Volume 14 | Issue 4 |

 Abstract

The advent of Artificial Intelligence (AI) in Human Resource Management has profoundly changed the conventional performance management systems by establishing data-based, objective and ongoing evaluation processes. This paper will discuss how the AI-driven performance management systems are perceived by employees and how this affects employee motivation and job satisfaction, with the main variables under consideration being trust, fairness, transparency, understanding, and comfort. A structured questionnaire was used to gather primary data on 60 respondents working in different industries. Statistical techniques such as descriptive analysis, correlation, and regression were used to examine relationships among variables and test hypotheses. The results show that employees have a somewhat positive but skeptical attitude towards AI-based systems. Perceived fairness was found to have the strongest impact on both motivation and trust, and transparency and trust had weaker and inconsistent impacts. The strongest predictor of job satisfaction was found to be understanding of AI systems, followed by motivation and comfort. Moreover, a hybrid (AI + human) solution was strongly preferred by employees, which suggests the value of human involvement should not be overlooked. Although AI has proven useful in terms of efficiency and bias reduction, issues associated with a lack of human interaction, data privacy, and algorithmic bias remain. The research concludes that the success of AI-based performance management systems is determined not only by the level of technological advancement but also by the perception and acceptance of employees. Successful implementation and long-term organizational results require a balanced, transparent and human-focused approach.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Artificial Intelligence (AI), Performance Management Systems, Employee Perception, Fairness and Transparency, Employee Motivation, Job Satisfaction, Human-AI Interaction

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The rapid advancement of Artificial Intelligence (AI) and Internet of Things (IoT) technologies has enabled the development of intelligent systems for modern agriculture. This project presents an integrated solution for automated mango grading and environmental monitoring using deep learning and IoT-based control mechanisms.The system employs a YOLOv5-based object detection model to identify and classify mangoes into three categories: ripe, medium ripe, and unripe. A live video stream from a webcam is processed in real time using computer vision techniques, allowing accurate detection and counting of mangoes. The trained model, deployed in ONNX format, ensures efficient and fast inference suitable for real-time applications.The detection results, including the count of each mango category, are transmitted to an ESP32 microcontroller through serial communication. The ESP32 acts as a central IoT node, integrating multiple environmental sensors such as DHT11 (temperature and humidity), soil moisture sensor, and LDR (light intensity sensor). These sensors continuously monitor the surrounding conditions to support smart agricultural decision-making.The system is further integrated with the Blynk platform, enabling remote monitoring and control through a smartphone interface. Real-time data, including mango grading results and environmental parameters, are displayed on the mobile application. Additionally, automation features are implemented using relay modules to control devices such as a water pump and lighting system based on sensor thresholds and user input.An LCD display is also incorporated for local visualization of sensor data, ensuring accessibility even without internet connectivity. The system operates efficiently with minimal human intervention, reducing manual labor and improving grading accuracy. Furthermore, it provides scalability for integration into larger agricultural or industrial environments. Overall, the proposed system demonstrates a cost-effective, reliable, and intelligent approach to fruit grading and smart farming, combining AI-driven decision-making with IoT-enabled monitoring and automation. This solution has the potential to significantly enhance productivity, reduce post-harvest losses, and support precision agriculture practices.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

YOLO, ESP32, IOT, Embedded systems _____________________________________________________

