International Journal of Science, Engineering and TechnologyLangDoc: A Hybrid AI-Powered Framework for Multi-lingual Document Understanding in Non-Searchable Visual Formats
Authors: Aman Verma, Prince Aaryan, Mr. Ankur Kaushik
Abstract: Multilingual document processing presents considerable challenges, particularly when content exists in non-searchable visual formats across diverse scripts and languages. Traditional optical character recognition (OCR) systems frequently encounter cascading errors in such complex environments, limiting accessibility and automated comprehension . The problem intensifies for languages with unique orthographies or those considered low-resource, where digital tools and datasets remain scarce . This document introduces LangDoc, a hybrid artificial intelligence (AI)-powered framework engineered to overcome these linguistic and accessibility barriers. LangDoc deviates from conventional "flat" OCR systems by adopting a novel Script-First approach. This methodology prioritizes accurate visual script identification as an initial, critical step, mitigating error propagation in subsequent processing stages . The architecture integrates a fine-tuned YOLOv8 model for robust visual script identification, dynamically routing to a specialized Tesseract OCR engine for precise text extraction . For multilingual interpretation, the system employs the M2M100 Many-to-Many Transformer, enabling direct translation across over 100 languages . Furthermore, the incorporation of Google Gemini 2.5 Flash augments the framework with context-aware reasoning and a conversational interface, facilitating interactive document comprehension . Experimental evaluations demonstrate significant reductions in Word Error Rate (WER) and superior Bilingual Evaluation Understudy (BLEU) scores, particularly for regional Indian languages, thereby validating the efficacy of this integrated approach.
DOI: http://doi.org/
Phishing Site URL Detection System
Authors: Khushi Agarwal, Vimal Kartik, Professor Shubhi Verma
Abstract: The rapid growth of internet usage has led to a significant rise in phishing attacks, posing serious threats to user security and data privacy. Traditional detection methods, such as blacklist-based approaches, are often ineffective against newly emerging and sophisticated phishing websites. This research presents a Machine Learning-Based Phishing Website URL Detection System designed to identify malicious URLs in real time with high accuracy. The proposed system utilizes multiple machine learning algorithms and extracts key features from URL structures and domain characteristics to effectively distinguish between legitimate and phishing websites. It is integrated with a user-friendly web interface that enables instant URL analysis and prediction. Experimental results demonstrate that the system achieves reliable performance across diverse scenarios, providing a scalable and efficient solution for enhancing web security. The proposed approach reduces reliance on traditional methods and offers proactive protection against evolving phishing threats.
DOI: http://doi.org/
Comparative Study of YOLOv8 and SSD for Real-Time Pothole Detection
Authors: Professor N.K.Gawade, Anushka Jadhav, Aditi Kumari, Yugandhara Topalmode, Pallavi Khamkar
Abstract: Road surface monitoring is a critical component in ensuring transportation safety and reducing vehicle damage. This paper presents the implementation and performance evaluation of two deep learning-based object detection models, YOLOv8 and Single Shot Detector (SSD), for real-time pothole detection. The proposed system is capable of processing both static images and live video streams, enabling automated road inspection. A custom dataset of road images captured under varying environmental conditions was used for training and testing the models. Experimental results demonstrate that YOLOv8 significantly outperforms SSD in both accuracy and real-time detection capability. YOLOv8 achieved an accuracy of 95%, whereas SSD attained 80% accuracy. Furthermore, during real-time video testing, YOLOv8 exhibited superior detection consistency by identifying a higher number of potholes with fewer missed detections compared to SSD.
Wanderlust – A Web-Based Travel Listing and Review Application
Authors: Professor S. P. Gunjal, Aniket Dhokle, Rohidas Kudnar, Chandrkant Solanke
Abstract: Wanderlust introduces a web-based travel planning guide system designed to offer personalized and interactive travel experiences. Leveraging modern web technologies, the platform provides users with tailored recommendations for destinations, attractions, accommodations, and activities based on individual preference. Through an intuitive interface and zestful features, travelers can efficiently plan their trips, access expert insights, and discover unique local experiences. Wanderlust is created using various web development technologies such as HTML, CSS, JavaScript, MySQL etc. It aims to revolutionize the way people explore destinations by offering a seamless and user-centric solution that enhances the overall travel experience by also providing cross-platforming and plan their trips according to their personal budgeting.
