and others. The Decision Tree Classifier yielded impressive results, achieving a 98% accuracy on the training set and a 95% accuracy on the test set. Similarly, the Bagging Classifier demonstrated robust performance, achieving a 98% accuracy on the training set and maintaining a 95% accuracy on the test set. These high accuracies indicate the effectiveness of the machine learning models in predicting and classifying crimes. The dataset encompasses 15 features, including address, case number, crime against category (Person, Property, or Society), neighborhood, occur date, occur time, offense category, offense type, open data latitude/longitude, open data X/Y, and offense count. These features provide a comprehensive and diverse set of information, enabling the models to make accurate predictions. The project's significance lies in its potential application for law enforcement agencies, city planners, and policymakers. By accurately predicting and classifying crimes, it facilitates proactive decision-making and resource allocation, contributing to the enhancement of public safety and the efficient utilization of law enforcement resources. The classification of crimes into specific categories allows for a more nuanced understanding of crime patterns, enabling stakeholders to implement targeted interventions and preventive measures. The high accuracy of the models demonstrates their reliability and effectiveness in handling real-world crime prediction scenarios. In summary, "Crime Prediction Using Machine Learning" presents a robust and accurate approach to crime prediction, leveraging advanced algorithms and a rich dataset. The project's success in accurately classifying diverse crime categories makes it a valuable tool for enhancing public safety and optimizing law enforcement strategies in urban environments.

In the realm of healthcare, timely and accurate drug recommendations during medical emergencies can significantly impact patient outcomes. This project presents a robust "Drug Recommendation System in Medical Emergencies using Machine Learning," implemented in Python. The system leverages two powerful classification algorithms, namely the Random Forest Classifier and the Decision Tree Classifier, attaining remarkable accuracies of 100% on both training and test datasets. The dataset employed in this project comprises 1200 records, each characterized by 30 features. These features encapsulate a diverse set of medical parameters, providing a comprehensive representation of patient health. The dataset spans 10 distinct classes, encompassing a spectrum of medical conditions: Allergy, Chickenpox, Chronic, Cold, Diabetes, Fungal, GERD, Jaundice, Malaria, and Pneumonia. The Random Forest Classifier, known for its ensemble learning capabilities, and the Decision Tree Classifier, recognized for its interpretability, were meticulously chosen to model the intricate relationships within the dataset. Both algorithms exhibited exceptional performance, achieving perfect accuracy scores on both training and test datasets, signifying the efficacy of the developed recommendation system. This project not only serves as a testament to the potency of machine learning in healthcare applications but also underscores the critical role of accurate drug recommendations in emergency medical scenarios. The achieved 100% accuracy underscores the reliability and precision of the system, instilling confidence in its potential deployment in real-world medical settings. As we navigate the intersection of technology and healthcare, this Drug Recommendation System stands as a testament to the transformative impact of machine learning on patient care in critical situations.

The demand for accurate and efficient gold price prediction has witnessed a surge in recent years, driven by its pivotal role as a global economic indicator and a haven asset.In an era defined by data-driven decision-making, the project "Efficient Machine Learning Algorithm for Future Gold Price Prediction" emerges as a groundbreaking endeavor. Developed using Python, this project leverages the formidable Random Forest Regressor algorithm to unlock the secrets of gold price trends, achieving an unparalleled accuracy score of 0.9995. The project's foundation is a rich dataset, consisting of 3,834 meticulously curated records. This dataset serves as the bedrock upon which the machine learning model is built, encapsulating a diverse range of insights into the gold market. As the project's core, the Random Forest Regressor algorithm exhibits remarkable prowess in capturing the intricacies of gold price dynamics. Its ensemble of decision trees harmoniously collaborates to forecast gold prices with astonishing precision. With an accuracy score that verges on perfection, this model stands as a testament to the synergy between cutting-edge technology and financial analysis. The dataset's 3,834 records provide a panoramic view of the gold market's evolution, offering crucial variables such as historical opening and closing prices, daily highs and lows, and trading volumes. These attributes encapsulate the essence of the gold market's fluctuations, enabling the algorithm to discern patterns and anticipate price movements.By utilizing Python, a versatile and widely adopted programming language for data science and machine learning, we ensure transparency, reproducibility, and scalability of the prediction model. In conclusion, this project marks a significant milestone in the realm of gold price prediction. With Python as its foundation and the Random Forest Regressor as its engine, it exemplifies the marriage of computational excellence and financial acumen. Its remarkable accuracy score underscores its potential to reshape the future of gold market forecasting, offering valuable insights to investors and stakeholders alike.