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Organizational Citizenship Behaviour (OCB) refers to the voluntary and discretionary actions of employees that contribute positively to the overall functioning of the organization beyond their formal job requirements. Employee performance reflects how effectively an employee carries out job responsibilities and achieves work-related goals. This study examines the impact of OCB on employee performance at Bhargave Rubber Pt. Ltd., Madurai, a manufacturing company engaged in the production of rubber seals and moulded rubber components. Data was collected from 100 employees using a structured questionnaire and analysed using correlation analysis, regression analysis, ANOVA, and coefficient analysis through SPSS. The results reveal a strong positive relationship between OCB and employee performance (r = 0.693, p = 0.000), with OCB explaining approximately 48% of the variation in employee performance. The findings confirm that promoting organizational citizenship behaviours such as helping colleagues, following organizational rules, and maintaining a positive attitude can significantly improve employee performance and organizational effectiveness.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Organizational Citizenship Behaviour, Employee Performance, Correlation, Regression, ANOVA, Bhargave Rubber Pt. Ltd.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Rapid urbanization has significantly increased road traffic, creating a strong demand for intelligent traffic monitoring systems. Accurate estimation of vehicle density and recognition of multiple vehicle classes remain challenging due to occlusion, illumination variation, and complex road environments. This study proposes a hybrid deep learning framework for real-time vehicle density estimation and vehicle classification in intelligent transportation systems. Initially, adaptive bilateral filtering is applied to improve image quality while preserving edge information. EfficientNet-B3 is then utilized for discriminative feature extraction from traffic scenes. Subsequently, YOLOv8 is employed for precise vehicle localization, and an attention-based BiLSTM classifier is used to determine traffic density levels and vehicle categories. Experimental evaluation demonstrates that the proposed framework achieves superior accuracy, precision, and computational efficiency compared with conventional CNN and Faster R-CNN-based approaches. The proposed system can support intelligent traffic management and smart city applications through reliable real-time performance.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Enhanced CNN, Automatic feature extraction, Faster R-CNN optimization, Vehicle detection, Traffic density analysis

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Bharat mein jaivik krishi ki samasyaen evam sambhavnayen - ek bhaugolik adhyayan

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Bharat mein jaivik krishi ki samasyaen evam sambhavnayen - ek bhaugolik adhyayan

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The rapid adoption of Artificial Intelligence (AI) in recruitment has created new opportunities for improving the efficiency, consistency, and scalability of interview evaluation. Traditional interview processes often rely on subjective human judgment, delayed feedback, and inconsistent assessment criteria, which can negatively affect hiring quality and candidate experience [1], [2]. Recent studies have shown that AI-assisted interview systems can improve evaluation accuracy by analyzing candidate responses, communication patterns, and behavioral cues in virtual interview settings [1], [4], [9]. In parallel, advancements in Large Language Models (LLMs) have demonstrated strong capability in contextual understanding, semantic reasoning, and automated feedback generation, making them highly suitable for intelligent interview assessment applications [1], [8], [14]. This paper presents an AI-powered automated interview feedback generation system that integrates Speech-to-Text (STT), Natural Language Processing (NLP), and Large Language Models (LLMs) to provide real-time and structured candidate evaluation. The proposed system begins with role-specific interview initialization through a lightweight web-based interface, where candidate details and job preferences are collected. Based on the selected role, the system dynamically generates technical and behavioral interview questions tailored to the candidate profile [3], [8]. During the interview session, candidate responses are captured through audio input and converted into text using speech recognition. The transcribed responses are analyzed using NLP techniques to evaluate communication clarity, response relevance, logical flow, and answer quality [9], [11], [12]. An LLM-based semantic evaluation module further assesses contextual understanding, identifies strengths and weaknesses, and generates personalized feedback. The system then produces a detailed audit report comprising competency-wise scores, overall performance summaries, skill gap analysis, and actionable recommendations. The proposed framework offers a lightweight, scalable, and practical end-to-end solution for intelligent interview assessment, reducing evaluator bias, improving feedback consistency, and enabling efficient decision-making in modern digital recruitment workflows [5], [6], [15].

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Artificial Intelligence, Speech-to-Text, Natural Language Processing, Large Language Models, Automated Interview Assessment, Candidate Evaluation, Real-Time Feedback, Semantic Analysis, Recruitment Analytics, Web-Based Interview System.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The rapid advancement of computer vision technolo-gies and the growing demand for contactless human-computer interaction have opened new possibilities for gesture-based con-trol systems. Traditional input devices such as the mouse and keyboard impose physical constraints on users and are unsuitable for hygienic, futuristic, or accessibility-oriented environments. This paper presents a Touchless Augmented Reality (AR) Virtual Workspace that enables users to interact with a computer system using only hand gestures captured through a standard webcam. The proposed system integrates MediaPipe for real-time hand landmark detection, OpenCV for frame processing, and PyAutoGUI for executing system-level actions. Gesture logic is applied to interpret specific hand positions and movements, enabling operations such as cursor movement, left and right click, scrolling, application launching, and system controls. Additionally, the system provides a floating AR workspace where virtual application icons are rendered on-screen with physics-based behavior including collision detection and body avoidance. Experimental results demonstrate that the system achieves real-time performance with low latency, natural inter-action, and efficient gesture recognition using only a standard webcam. The proposed system proves that gesture-based control combined with augmented reality can significantly enhance user experience and eliminate the need for physical input devices.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Augmented Reality, Hand Gesture Recognition, MediaPipe, OpenCV, PyAutoGUI, Human-Computer Interac-tion, Touchless Interface, Real-Time Processing, Cursor Control, Virtual Workspace