Job Hook- An Intelligent Job Aggregator Portal With AI-Based Recommendation Engine
Authors: Dr. A. C. Sawant, Aniket Solanke, Laxman Ghodke, Rohan Gore, Sumit Lohar
Abstract: The Job Aggregator Portal is an intelligent platform that consolidates job listings from multiple sources to simplify the employment search process for users. It leverages APIs such as Adzuna, Jooble, and JSearch to provide real-time, up-to-date job information across various domains. The portal incorporates AI-ML– based resume analysis to recommend suitable jobs based on user skills and experience, enhancing personalization and efficiency. Built using React for the frontend and Spring Boot for the backend, the system ensures seamless data management, user authentication, and fast performance. By integrating smart algorithms and modern web technologies, the Job Aggregator Portal significantly improves the job search experience for users while assisting companies in reaching the right candidates efficiently. Furthermore, the Job Aggregator Portal addresses major challenges in the job search ecosystem—such as data redundancy, fake listings, and inefficient filtering—by implementing real-time synchronization, data validation, and intelligent ranking algorithms. The system’s modular architecture allows easy scalability and integration of additional APIs, ensuring adaptability to evolving market trends. It also emphasizes data security and privacy by using secure API communication and encrypted user authentication. The AI-driven recommendation engine continuously learns from user interactions, improving job relevance over time. Additionally, the platform provides valuable insights through analytics dashboards, helping recruiters understand user preferences and optimize postings.
AR-School
Authors: Professor C. P. Lachake, Anirudh Mohite, Chinmay Bomble, Vedant Gund, Professor S. P. Gunjal
Abstract: Education is undergoing a major transformation through the integration of immersive and intelligent technologies. Augmented Reality (AR) and Virtual Reality (VR) provide new ways to visualize, interact, and understand complex concepts, while Artificial Intelligence (AI) introduces personalized and adaptive learning. This paper presents a student prototype of an AI-Powered Virtual School Environment developed using Unity and AR Foundation. The system enables learners to interact with 3D educational objects through AR and hand-gesture controls, enhancing engagement and accessibility. This prototype aims to make learning interactive, inclusive, and effective, particularly for remote learners and students with disabilities. Future extensions include AI-driven adaptive modules, voice-based learning, and VR-classroom integration.
Medical Data Privacy and Security in Wireless Networks via Smart Health Card
Authors: Mr. S. P. Gunjal, Ganesh Maddewad, Ankur Takale, Kapil Belure
Abstract: In the present generation, healthcare has become the foremost imperative sector in today's medicinal eon. The massive private documents, responsive details are kept in a scalable manner. The healthcare industry has become more competitive in the digital world. As a thriving industry, it's challenging for doctors to understand the moving technology in the healthcare sector. This also deals with the patient’s nursing and maintains their portfolios. The overview of the project depicts a role played by the doctors, patients, management, and resource suppliers by implementing cloud- technology in the healthcare industry. The platform was designed and developed for user-friendly interactions where patients can connect with the management and doctors at any corner of the world. The peculiarity of the project was to withdraw the pen-paper method followed by the sector for ages. Cloud computing (CC) has played a vital role in the project that helped and managed to store, secure large data files. The features while operating the system were QR codes, generating e-mails, SMS text, and free-trunk calls. This approach assists on track with each individual's health-related documents, henceforward approving with the doctors to access the knowledge throughout the flow of emergency and firmly access policy. Besides the facts, it rescues the lifetime of the patients and mutually helps the doctors figure it out comfortably. The utilization of mobile aid applications may be a dynamic field and has received the attention of late. This development provides mobile technology additional enticing for mobile health (m-health) applications. The m-health defines as wireless telemedicine involving the utilization of mobile telecommunications and multimedia system technologies and their integration with mobile health care delivery systems. As well as human authentication protocols, whereas guaranteeing, has not been straightforward in light-weight of their restricted capability of calculation and remembrance.
