Volume 14 | Issue 6 |

 Abstract

Humanism occupies a central position in modern Indian literature as a philosophical and literary approach that places human beings, their dignity, suffering, freedom, and ethical responsibility at the centre of literary discourse. Among the foremost Indian writers who successfully transformed human suffering into universal literary experience is Mamoni Raisom Goswami (Indira Goswami). Her novels reveal profound compassion for marginalized communities, women, widows, labourers, tribal groups, refugees, victims of communal violence, and economically deprived people. Rather than merely portraying social realities, Goswami explores the psychological, cultural, and existential dimensions of human suffering and resilience. Her fiction transcends regional boundaries and reflects universal human values such as love, equality, justice, forgiveness, empathy, and hope.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Humanism, Mamoni Raisom Goswami, Indira Goswami, Assamese Novel, Humanity, Social Justice, Women, Marginalization, Compassion, Indian Literature.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

: Early and accurate diagnosis is important since cardiovascular diseases (CVDs) are among the leading causes of death worldwide. Electrocardiography (ECG) is a useful diagnostic tool that allows physicians to detect and monitor a variety of heart disorders by observing the electrical activity of the heart. However, manually identifying ECG characteristics and classifying heartbeats is a challenging and time-consuming operation that requires a high ability level. To deal with the stated issue, we developed a novel system, named ArythmiAR, by combining Convolutional Neural Networks (CNNs) and Augmented Reality (AR) for interactive diagnosis with 3D visualisation and real-time interaction. The major features of the ArythmiAR are: 1. Deep learning ECG classification for better arrhythmia detection, 2. 3D heart modelling and assembly for better visualisation, 3. AR interface for deployment of CNN model, 4. 3D location of the heart sub-regions responsible for arrhythmia anomalies, 5. Improved 3D visualisation and interaction. In this work we explore different strategies for ECG classification, using data rebalancing techniques to improve the performance of the models. We focus on CNN and Multilayer Perceptron (MLP) models which achieved 99.07% accuracy with the MLP model and were extremely competitive on the PhysioNet MIT-BIH Arrhythmia dataset. The paper also demonstrates the use of the deep learning model for ECG categorisation in an augmented reality environment. It is an augmented rendering prototype that allows the user to identify, view and interact with specific cardiac regions causing arrhythmia. It helps doctors to better identify patients and find better ways to treat them.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

ECG Classification, Arrhythmia Detection, Deep learning, Convolutional Neural Networks (CNN), Multilayer Perceptron (MLP), Augmented Reality (AR), 3D Heart Visualisation, Medical Imaging, Healthcare AI, PhysioNet MIT-BIH Dataset.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Most crime scenes contain fingerprints, which are among the most crucial pieces of identification evidence. For the creation of latent fingerprints on various surfaces, including porous, non-porous, and semi-porous surfaces, various kinds of fingerprint powders are utilized. Developing efficient, economical, and ecologically conscious methods for latent fingerprint visualization on non-porous surfaces remains challenging. In this work, off-white and grey powders derived from crab shells, an abundant marine waste material, were used to develop a novel fingerprint powder for latent fingerprint visualization. White powder from crab shells was prepared by cleaning, boiling, drying, crushing, and sieving. Grey powder from crab shells was prepared by using muffle furnace it using analytical techniques in order to assess its particle size, shape, and chemical properties. The material was tested for its viability in visualizing fingerprints on non-porous surfaces by using various deposition techniques, such as TV glass, Polished wood, Metal door, mobile glass, plastic covers. The developed fingerprints were evaluated for contrast, clarity and ridge detail. Results show that crab shell powder has good affinity with sweat and oil residues left in latent fingerprints and produces clear and well-defined ridge patterns. The results suggest that crab shell ash can be a low-cost, eco-friendly and efficient alternative for latent fingerprint visualization. This research adds to the field of forensic science, as well as demonstrates the potential to use marine waste for sustainable technology.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Latent fingerprints, Crab shell Powder and Ash, Non-porous surfaces, Semi- porous surfaces, Eccrine Fingerprints, Sebaceous Fingerprint, Eco-friendly fingerprint powder.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Emotion recognition is an increasingly important part of improving human-computer interaction. This is because emotions have a large part in how people behave towards each other and how they feel in general. Many industries want robots that can recognise and respond to emotional cues as people do. Affective agents can be useful in many fields such as education, health care, gaming, marketing, customer service, human-robot interaction and entertainment. The aim of this study is to investigate the potential of multimodal (artificial intelligence) AI to improve virtual assistants. More effective and sympathetic virtual assistants are developed using different emotion recognition techniques. The proposed approach enhances the system's emotion awareness and improves the user satisfaction with sympathetic dialogue by using written cues and facial expressions. The proposed MER (Multimodal Emotion Recognition) is effective as the FER (Facial Emotion Recognition) model reaches 71% accuracy in real-time and the TER (Textual Emotion Recognition) model reaches 59% accuracy in validation. Our lightweight architecture merges DialoGPT-based answer generation with face and text-based emotion identification for real-time inference. It is different from other multimodal emotion-aware systems in that it shows how it can work with large language models to have empathetic conversations