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Landslides are the most dangerous natural hazards, often occurring without warning and causing significant loss of life, property damage, and environmental destruction. Accurate and timely prediction of landslides is essential for effective disaster management and risk mitigation. This project presents a machine learning-based landslide prediction system using a Random Forest classifier to analyse multiple environmental and geological parameters influencing landslide occurrence.The proposed system utilizes features such as rainfall intensity, slope angle, soil saturation, vegetation cover, earthquake activity, proximity to water bodies, and soil type to predict the likelihood of a landslide. Standard normalization methods are used to preprocess and scale the input data in order to boost model performance. The trained machine learning model determines whether the situation is safe or high-risk and provides an estimate of the likelihood of a landslide occurring. To enable real-time integration with embedded and IoT systems, the prediction results are transmitted through serial communication at 9600 baud rates in a simple comma-separated

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Random Forest algorithm, IOT, Embedded systems

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Abstract: The ageing population has emerged as a significant concern in both developed and developing countries, including India. In the contemporary world, the proportion of elderly individuals is steadily increasing, leading to complex social and healthcare challenges. In India, the issue is more pronounced due to the gradual decline of the traditional joint family system, while adequate institutional care facilities are still insufficient. This review aims to explore the symptoms of depression, quality of life, and the major challenges faced by elderly individuals, particularly in India and West Bengal. Existing literature indicates that many older adults experience feelings of loneliness, isolation, helplessness, and reduced self-worth, even when living in well-equipped care homes. Studies have shown that chronic diseases play a crucial role in determining quality of life, whereas factors like age, education, and marital status have limited influence. Overall, the review underscores the increasing burden of depression among the elderly and highlights the urgent need for comprehensive primary healthcare services and community-based interventions to improve their quality of life and ensure early detection and management of mental health issues.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Keywords: Depression, Quality of life, Elderly People, West Bengal, India

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Indian Classical Dance is a rich cultural heritage that employs intricate hand gestures (Mudras), facial expressions (Bhavas), and body postures to convey profound emotional and narrative content. However, the computational analysis of these elements has remained a challenging and largely unexplored domain. In this paper, we present NrityaAI, a comprehensive AI- driven framework that automates the recognition and analysis of Bharatanatyam dance performances from video input. The pro- posed system integrates multiple deep learning models including MediaPipe for skeletal pose estimation, a custom Convolutional Neural Network (CNN) based on EfficientNetB0 for Mudra classification, and DeepFace for facial expression analysis with Rasa mapping. The system processes dance videos through a multi-stage pipeline comprising video validation, frame extrac- tion, Region of Interest (ROI) detection, Mudra identification, expression analysis, context fusion, and automated storyline generation. Our framework achieves high accuracy in Mudra detection while providing rich contextual output including anno- tated videos with text overlays, SRT subtitle files, and human- readable narrative descriptions that map detected Mudras to their cultural Rasa/Bhava significance. The system is deployed as a full-stack web application with a React TypeScript frontend and Python FastAPI backend, supporting user authentication, analysis history, and real-time video playback with synchro- nized captions. Experimental evaluations demonstrate that the proposed system provides accurate, efficient, and culturally meaningful analysis of classical dance performances, bridging the gap between traditional art and modern artificial intelligence.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Indian Classical Dance; Mudra Recognition; Convolutional Neural Network; EfficientNet; MediaPipe; Deep- Face; Multimodal Analysis; Narrative Generation; Dance Video Analysis; Cultural Computing

  License

Creative Commons Attribution 4.0 and The Open Definition