Yoga Pose Estimation and Prediction
Authors: Dhruv Gajanan Lokhande, Abhishek Mukesh Agarwal, Shivraj Hanumant Akhade, Kailas Anil Yadav, Assistant Professor Chandani P. Lachake
Abstract: Physical activity is essential for maintaining health and fitness. Recent advancements in computer vision and machine learning have enabled highly accurate human pose estimation models that can track and analyze body movements in real time. This paper reviews state-of-the-art techniques in pose estimation and prediction, highlighting their applications in sports science, rehabilitation, and computer animation. It also presents the development of a virtual fitness trainer using MediaPipe, integrated with deep learning frameworks like TensorFlow and Keras. The system leverages posture estimation algorithms to detect key body landmarks, enabling accurate tracking of exercises and performance evaluation. By analyzing user movements, it provides feedback and predicts future poses to improve exercise quality.With the rapid evolution of deep learning techniques, pose estimation models continue to improve in accuracy and efficiency. These advancements hold strong potential to transform fitness and health monitoring by enabling smarter, real-time movement analysis and personalized training systems.
Attendo: BLE-Based Smart Attendance System
Authors: C.P. Lachake, Pratima Wadikar, Sujal Bele, Vishal Gupta, Ankit Metkari
Abstract: This paper introduces a prototype Bluetooth Low Energy (BLE)-based smart attendance system developed to overcome common technical and operational challenges in educational environments, including cost, scalability, hardware dependence, and hygiene concerns. The proposed system utilizes two dedicated mobile applications: one for teachers and another for students. In contrast to conventional BLE-based models where students broadcast signals, this design reverses the communication direction. The teacher’s device functions as the sole BLE beacon, transmitting a session-specific encrypted UUID, while student devices operate as passive scanners that detect proximity and send encrypted check-in requests to a central server. The system emphasizes energy efficiency, secure data exchange, and minimal infrastructure requirements, ensuring reliable operation even in large classrooms. Experimental testing demonstrates strong performance, high reliability, and ease of use. Comparative analysis with existing methods and id limitations validate its potential as a scalable and low-maintenance alternative to traditional
Explainable Deepfake Detection System Using Xceptionnet, Grad-Cam, and Mediapipe
Authors: Mr.S.P.Gunjal, Yash Kohakade, Adarsh Gurekar, Sakshi Gadkar
Abstract: Deepfake technology has emerged as a major threat to digital media authenticity in recent years. Deepfakes are manipulated images or videos generated using deep learning models that can closely resemble real human faces. These fake media contents are often misused for spreading misinformation, identity theft, and reputation damage. Although many deep learning models have been developed for deepfake detection, most of them work as black-box systems and do not provide any explanation for their predictions. This paper presents an Explainable Deepfake Detection System for facial images using XceptionNet, Grad-CAM, and MediaPipe Face Mesh. The proposed system is implemented as a Python Flask web application with SQLite as the backend database. XceptionNet is used as the classification model to detect whether an image is real or fake. Grad-CAM is applied to generate heatmaps highlighting suspicious facial regions, and MediaPipe Face Mesh maps these regions onto facial landmarks to provide meaningful visual and textual explanations. The system offers real-time detection with interpretable results, making it suitable for forensic analysis, journalism, and educational purposes. Experimental results show that the proposed system achieves high detection accuracy while maintaining transparency and usability.
AMMVMS: An Accessible Multi-Modal Adaptive Virtual Mouse System
Authors: Professor A. C. Sawant, Aditya Deshmukh, Kaustubh Kale, Deepak Agawane, Kunal Jadhav
Abstract: The Accessible Multi-Modal Adaptive Virtual Mouse System (AMMVMS) is proposed as an inclusive, low-cost, and software-driven alternative that utilizes Artificial Intelligence (AI) and Computer Vision (CV) to redefine digital accessibility. The system integrates gesture-based spatial control using OpenCV and MediaPipe with voice-based contextual control via the SpeechRecognition library. Together, these modules enable users to interact with computers through simple hand movements and spoken commands, eliminating the need for physical touch or precision devices. The Adaptive Control Layer of AMMVMS further enhances usability by incorporating intelligent filters such as tremor smoothing, click cooldowns, and context-based mode switching (Normal, Gaming, and Entertainment).