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Multimodal Emotion Recognition; Facial Emotion Recognition; Textual Emotion Analysis; Affective Computing; Human-Computer Interaction; Empathetic Virtual Assistants;

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Cardiovascular diseases (CVDs) are still one of the leading causes of mortality worldwide, which reveals the importance of early and precise diagnosis. ECG stands for electrocardiography and is a helpful diagnostic tool that allows doctors to identify and track different heart conditions by examining the electrical activity of the heart. However, manual identification of ECG features and classification of heartbeats is a time consuming and difficult task which requires a lot of skills. We developed a novel system named ArythmiAR to solve the above problem by integrating Convolutional Neural Network (CNN) and Augmented Reality (AR) for interactive diagnosis with 3D visualisation and real-time interaction. The main features of the ArythmiAR are: (1) deep learning ECG classification for efficient arrhythmia detection; (2) 3D heart modelling and assembly for better visualisation; (3) AR interface for deploying CNN models; (4) 3D location of the sub-regions of the heart responsible for arrhythmia anomalies; and (5) improved 3D visualisation and interaction. In this work we explore different ECG classification approaches, using data rebalancing techniques to improve the performance of the models. Our focus is on Multilayer Perceptron (MLP) and Convolutional Neural Network (CNN) models that performed very competitively on PhysioNet MIT-BIH Arrhythmia dataset and reached 99.07% accuracy with MLP model. The work also demonstrates the use of the deep learning model for ECG classification in an AR environment. It is a prototype for augmented rendering, allowing the user to locate, see and interact with specific parts of the heart that cause arrhythmia. The platform gives doctors the tools they need to make better diagnoses and create better treatment plans, improving care for all patients

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

ECG Classification, Arrhythmia Detection, Deep Learning, Convolutional Neural Networks (CNN), Multilayer Perceptron (MLP), Augmented Reality (AR), 3D Heart Visualization, Medical Imaging, Healthcare AI, PhysioNet MIT-BIH Dataset.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

: Help desk systems are an integral part of any institution or public service environment as visitors often need instant help for navigation, enquiries and access to services. In many cases, manual help desks are not available around the clock, and visitors have to wait for a human operator. The Fixed-Place Help Desk Robot with Face Recognition and Voice Q&A for Visitor Assistance addresses this issue by providing an automated, voice activated, fixed-position assistant that can answer frequently asked questions and provide directional assistance. The Raspberry Pi 5 is the heart of the system and is the central controller for processing voice input, responses and the camera. A microphone listens to the user's speech . Speech recognition converts it into text . A response engine searches the query against a set of preprogrammed answers in the system. The generated response is then converted into speech via text to speech output and played through a speaker. At the same time, live visual monitoring is provided by a camera at the help desk location.This project is particularly useful in colleges, hospitals, offices, libraries and other public places where visitors repeatedly and frequently ask questions. The system reduces manual work load and improves response time and provides a uniform user-experience. It also shows how embedded hardware, speech technologies and human-computer interaction can be practically integrated into a compact and low-cost assistive system..

