Volume 14 | Issue 4 |

 Abstract

The rapid rise of cloud computing, fast digital networks, and the Internet of Things has given attackers way more ways to break into today's computer systems. Old-school network intrusion detection systems (NIDS) that just look for known patterns are having a hard time catching tricky, low-profile attacks like polymorphic malware or brand-new security flaws. While deep learning is better than older machine learning for finding patterns and sorting data, it's not always practical for important security operations centers (SOCs) because it can be slow and the deep neural network (DDNN) models aren't very clear about how they make decisions. This research introduces a new NIDS that uses a mix of deep learning methods. It uses Convolutional Neural Networks (CNNs) to pick out important patterns in network traffic, Long Short-Term Memory (LSTM) networks to understand the order of data packets, and Random Forest (RF) classifiers to make the overall results more reliable. A logistic regression model then combines the predictions from these three. To help with the lack of clarity in DDNN models, the study also incorporates SHAP values from explainable AI. Analyzing network traffic in near real-time is done using Python's Scapy library and a sliding window approach. This setup lets you build a model offline and then use it live in production through a web service that connects the two. The system was tested on the NSL-KDD and CICIDS2017 datasets, and it achieved classification accuracy over 99% with a Matthew correlation coefficient (MCC) of 0.90 or higher.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Network Intrusion Detection System (NIDS), Deep Learning, Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), Random Forest (RF), Explainable AI (XAI), SHAP, Stacking Ensemble, Real-Time Detection, Cybersecurity.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The proliferation of online news has created challenges for users to access relevant, credible, and timely information. This project presents an AI-Powered Personalized News Aggregator and Fake News Detection System that addresses these challenges by leveraging user preferences, location data, and advanced generative AI. Users can register, set news categories, and receive curated news feeds tailored to their interests. The system integrates Google's Gemini AI for content classification and filtering, ensuring that recommended news matches user preferences while eliminating irrelevant articles. Additionally, a fake news detection module analyzes news content for authenticity using AI-driven verification techniques, providing users with verdicts and confidence scores. The system stores user interactions, enabling automated preference updates based on liked articles. By combining machine learning, NLP, and real-time API integration, this system offers a robust, personalized, and trustworthy news experience, reducing misinformation exposure and enhancing user engagement with relevant content.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Personalized News, AI, Fake News Detection, Gemini AI, User Preferences, News Aggregator, Machine Learning, NLP, Content Filtering, Real-Time API Integration

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The rapid growth of e-commerce and digital plat- forms has significantly increased the demand for interactive and realistic product visualization systems. Traditional methods such as images and videos often fail to provide a clear understanding of how products appear in real-world environments, leading to uncertainty in user decision-making. This paper presents an AI-Driven Augmented Reality System for Real-Time Virtual Try-On and Visualization that enables users to interact with virtual objects through a live camera feed. The proposed system integrates a Unity-based augmented reality environment with a Python backend using FastAPI for efficient real-time communication. Computer vision techniques are implemented using OpenCV for image processing, while MediaPipe is used for accurate detec- tion of facial, hand, and body landmarks. The system supports multiple detection modes, including face-based glasses try-on, hand-based watch placement, pose-based dress visualization, and surface detection for environmental interaction. The detected landmark data is converted into JSON format and transmitted back to Unity, where virtual objects are aligned using anchor-based positioning techniques. Experimental results demonstrate that the proposed system achieves accurate feature detection, stable object alignment, and low-latency performance during real-time interaction. The system effectively combines artificial intelligence and augmented reality to enhance user experience and provides a scalable solution for applications in e-commerce, fashion, and interactive visualization.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Augmented Reality, Artificial Intelligence, Vir- tual Try-On, Computer Vision, MediaPipe, FastAPI, Real-Time Processing, Landmark Detection

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Screen time among Indian children has risen sharply in the post-pandemic period, yet most research treats all screen use as equivalent. This study examined whether the type of screen time -- educational versus recreational -- predicts children's cognitive and social development outcomes differently from total screen time alone. A correlational, cross-sectional design was employed with a sample of 54 parents of children aged 8-14 years recruited via WhatsApp groups. Parents completed the Screen Time Questionnaire (STQ), a five-point Academic Performance Rating for Mathematics and Language (English), and the Strengths and Difficulties Questionnaire (SDQ; Goodman, 1997). Total screen time showed no significant association with academic performance but correlated significantly with SDQ total difficulties (r = 0.398, p = 0.003). Recreational screen time significantly predicted Language performance (r = 0.352, p = 0.009) and SDQ total difficulties (r = 0.457, p < 0.001). Multiple regression confirmed recreational screen time as the only significant predictor of Language rating (B = 0.266, p = 0.012) and SDQ total difficulties (B = 1.985, p < 0.001) after controlling for educational screen time and age. Educational screen time showed no significant associations. Findings support a content-differentiated model of screen time effects and have implications for parents, school counsellors, and digital wellness practitioners.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Screen time, cognitive development, social development, children, recreational screen time, SDQ, academic performance, India, digital media, counselling psychology