On the Daksh-Majumder-Balaji Geometry
Authors: Daksh Jain, Soham Majumder, Amitesh Balaji, Daksh Saksena, Amitesh
Abstract: We introduce the Daksh-Majumder-Balaji (DMB) geometry, a pro-posed noncommutative geometric model for the quantum vacuum. The construction is defined by a Golden-ratio spectral dimension DDMB = 3 + ϕ−2, a fractional Dirac operator, and a ϕ−2-scaled deformation of the Moyal product on spacetime. We formulate the associated spectral triple, define the metric structure, and derive the foundational quantum mechanical framework using the Dirac–von Neumann axioms expressed on the Hilbert space HD . A preliminary formulation of the spectral ac-tion is presented, establishing the pathway toward extracting gravitational and gauge dynamics. This work provides the structural basis for future physical predictions arising from the DMB spectral geometry.
DOI: http://doi.org/
A QR Code Attendance System And Management
Authors: Mr.S.P.Gunjal, Prashant Singh, Sudarshan Vekhande, Rupali Dhamale, Ashlesha Gadade
Abstract: This paper details the design and implementation of the QR Attendance System, a contemporary, web-based solution intended to modernize and streamline student attendance management. The system is engineered around a mobile-first, dark-mode user interface (UI) developed using React.js and TailwindCSS, ensuring optimal usability and speed via device camera scanning (html5-qrcode). The back-end, powered by Flask/Python, processes attendance in realtime, applying a key feature: configurable, time-based logic to categorize status as On Time, Late, or Very Late. Attendance records are stored in a simple, portable CSV format for easy data manipulation and reporting. This architecture prioritizes cost-effectiveness, high performance, and institutional policy adherence, presenting an efficient alternative to traditional and more complex biometric attendance methods. its scalability and performance optimization, the system prioritizes cost-effectiveness, security, and maintainability by relying on open-source technologies and lightweight architecture. Thssis approach significantly reduces infrastructure complexity while maintaining high accuracy and reliability in data processing. Comparative analysis with traditional biometric, RFID, and manual attendance methods highlights the proposed system’s superiority in terms of deployment speed, adaptability, and user engagement. Furthermore, the system supports institutional customization, enabling administrators to configure academic schedules, attendance thresholds, and reporting periods. Future enhancements may include cloud-based data storage, machine learning-driven analytics for attendance trend prediction, and integration with Learning Management Systems (LMS) for automated grading and participation tracking.
Design and Development of an AI-Powered DSA Practice and Learning Platform
Authors: Prof. A. C. Sawant, Mayuri Bhujbal, Harshada Salunke, Harshada Ahire, Dhruv Mene
Abstract: This paper presents a comprehensive DSA (Data Structures and Algorithms) practice platform designed to enhance problem-solving skills through personalized learning and performance-driven feedback. The system integrates user and admin modules, enabling efficient management of topics, quizzes, coding problems, and assessments. Users can learn concepts, attempt MCQ-based quizzes, and solve coding problems, earning experience points (XP) to encourage engagement. A machine learning based recommendation engine analyzes user performance, including accuracy, time complexity, and coding efficiency, to suggest optimized learning paths and identify weak areas. Additionally, an AI powered code analysis module evaluates submissions and provides improvement insights. The platform includes coding tests covering multiple topics, generating detailed performance summaries to guide learners effectively. Supporting features such as real-time notifications, flash notes, admin-user chat support, and topic-based community discussions further enrich the learning experience. The proposed system aims to bridge the gap between theoretical understanding and practical coding proficiency by offering an adaptive, interactive, and scalable environment for DSA preparation.
