VFAST - Virtual Foundation for Advancement of Science and Technology (Pakistan)
Not a member yet
1255 research outputs found
Sort by
A Next Generation Real Time Frame work for Drone Video Decoding Leveraging IoT-Enabled Communication Network
Full text linkThe new data processing systems required reliable. Fast and low latency data services for smart cities operations and unmanned vehicle systems for quick and fast decisions. At present the low latency and speed data for video decoding for real time are required for intelligent decisions. In this research work we present video decoding model based on for decoding data in real time. This proposed model is based on hybrid Edge-Fog-Cloud orchestration layer that perform decoding task in real time according to the network congestion and this technique ensure data integrity, traceable task distribution and protect the data from tempering by using IoT backbone secured by blockchain technology. To reduce the risk of end-to-end latency and packet loss in worst conditions a novel Temporal-Spatial Predictive Decoding (TSPD) method is used. The AI model deep reinforcement learning is used for fast decisions. After analyzing it can be concluded that a 47.8% improvement in decoding throughput, a 62% reduction in jitter and 38% improvement QoE. This shows satisfactory performance from proposed model. By optimizing energy-latency and combining decentralized system with IoT-driven communication for autonomous aerial system can be used in future 6G network
The introduction of Artificial Intelligence in Medical Education: A Narrative Review of Implementation Practice, Evaluation, and Methodological Barriers
Full text linkArtificial Intelligence (AI) is changing the face of medical education at an unprecedented rate, providing new opportunities for personalized learning, automated assessment, intelligent tutoring, clinical simulation, and data-driven decision support. Although this adoption is gaining momentum, the use of AI in medical curricula is still disjointed, and there is considerable diversity in the practices and approaches of implementation, assessment, and institutional preparedness. This is a narrative review of the existing data on the applications of AI tools in medical training, their educational utility, practical application, and effects on student learning. The review also presents methodological challenges, such as the inconsistent evaluation schemes, insufficient empirical validation, ethical issues, problems with the curriculum, readiness by faculty readiness, and unclear structural constraints, which are obstacles to scalable and standardized integration. This review offers a clear basis to support the evolution of AI-enhanced medical education by the educator, policymaker, and institutions by mapping the existing practices, defining gaps, and identifying recommendations to be implemented and researched in the future
Powering Tomorrow Safely: Handling IoT Cybersecurity Issues and Ethical Energy Systems Environments
Full text linkIoT integration is transforming global energy networks, leading to previously unheard-of sustainability and efficiency. Nevertheless, there are significant cybersecurity risks and moral dilemmas that are closely linked to this potential. IoT devices with limited resources are widely deployed, creating a huge attack surface for sophisticated attacks, including those by nation-state actors. At the same time, gathering energy data in real time presents serious ethical issues with algorithmic bias, data ownership, and privacy. Resolving these intricate issues calls for a multifaceted approach that combines robust security frameworks, moral governance, and proactive policy development in order to create a safe and equitable energy future.
Improved Traffic Safety with YOLO-v8 Driven Smart Helmet Detection
Full text linkHelmet compliance is an essential need yet the monitoring of helmet remains a challenge, due to which various accidents take place and people loss their precious lives due to various head injuries. This paper proposes a YOLO-v8 based biker helmet detection system to enforce helmet wearing by riders. The system uses the advanced YOLO-v8 system that is trained on a self-prepared & labelled data set consisting of images captured from various roads & angles, this architecture ensures prompt communication with traffic regulations by achieving an accurate and exact detection and generating a report. The model achieved a detection accuracy of 97% demonstrating its reliability and robustness. The system works by processing input videos provided via dataset and analyzes each frame to detect helmet violations and non-compliance. This provides the authorities and government formalize a plan for traffic solutions that helps in reduce accidents foster awareness. By addressing the in-efficiencies of traditional methods the system offers high accuracy, adaptability in various conditions like lowlight, weather, etc. making it suitable for smart cities and also supports integration with other systems like traffic management. This model can be optimized with edge devices, integration with vehicle detection to further enhance enforcement strategies. This strategy can be easily implemented into current traffic management systems because to its scalability and practicality, utilizing AI tools such as YOLOv8 to solve problems efficiently and improve road safety
Analyzing the Impact of User-Generated and Firm-Generated Content on Online Shopper Behavior in a Developing Market: A Technology Acceptance Model (TAM) Approach
Full text linkIt is important to note that online content is central to the development of consumer behavior in the digital age of the marketing transformation wherein consumers have become heavily dependent on the nature, quality, and validity of information they see online prior to making purchase-related decisions. The assumption as to whether user-generated or firm-generated content is more important issue is, however, under-researched and especially in such countries as Pakistan, where the adoption of digital is growing at a rapid pace, but academic and empirical interest in the given issue has been limited. The study is based on analyzing the impact of user-generated content and firm-generated content on evaluation perceptions and attitudes of online shoppers in Pakistan and employs the Technology Acceptance Model (TAM) as the theoretical framework to interpret the research results. The results show that there is no significant difference between the two types of content in regard to the perceived usefulness or intention to purchase, but the user-generated content has a significantly greater impact on the attitudes towards the advertisements and brand perception. This implies that consumers are more likely to consider as naturally, relatable and credible content that is produced and shared by other users than that which is specifically produced and distributed by companies. These discrepancies point to the role of consumer voices within digital ecosystems in which online trust and credibility is highly involved in decision-making, and found that in situations where both types of content deliver helpful information, user-generated content is more sensitive to content influence than the content produced by the company. Moreover, it provides useful lessons to the marketers who work in the developing markets with particular reference to Pakistan because of the necessity to incorporate user-driven stories, reviews, and experiences into the digital strategies to enhance consumer engagement, brand image, and effectiveness of the marketing campaigns