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Raspberry Pi 5, Fixed-Place Help Desk Robot, Face Recognition, Voice Recognition, Speech-to-Text, Text-to-Speech (TTS), Human-Machine Interaction (HMI)Raspberry Pi 5, Fixed-Place Help Desk Robot, Face Recognition, Voice Recognition, Speech-to-Text, Text-to-Speech (TTS), Human-Machine Interaction (HMI)

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

In civil engineering, the structural integrities of bridges are very important. This is because problems with the bridge surface can make bridges unsafe, more costly to maintain, and even lead to catastrophic failure. Most traditional methods for inspecting bridges are manual, where trained personnel look for cracks, spalling, corrosion and other issues on the surfaces of. These manual inspections work to some extent, but they are time-consuming and labour-intensive and are prone to error. Moreover, manual assessments are subjective and often do not agree with each other, especially when there are multiple defects or when surface conditions are made more difficult by lighting, noise, or other environmental factors. In recent years, machine learning and deep learning have emerged as powerful tools to automate the detection of defects in civil infrastructure. Among them, convolutional neural network (CNN) based object detection models like the YOLO (You Only Look Once) family have shown a lot of promise for real-time detection tasks due to their speed and efficiency. But the models that are already out there have some issues. Many of them are designed to find only one kind of defect, making them less useful in the real world where multiple defects often happen at the same time. Standard models also lack mechanisms to focus on the most important parts of an image, making it more difficult for them to detect small, subtle or overlapping defects. Bounding box regression uses standard loss functions such as IoU or GIoU but these may not be effective with defects that are not in the shape of a box which can lead to errors in localisation . These limitations highlight the necessity of a more powerful and precise system capable of detecting various kinds of defects on bridge surfaces under different challenging and variable circumstances. We propose a YOLOv8 model with an improvement using the Convolutional Block Attention Module (CBAM) and the Wise-IoU loss function in this paper. This new model is named YOLOv8-CBAM-Wise-IoU. The CBAM module enhances the feature representation by employing channel and spatial attention. This guides the model to focus on relevant portions of the image and ignore irrelevant background noise. This is even more important for detection of small cracks, subtle corrosion or overlapping defects, that could be missed by standard detection models. The Wise-IoU loss function improves bounding box regression by dynamically adjusting the loss weight according to the object characteristics. This leads to more accurate localisation and reduces false positives. The proposed model can detect seven different types of defects on bridge surfaces simultaneously. They consist of cracks, spalling, corrosion, delamination, surface scaling, efflorescence and potholes. We collected a large set of labelled bridge pictures for training the model, and augmented it with methods such as rotation, scaling and brightness changes to make it more generalisable. We put a lot of effort into improving the training, tuning hyperparameters, performing ablation studies to evaluate the contribution of each component (YOLOv8 backbone, CBAM, and Wise-IoU) to the overall performance. Tests show that YOLOv8-CBAM-Wise-IoU performs much better than the standard YOLO models and other common methods. The model could detect 97.9% of the defects, 76% of the times it was supposed to, 58% of the times it was supposed to and 55.4% of the times it was supposed to. It was also able to detect defects in real world scenarios such as variations in lighting, complex backgrounds and non-uniform surface textures. The system also had a faster detection speed which was good for real time use and this meant it could be used for inspection systems that use drones or cameras. In summary, the YOLOv8-CBAM-Wise-IoU model is a reliable, scalable and efficient method for multi-defect detection on bridge surfaces. To solve the problems of traditional detection methods, the system adopts the attention mechanism and more advanced loss function. Enables the complete, accurate and automated monitoring of the health of structures. It could enhance preventative maintenance schedules, lower the cost of inspections and make bridges safer overall. This study adds to the development of intelligent infrastructure inspection systems, and also shows the effectiveness of combining deep learning, attention mechanisms, and optimised loss functions to address practical civil engineering applications.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

