Journal of Science Engineering Technology and Management Science

Timely and accurate crop identification and health monitoring across large agricultural landscapes remain challenging, especially where field-level inspection is impractical due to scale, cost, or terrain. This paper presents an AI/MLbased system for automated crop classification and health monitoring using satellite remote sensing imagery. A Convolutional Neural Network (CNN), trained on the EuroSAT benchmark dataset, learns discriminative spatial and spectral features directly from Sentinel-2 derived images, eliminating dependency on handcrafted features and manual interpretation. The architecture integrates image preprocessing, a deep convolutional feature extractor, and a softmax classification layer to categorize imagery into land-cover and agricultural classes, complemented by a health-assessment module estimating vegetative condition. The system is delivered through a Flask-based web application with an SQLite backend, offering user authentication, image upload, real-time inference, confidence reporting, historical records, and an administrative dashboard. Experimental evaluation demonstrates classification accuracy between 95-98%, indicating strong generalization across visually similar categories, positioning the framework as a scalable alternative to manual crop surveys for precision agriculture applications.

Keywords : Crop Classification, Convolutional Neural Network, Deep Learning, Remote Sensing, EuroSAT, Satellite Imagery, Precision Agriculture, Crop Health Monitoring, Image Classification, Computer Vision, TensorFlow, Keras, Flask.

Author : Dr. PS. Naveen Kumar, 2Veera Anish Kumar, 3Bhargav, 4Prem Kumar, 5Sai Charan

Title : AI/ML-Based Crop Classification and Health Monitoring Using Remote Sensing Data

Volume/Issue : 2026;03(08)

Page No : 813-821

The rapid growth of satellite and aerial imaging technology has made high-resolution remote sensing data widely available for environmental monitoring, urban planning, defense surveillance, and intelligent transportation. Manual interpretation of such imagery is slow, labourintensive, and prone to human error, motivating the need for automated, learningbased analysis. This paper presents the design and development of an AI/ML-based object detection system for high-resolution remote sensing and aerial imagery, built around the YOLOv8 deep learning architecture. The system is delivered as a secure, web-based platform (Flask, Flask-SQLAlchemy, SQLite) offering two intelligent modules: an Object Detection module that identifies buildings, roads, vehicles and other ground objects from an uploaded satellite/aerial image, and a Traffic Management module that analyses traffic-surveillance video to detect vehicles, estimate traffic density and recommend an appropriate signal state. Each detected object is annotated with a bounding box, class label and confidence score. Experimental results obtained from the deployed prototype show that the system reliably detects multiple object classes—including cars, motorcycles, buses, trucks and pedestrians—under varying scene complexity, and produces real-time, humaninterpretable outputs suitable for practical remote sensing and traffic-monitoring applications.

Keywords : Remote Sensing, Object Detection, YOLOv8, Deep Learning, Satellite Imagery, Aerial Imagery, Traffic Density Estimation, Flask, Computer Vision.

Author : ¹V. Prasad, ²G. Namal Sai, ³D. Anil, ⁴P. Madhan, ⁵U. Kiran Sai

Title : Development of AI/ML Model for Object Detection from High-Resolution Remote Sensing Images

Volume/Issue : 2026;03(08)

Page No : 822-829

Vehicle damage assessment is an essential process in automobile insurance and repair services. Traditional inspection methods are manual, timeconsuming, and prone to errors. This paper presents a Deep Learning-Based Vehicle Damage Assessment System that uses the YOLO (You Only Look Once) object detection algorithm to automatically detect and classify vehicle damages such as dents, scratches, broken glass, and bumper damage from uploaded images. Developed using Python, Flask, OpenCV, TensorFlow/Keras, and SQLite, the system provides fast, accurate, and reliable damage detection by highlighting damaged regions with bounding boxes and labels. It reduces inspection time, minimizes human intervention, and supports efficient insurance claim processing and repair estimation, making it a practical solution for intelligent vehicle damage assessment.

Keywords : Deep Learning, Vehicle Damage Assessment, YOLO, Computer Vision, Object Detection, Image Processing, Flask, TensorFlow, OpenCV, Artificial Intelligence.