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Background and Objective: It is a musculoskeletal condition characterised by progressive loss of active and passive shoulder motion, causing pain, reduced range of motion and muscle weakness. It can be treated using Conventional physiotherapy to control pain, improving range of motion and function but Extracorporeal shock wave therapy (ESWT) is different from Conventional physiotherapy and it is not much practiced in clinical sittings for Periarthritis Shoulder. The objective of the study was to compare the effect of ESWT along with Conventional physiotherapy versus Conventional physiotherapy alone on pain, range of motion and function in subjects with Periarthritis Shoulder. Methods: Quasi Experimental Study Design. This study included 84 subjects with a mean age of 51.4 years have clinically diagnosed with Grade 1 and 2 Periarthritis Shoulder. The subjects were randomly allocated into 2 groups. In Group A (n=42) subjects were treated with ESWT and Conventional Physiotherapy where Group B (n=42) received Conventional physiotherapy alone. Participants were treated thrice a week for 4 weeks. The outcome measures of this intervention were measured in terms of VAS for pain, Universal Goniometer for Shoulder Range of Motion and SPADI for Function. Results: Independent -t? test was used to compare the mean significance difference between continuous variables. Paired -t? test was used to assess the Statistical significance difference between Pre and Post test scores. Statistical analysis of the data revealed that within the group comparison, both groups showed significant improvement in all parameters. Whereas, in between the group's comparison Extracorporeal Shock Wave Therapy combined with conventional physiotherapy showed better improvement compared to Conventional Physiotherapy alone. Conclusion: After 4 weeks of intervention both Extracorporeal Shock Wave Therapy (ESWT) combined with Conventional Physiotherapy and Conventional Physiotherapy alone showed significant improvement on reducing Pain, improving Range of Motion and Function. However, Extracorporeal Shock Wave Therapy combined with Conventional Physiotherapy group was found to be more effective when compared to Conventional Physiotherapy alone. From the findings of the current study, it can be recommended that the Extracorporeal Shock Wave Therapy can be prescribed to the participants with periarthritis shoulder.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Periarthritis Shoulder, Extracorporeal Shock Wave Therapy, Short Wave Diathermy, Visual Analogue Scale, Universal Goniometer and Shoulder Pain and Disability Index.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Because of the growing need for electricity and worries about the environment, renewable energy sources are becoming more and more important. One of the most dependable and long-lasting sources of energy today is solar power. The Photovoltaic-Driven Resonant Inductive Energy Transfer Platform is made to charge electronic devices using solar power and wireless charging. A solar panel collects solar energy, changes it into electrical energy, and stores it in a rechargeable battery. Using electromagnetic induction, the stored energy is then sent wirelessly to charge devices like smartphones. This project shows an eco-friendly charging system that cuts down on the need for traditional electricity sources and does away with wired connections. With wireless charging technology, you don't need to use physical cables or connectors, which makes it easier and more flexible. You can use the system outside, in remote areas, or anywhere else where there isn't a normal electricity supply. The main goal of this project is to make a portable and eco-friendly charging system that uses solar energy and wireless power transfer technology. The system has a solar panel that turns sunlight into electricity. A battery management system stores this energy in a rechargeable battery. Using electromagnetic induction between transmitter and receiver coils, the stored energy is then sent wirelessly to electronic devices. This project shows that it is possible to combine solar power generation with wireless charging technology into one system. The proposed system lessens reliance on traditional electricity, encourages the use of renewable energy, and offers a portable charging option.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Solar Energy, Wireless Charging, Photovoltaic System, Electromagnetic Induction, Rechargeable Battery, Renewable Energy, Portable Charging System.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Access to adequate and clean water is one of the most crucial prerequisites for human survival. Water scarcity is one of the critical challenges facing the world today. With 2.4 percent of the world's land area, India is home to more than 16 percent of the human population. The quantity of freshwater available for such a large population is limited, resulting in a per capita availability of water far below the world average. Water is a key resource necessary for national agricultural production. Growing industries, infrastructure, and urbanization require a tremendous amount of water. However, increasing population and fast-growing urban development are putting immense pressure on water resources, resulting in high levels of water stress for many coastal and river basin states in India. The role of wetlands in addressing the water scarcity issue is critical and of great significance because they can store surface and groundwater, reduce runoff, purify water, and improve groundwater recharge. However, the nation's wetlands have been reduced to one-third. The wetlands have seen a decline of 70%, whereas urban wetlands have been reduced due to encroachment and improper utilization. This Research Paper is a humble attempt to analyze the role of wetlands in addressing water scarcity and suggest certain solutions to conserve these wetlands.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Water Scarcity, Wetlands, Role of Wetlands, Water Management, Challenges