YOLOv8, Multi-Defect Detection, Bridge Surface Inspection, Convolutional Block Attention Module (CBAM), Wise-IoU Loss Function.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The importance of emotion recognition is increasing for improving human-computer interactions. People's feelings strongly affect people's feelings and interaction with each other. In many sectors, machines that can recognise and respond to emotional cues like humans do are needed. Emotionally sensitive agents can be useful in many industries such as education, healthcare, gaming , marketing, customer service, human-robot interaction and entertainment. We investigate in this paper how to improve effectiveness and compassion of virtual assistants by upgrading them with multimodal Artificial Intelligence (AI) and a variety of emotion identification methods. The approach proposed uses written signals and facial expressions to increase the awareness of emotions of the system and to improve the satisfaction of the user by means of sympathetic dialogue. Multimodal Emotion Recognition (MER) is effective as the Facial Emotion Recognition (FER) model is 71% accurate in real time and the Textual Emotion Recognition (TER) model is 59% accurate in validation. The lightweight design allows for real-time inference and combines DialoGPT-based answer generation with textual and facial emotion recognition. This shows that it works with large language models for empathetic interaction, unlike previous multi-modal emotion-aware systems.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Multimodal Emotion Recognition; Facial Emotion Recognition; Textual Emotion Analysis; Affective Computing; Human-Computer Interaction; Empathetic Virtual Assistants

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Contribution: This paper discusses the advantages and difficulties of the construction, implementation and assessment of MoodleBot, a large language model (LLM) chatbot, in computer science education settings. It explores possible LLM applications in LMSs (e.g. Moodle) for supporting self-regulated learning (SRL) and help-seeking behaviour. Computer science teachers find it challenging to add new features to LMSs that make the learning environment more engaging and helpful. MoodleBot solves this. MoodleBot is a platform for communication between teachers and students. Research Questions: Despite teachers' and educators' hesitation to embrace new AI technology and the problems of bias and hallucinations, this study answers two questions. RQ1: What are students' perceptions of MoodleBot as a teaching tool? (RQ2) How accurate are the MoodleBot responses and how close are they to the assigned course material? technique: This study reviews the pedagogical literature on AI-powered chatbots and applies the retrieval-augmented generation (RAG) technique in the design and data processing of MoodleBot. The technology acceptance model (TAM) examines the degree of user acceptance via factors like perceived utility and perceived ease of use. The study included 46 participants and 30 of them filled out the TAM questionnaire. Results: LLM-based chatbots such as MoodleBot can improve the teaching and learning experience significantly. The success rate of help with course-related tasks (88%) was high in this study. The positive reaction of the students is a sign of the success and practicality of AI-powered instructional tools in real life. The results show that educational chatbots can be used in courses to tailor learning and to support teachers' work but that automated fact-checking needs to be improved.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Artificial Intelligence, Large Language Models, Retrieval-Augmented Generation, Learning Management Systems, AI Chatbot, Personalized Learning, Higher Education.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Non-Technical Losses (NTLs), such as stealing electricity, messing with meters, making illegal connections, and billing fraud, are a big problem for modern power distribution systems because they cost a lot of money and slow things down. Old-fashioned ways of finding fraud, like manual inspections and audits based on fixed rules, don't work well, cost a lot, and can't handle the huge amounts of data that smart meters send. This paper talks about how well a Machine Learning-based Apriori Algorithm works to automatically find NTLs. The proposed system analyses three years' worth of monthly electricity usage data from about 15,000 customers. We used and compared a number of models, including Support Vector Machine (SVM), Deep Neural Network (DNN), Gradient Boosted Reinforcement Learning (GBRL), and Apriori-based association rule mining. Experimental results show that the Apriori-based model is more accurate, has a higher recall rate, is more precise, is more specific, and has a lower false positive rate than traditional methods. The system helps utility companies save money, run their businesses more efficiently, and make the smart grid more reliable.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Non-Technical Losses, Electricity Theft, Smart Grid, Apriori Algorithm, Machine Learning, Fraud Detection, Support Vector Machine, Deep Learning

  License

Creative Commons Attribution 4.0 and The Open Definition