Author : 1D. Madhuri,2Damarla Keerthi,3Bapatla Chandini,4Dasari Rajeswari

Title : DEEP LEARNING – BASED VEHICLE DAMAGE ASSESSMENT SYSTEM

Volume/Issue : 2026;03(08)

Page No : 830-836

Artificial intelligence (AI) is reshaping how investors select securities, construct portfolios and monitor risk, yet the evidence on how human judgement and machine intelligence work together remains thin, especially in emerging markets. This study examines whether AI assistance improves perceived investment decision quality and risk-management effectiveness, and whether human oversight strengthens these outcomes. Primary data were collected through a structured questionnaire from 250 retail, high-net-worth and institutional investors in the Hyderabad–Secunderabad region of Telangana, supported by secondary literature and regulatory reports. Reliability testing, correlation, multiple regression, independent-samples t-tests and one-way ANOVA were applied. AI usage intensity, perceived usefulness, trust in AI and human oversight jointly explained 41.2% of the variance in decision quality, and 40.5% of the variance in risk-management effectiveness was explained by AI usage, trust, oversight and decision quality. AI-assisted investors scored significantly higher on both outcomes than non-AI investors. The highest scores were recorded by respondents combining high AI usage with high human oversight, and 64% preferred a collaborative model over AI-only or human-only advice. The results support a human-in-the-loop approach and offer suggestions for investors, financial institutions and regulators.

Keywords : human–AI collaboration, AIassisted investing, risk management, financial decision-making, trust in AI, human oversight, behavioural finance

Author : Dr. Ramadevi Puli

Title : Human–AI Collaboration in Financial Decision-Making: An Empirical Study of AI-Assisted Investment and Risk Management

Volume/Issue : 2026;03(08)

Page No : 837-846

The present study highlights on the role of social work and contributions of Shikshan Maharshi, Shrihari Jivatode who acquired a key role in education, social awareness and community development. Education has always been one of the powerful weapons that can be used for social change especially in addressing the inequalities, poverty and lack of educational opportunities for the disadvantaged sections of the society. He explores his vision, initiatives and contributions to the field of education and society. It aims to gain insight into what his body of work has been Englanding — educational awareness, social responsibility and to empower people to get involved in community development. The study also looks into the impacts of his ideals in the current context, especially for the promotion of inclusive education, social equality and human values. His contributions are reviewed in terms of education for empowerment and social development with a particular focus on the "education/sustainable community development" interface. This is descriptive-analytic research using secondary sources of information (from literature and historical documents) with data from published literature and other information related to his activities and life. The study seeks to document his oeuvre and shed light on the importance of stewardship in the areas of education and social development of a man. It culminates all the ideals which need to be instilled in the forthcoming generations to make the present society educated, equitable and socially responsible as inspired by Shikshan Maharshi Shrihari Jivatode.

Keywords : Shikshan Maharshi Shrihari Jivatode, Social Work, Educational Development, Social Reform, Community Development, Educational Empowerment, Social Equality, Social Awareness, Human Values, Community Welfare

Author : Rahul Keshavrao Thombare, Dr. Shrikant Panaghate

Title : SOCIAL WORK OF SHIKSHAN MAHARSHI SHRIHARI JIVATODE

Volume/Issue : 2026;03(08)

Page No : 847-853

The growing complexity and release frequency of software applications increase the effort required to design, maintain, and execute end-to-end (E2E) tests. Generative artificial intelligence (GenAI) and large language models (LLMs) can assist with test authoring, but generated outputs may contain unsupported actions, invalid assertions, inconsistent test data, or code that cannot be executed. This paper presents an implementation-oriented framework, referred to as Rapid TestGen AI, for organizing the workflow from input preparation and feature extraction to AI-assisted test generation, validation, hallucination analysis, execution reporting, and comparative assessment. The supplied implementation evidence consists of screenshots of a web-based application showing project overview and workflow pages, administrative and user dashboards, dataset management and preprocessing, feature extraction and training, generated test-case views, multi-layer validation, hallucination detection and mitigation, execution-system results, and comparative-analysis pages. These screens establish the presence of corresponding user-interface modules; they do not, by themselves, establish the correctness of the underlying algorithms or the validity of displayed performance values. Accordingly, the paper formalizes a reproducible evaluation protocol and defines metrics for syntactic validity, requirement traceability, assertion support, execution success, functional coverage, hallucination incidence, and human review effort. No new experimental scores are claimed because the source package does not include raw datasets, model configuration logs, test execution traces, ground-truth labels, or independently verifiable result files. The contribution is therefore an implementation-grounded system description and an evaluation design that can be completed with auditable experiments. The framework is intended to support human-supervised test engineering rather than replace expert review.

Keywords : Generative artificial intelligence, large language models, automated software testing, end-to-end testing, test-case generation, hallucination detection, software quality assurance, test validation.