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Managing LPG (Liquefied Petroleum Gas) use well is a big problem in both homes and businesses. This study suggests an IoT-enabled autonomous system for monitoring LPG depletion in real time and planning when to refill it. The proposed framework uses IoT hardware like the ESP8266 microcontroller, MQ gas sensors, load cells, and GSM modules to keep track of the weight and usage patterns of gas cylinders all the time. The system looks at both real-time sensor data and past usage patterns to figure out exactly when a refill will be needed. The system lets you schedule refills in advance by using predictive algorithms and automated alert systems. This gets rid of the chance of petrol shortages happening unexpectedly and cuts down on the need for manual intervention. The platform also makes sure that the monitoring device and the end user can talk to each other reliably through GSM-based SMS notifications.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

LPG monitoring based on the Internet of Things (IoT), predicting gas consumption, the ESP8266 microcontroller, MQ gas sensors, load cell measurement, a predictive refill system, and GSM communication.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

This study focuses on analyzing consumer satisfaction and brand perception of automotive air conditioning systems with reference to Wonjin Autoparts in Chennai. The research examines how factors such as cooling performance, reliability, pricing, and brand image influence customer satisfaction.Primary data was collected from 114 respondents using a structured questionnaire. Statistical tools including percentage analysis, chi-square test, and one-way ANOVA were applied to evaluate relationships between variables.The results indicate that most consumers are satisfied with the performance of automotive AC systems, particularly in terms of cooling efficiency and comfort. Brand reputation plays an important role in shaping customer perception, while awareness about OEM suppliers remains relatively low. The study also reveals that price perception significantly affects satisfaction levels, highlighting the importance of value for money.The study concludes that improving product durability, strengthening brand visibility, and adopting competitive pricing strategies are essential for enhancing customer satisfaction and brand perception in the automotive components sector.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Consumer satisfaction, brand perception, automotive air conditioning, OEM suppliers, price perception, Wonjin Autoparts.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

For intelligent robotic manipulation systems to work, they need to be able to recognise objects and detect strong grips. They are especially important in factories and other places where the lighting is often complicated and changes all the time. Shadows, reflections, glare, and low light levels can all cause traditional computer vision methods and simple deep learning-based methods to fail, which can lead to false object detection, false pose estimation, and false robot grasp. So, the goal of this project is to make an object work piece recognition and grip detection system that uses deep learning. This will be done by combining an improved YOLO Net with a feature fusion and attention module, and then using ROS-based robotic control to get the pose right. The first part of the proposed system is a YOLO-based object detection network that uses improved feature fusion to combine features of different sizes so that it can find both small and complex workpieces. The network has an attention mechanism built in that brings out the features of the object and reduces the noise from an unevenly lit environment. This makes the detection more accurate and reliable. Once the target object has been found in the picture, hand and eye calibration can be done to find the exact change from the camera frame to the robot hand frame. This lets us figure out the object's exact position in the world coordinate frame. Accurate pose information about the object may be very important to avoid mistakes in positioning while the robotic arm moves toward, grasps, and drops off the target object, which can lead to failures in grasping. The robot modelling module sets the manipulator's poses, like its kinematic information, dynamic parameters, joint limits, degree of freedom, and so on. This makes sure that the trajectory planning is possible. The ROS environment brings all the parts together. The control module gets the detection results and pose information, and the robot can then smoothly and without colliding grasp and place things in real time. 3.4. Results of the experiment Experimental tests demonstrate that the proposed system effectively mitigates the issues arising from unbalanced illumination and environmental interference. The model's recognition accuracy is 92.2%, and the overall average success rate for grasping is 93.75%. The experimental results confirm the feasibility, robustness, and operational efficiency of the proposed system in actual industrial automation applications. The system can be easily adapted to different objects, lighting conditions, robotic arm working levels, and other robots that work together.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Deep Learning-Based Object Detection, YOLO-Net Feature Fusion Network, Robotic Workpiece Grasp Detection, Hand-Eye Calibration Pose Estimation, ROS-Based Robotic Control System

  License

Creative Commons Attribution 4.0 and The Open Definition