Efficient and Sustainable Video Surveillance Using CNN-LSTM Model for Suspicious Activity Detection
Full text linkThis study presents a novel approach for enhancing the automation and effectiveness of real-time threat detection in video surveillance systems. Traditional surveillance methods require continuous human monitoring, are resource-intensive, and often fail to consistently identify suspicious activities with precision. Addressing these challenges, we propose the Mono-Scale CNN-LSTM Fusion Network, an advanced deep-learning model designed for automated, sustainable, and high-accuracy CCTV systems. The model utilizes Convolutional Neural Networks (CNN) in combination with Long Short-Term Memory (LSTM) networks to improve recognition capabilities by capturing temporal and spatial features. For feature extraction, the Oriented FAST and Rotated BRIEF (ORB) techniques are employed to enhance detection efficiency. The model was tested using the UCF crime image dataset and achieved an accuracy rate of approximately 99%, surpassing traditional models like CNN, VGG-16, VGG-19, ResNet-50, and DenseNet. This study highlights the contributions of our approach, which offers a significant reduction in the need for human oversight and sets new standards in the field of automatic threat detection. Furthermore, it emphasizes the model’s capability to support contemporary security systems with high precision, reliability, and scalability, making it a valuable tool for the next generation of intelligent surveillance systems
Practices of Requirement Engineering Process Maturity in Global Software Development
Full text linkThe software development life cycle places requirement engineering (RE) at its core because of its fundamental importance. Global Software Development (GSD) creates significant difficulties for requirements collection and implementation because it faces challenges including geographical distance and time zone differences and language barriers and cultural differences. Due to difficulties in traditional requirements collection and analysis procedures in GSD projects organizations need to advance their RE process. Over several investigations’ researchers have established that RE process maturity operates as a critical determinant for achieving successful requirements implementation in GSD. The research identifies every potential practice to enhance RE process maturity by analyzing literature through Systematic Literature Review (SLR). The Systematic Literature Review revealed that RE process maturity includes 20 distinct practices as its outcome
Evaluation of Machine Learning–Based Methods to Detect Bipolar Disorder in Individuals With Mental Health Conditions
Full text linkBipolar disorder (BD) is still one of the most incapacitating of neuroaffective disorders in psychiatry. The strong mood swings from states of euphoria to depression often destabilize interpersonal relationships and can undo occupational stability. Early and reliable diagnosis facilitates prompt pharmacological intervention and mental-health education that may protect not only the patient and their immediate social circle but also the entire social structure from general distress. In this research study the performance of machine learning algorithms such as random forest (RF), support vector machine (SVM) and gradient boosting (GB) has been investigated for classification and prognostication of BD and its subtypes. The machine learning models were validated using a clinical dataset, which included 120 participants: 28 of BD I, 31 of BD II, 31 of Major Depressive Disorder and 30 healthy controls. Model performance was evaluated with stratified cross-validated train-test-split and a set of metrics, including accuracy, precision, recall, F1-score, and Receiver Operating Characteristic - Area under the Curve (ROC vs. AUC). In other words, the RF model had the highest accuracy (88%), precision (90%), and recall (88%). The discriminative performance of RF and SVM models was comparable with an ROC-AUC of 97\%. These results emphasize the potential of machine learning (ML), specifically ensemble techniques like Random Forest (RF), as an effective supplement to traditional early clinical diagnosis in bipolar disorders and related psychiatric illnesse
HAR-AttenNet: Multi-Head Transformer for Precise Human Activity Recognition Using Wearable Devices
Full text linkHuman Activity Recognition has become inevitable in wearable technology in healthcare and fitness tracking applications. Precise recognition of daily routine human activities using wearable sensors remains a challenging task due to sensor modality, placement, and environmental factors. Indeed, Machine Learning and Deep Learning models have become robust in human activity recognition, yet these face numerous challenges in precisely recognizing the daily human activities. One major challenge is the variability in sensors readings where the same activity may lead to different sensor readings, intra-class variability, as different people perform the same activity differently or use different devices. To address this, we propose a multi-head transformer model with a multi-head attention mechanism that explicitly handles the intra-class variability. We demonstrate that the multi-head Transformer model exhibits enhanced robustness and performs better even in the face of such variability, although variability is a major problem in the popular PAMAP2 dataset, since it directly impacts the performance of deep learning models
Integrated HealthAI: Revolutionizing Healthcare through Advanced Diagnostic Systems
Full text linkThe incidence of skin cancer has increased dramatically and the need for early detection, management of the disease, and effective public health campaigns are therefore paramount. Here, we introduce an AI-powered health application, encompassing novel features for skin cancer detection, heart rate estimation, and a health-related chatbot, providing new opportunities for accessible healthcare solutions. With the chosen Skin Cancer Classification data set and the YOLOv8 model, the system successfully classifies skin cancers including actinic keratosis, basal cell carcinoma, dermatofibroma, and vascular lesions. It allows users to upload images of skin conditions and get instant data clinical insights, while for heart rate monitoring, it uses video analysis of the user\u27s uploaded facial videos. The chatbot also gives individual health recommendations so that users can better make health-related decisions. The performance metrics (accuracy, precision, recall, and F1 score) present the effectiveness of the application for dermatology and health monitoring. Such tool is promising of better access to healthcare, encouraging screening and preventive care leading to better public health