Author : G RAMESH , B YASHASWINI , N JASHWANTH

Title : AN INTELLIGENT GENERATIVE AI FRAMEWORK FOR AUTOMATED ENDTO-END SOFTWARE TEST GENERATION, MULTI-LAYER VALIDATION, AND HALLUCINATION MITIGATION

Volume/Issue : 2026;03(08)

Page No : 854-866

Weed infestation is one of the major challenges affecting agricultural productivity because weeds compete with crops for nutrients, water, sunlight, and space. Conventional weed management techniques generally depend on manual labour, mechanical operations, or indiscriminate application of herbicides, which can be expensive, labour-intensive, environmentally harmful, and ineffective under varying field conditions. This paper proposes a Smart Vision-Based Robotic System for Automated Weed Management that combines computer vision, artificial intelligence, embedded control, and mobile robotics to identify and manage weeds with minimal human intervention. The proposed system uses a camera mounted on a mobile robotic platform to continuously capture crop-field images. Imageprocessing and machine-learning techniques are applied to distinguish crops from unwanted vegetation based on visual characteristics such as colour, shape, texture, and spatial distribution. Once a weed is detected, its location is estimated and transmitted to the robotic control unit, which navigates the actuator toward the identified weed and performs targeted removal. The system architecture integrates an image-acquisition module, weed-detection algorithm, localization module, microcontroller, motor driver, navigation mechanism, and weed-removal actuator. The proposed approach aims to reduce chemical herbicide consumption, labour requirements, crop damage, and operational costs while improving precision in agricultural weed management. The system provides a scalable foundation for autonomous precision-agriculture applications and can be further enhanced using deep-learning models, multispectral imaging, GPS/RTK positioning, and edge computing.

Keywords : Smart Agriculture, Weed Detection, Computer Vision, Robotics, Machine Learning, Precision Agriculture, Autonomous Robot, Automated Weed Management.

Author : P PADMAJA , CH VENKATESWARA RAO , K PRASAD

Title : AN INTELLIGENT VISION-BASED ROBOTIC FRAMEWORK FOR AUTOMATED WEED DETECTION AND PRECISION WEED MANAGEMENT IN SMART AGRICULTURE

Volume/Issue : 2026;03(08)

Page No : 867-876

Rapid urbanization and population growth have increased the generation of sanitary and municipal solid waste, creating challenges for collection, treatment, disposal, and environmental protection. Sanitary waste streams may include biodegradable organic matter, paper, packaging, plastics, and other nonrecyclable or contaminated materials. If waste is not segregated and treated appropriately, it can contribute to land and water pollution, greenhouse-gas emissions, public-health risks, and loss of recoverable energy. Waste-to-energy (WtE) approaches can recover useful heat, electricity, or fuel from suitable waste fractions while reducing the volume of material requiring final disposal. This report presents a conceptual design for a Sanitary Waste-to-Energy System that integrates waste segregation, preprocessing, conversion, energy recovery, and monitoring. The proposed system separates biodegradable and combustible fractions from recyclables and hazardous or biomedical materials. Depending on feedstock characteristics, conversion pathways may include anaerobic digestion for wet biodegradable waste, controlled thermal treatment for suitable dry combustible fractions, or refuse-derived fuel (RDF) preparation. Energy conversion is coupled with appropriate gas cleaning, emission control, residue handling, and safety monitoring. An IoT-based monitoring layer is proposed for measuring process temperature, pressure, moisture, pH, gas composition, flow, and energy output, with data logging and alarms through a local controller and cloud dashboard. The report discusses the motivation, literature, problem definition, existing and proposed systems, architecture, requirements, waste-conversion methodology, monitoring and cloud integration, operational reporting, algorithm, implementation plan, evaluation metrics, limitations, future scope, and references. The appropriate conversion route depends on local waste composition, contamination, moisture, scale, regulatory requirements, and economic feasibility. No specific plant capacity, measured energy yield, emission result, or experimental performance is claimed because project-specific feedstock data and test results were not supplied.

Keywords : Sanitary waste management, waste-to-energy, anaerobic digestion, biomass, refusederived fuel, thermal conversion, IoT monitoring, biogas, energy recovery, pollution control.

Author : Dr K RAJESH , D HATHIRAM , G RAMESH

Title : AN IOT-ENABLED SMART SANITARY WASTE-TO-ENERGY CONVERSION SYSTEM FOR SUSTAINABLE WASTE MANAGEMENT

Volume/Issue : 2026;03(08)

Page No : 877-887