The "House Price Prediction using Machine Learning" project presents a comprehensive approach to predicting real estate prices by harnessing the power of advanced data analysis techniques. Developed primarily using Python programming language, the project employs the Random Forest Regressor algorithm as its core predictive model. The objective is to accurately estimate the prices of residential properties, contributing to informed decision-making in the real estate market.In this project, a dataset containing 42,703 individual data points from the United States of America is utilized for training and evaluation. The dataset encompasses various essential features that influence property prices, including location, square footage, number of bedrooms and bathrooms, amenities, and more. By leveraging this diverse set of attributes, the Random Forest Regressor algorithm learns intricate patterns and relationships within the data, enabling it to make reliable predictions.The project's success is measured by the achieved performance metrics. During the training phase, the model attains a Mean Absolute Error (MAE) of 1.4606, indicating the average absolute difference between predicted and actual prices on the training set. Furthermore, on the test set, the model demonstrates its generalization capability by achieving a MAE of 3.8313. These metrics underscore the model's ability to make accurate predictions on unseen data, enhancing its practical utility in real-world scenarios.The Proposed House Price Prediction using Machine Learningshowcases the efficacy of the Random Forest Regressor algorithm in forecasting residential property prices. The Python-based implementation leverages a dataset comprising thousands of data points from the United States, contributing to a robust and reliable predictive model. The achieved low Mean Absolute Error values on both training and test sets emphasize the model's accuracy and generalization potential. This project holds significant implications for individuals, investors, and real estate professionals seeking data-driven insights to navigate the dynamic real estate market.

Stress, an increasingly prevalent aspect of modern life, can significantly impact an individual's physical and mental well-being. Hence, understanding and monitoring stress levels play a crucial role in promoting overall health and quality of life. The project "Human Stress Detection Based on Sleeping Habits Using Machine Learning with Random Forest Classifier" presents a novel and effective approach to detect human stress levels by analyzing their sleeping habits. Leveraging the powerful capabilities of Python programming language, the study employs the Random Forest Classifier algorithm, known for its versatility and accuracy in classification tasks.The primary objective of this research is to develop a reliable stress detection system that can provide valuable insights into individuals' stress levels, enabling timely interventions and promoting better mental health.The dataset used in this study is carefully curated and comprises various essential parameters related to both sleep patterns and stress levels. These parameters include the user's snoring range, respiration rate, body temperature, limb movement rate, blood oxygen levels, eye movement, the number of hours of sleep, heart rate, and stress levels categorized into five classes: 0 (low/normal), 1 (medium low), 2 (medium), 3 (medium high), and 4 (high). The inclusion of these diverse parameters ensures a comprehensive analysis of sleep patterns and their correlation with stress levels.To achieve accurate stress detection, the Random Forest Classifier is chosen as the machine learning model due to its ability to handle complex data relationships, mitigate overfitting, and offer high predictive accuracy. The model is trained using the dataset, and its performance is evaluated on a separate test dataset to ensure generalization and unbiased assessment.The results of the experiments reveal a Training score of 100% and an impressive Test score of 97%, demonstrating the effectiveness and robustness of the proposed methodology. The achieved high accuracy showcases the model's capability to learn intricate patterns from the dataset and make accurate stress predictions based on the user's sleeping habits.This stress detection system's potential applications are vast, ranging from personal health monitoring to medical research and interventions. By enabling individuals to gain insights into their stress levels through analysis of their sleep habits, the system empowers them to take proactive measures to alleviate stress, improve sleep quality, and foster overall well-being.