Scholaraid: Government Scholarship & Scheme Recommender
Authors: Associate Professor C.P. Lachake, Nitin Naik, Pranit Jatal, Swapnil Kharade, Vineeth Ghabak
Abstract: Government schemes and scholarships are often difficult for citizens to find because information is scattered across many official websites. SCHOLARAID is a Flask-based web application that solves this problem by bringing all scheme and scholarship details into one centralized platform. The system collects verified data through manual web scraping and stores it in a structured database for easy access. Users can search and filter schemes based on category, eligibility, state, or income level. The platform also includes an AI- powered chatbot that helps users find suitable schemes through simple, natural conversation. By making information easy to understand and access, SCHOLARAID supports Digital India and helps citizens quickly discover the benefits they are eligible for.
Green Fund Connecting Farmers and Investors Through Transparent Crowdfunding
Authors: Prof. Chandani Lachake, Omkar Memane, Tushar Rathod, Kshitij Watmare
Abstract: With the increasing financial challenges in the agricultural sector, the need for transparent, efficient, and accessible funding solutions has become critical. Traditional financing methods often involve intermediaries, lack transparency, and place significant financial risk solely on farmers. This paper presents GreenFund: A Direct Farmer–Investor Crowdfunding Platform with Real-Time Profit Tracking, a system designed to bridge the gap between farmers and investors through a secure and transparent digital platform. The proposed system enables farmers to publish detailed agricultural projects, including crop information, funding requirements, and expected returns, while allowing investors to directly fund these projects. The platform integrates real-time monitoring and tracking mechanisms to analyze project progress and calculate profit or loss based on actual outcomes. It incorporates features such as automated notifications, secure transaction management, and risk visibility to enhance trust and decision-making. The system is capable of reducing dependency on traditional intermediaries, improving financial accessibility for farmers, and providing investors with a structured and transparent investment environment. Experimental results indicate that the proposed solution enhances operational efficiency, reduces manual effort, and offers a scalable and cost-effective approach for agricultural funding. The architecture is designed to be flexible and suitable for both small-scale farmers and large agricultural ecosystems.
Smart Agriculture: Image Processing-Based Automated Detection of Cotton Leaf Diseases
Authors: Prof. C. P. Lachake Girish Khamkar, Kishor Handge, Aryan Kakade, Siddhant Gadkari
Abstract: Nowadays in India Cotton is considered one of the most important cash crops i.e. White Gold, as most farmers cultivate cotton in large numbers. Agriculture plays a vital role in the economic development of our country. Farmers face various challenges due to unexpected weather changes and plant diseases that reduce crop yield and quality. Cotton, one of the major cash crops, is highly affected by leaf diseases, which are often difficult to detect at an early stage using manual methods. To overcome this issue, the proposed system focuses on automating the detection and classification of cotton leaf diseases using image processing and machine learning techniques. In this project, images of cotton leaves are captured and processed to identify disease symptoms such as color variation, texture, and shape. Using trained machine learning models, the system can accurately classify the type of disease and provide information about suitable remedies or preventive measures.
ConnectMe – A Networking and Socializing App Using Flutter and Firebase
Authors: Prof. A. C. Sawant, Vaishnavi Kale, Harshal Saudagar, Anant Paymode, Kartik Borude
Abstract: In today's world, making meaningful connections with like-minded individuals has become increasingly difficult due to a lack of efficient platforms for real-time networking and collaboration. The ConnectMe app aims to bridge this gap by leveraging modern technologies, transforming the process of finding, connecting, and interacting with others based on shared interests such as travel, movies, treks, or other activities. The core aim of our project is to create a platform that allows users to seamlessly find individuals with similar passions and meet up with them in real-time. The traditional methods of connecting with others, like social media platforms, often lack the personalization and spontaneity needed for real-life meetups. ConnectMe addresses these issues by providing an easy-to-use mobile app built with Flutter and backed by Firebase, which enables realtime interactions. connections, notifications, and The mobile application includes features such as instant activity matching, real-time notifications, and seamless user interaction. The app employs Firebase for authentication, data management, and event notifications, while Flutter provides a crossplatform solution for mobile development.
An AI-Based Document Analyzer Using Gemini API for Conversational Knowledge Retrieval
Authors: Prof. S. P. Gunjal, Kunal Jagtap, Hrishikesh Joshi, Bhavya Bhat, Anushka Gaikwad
Abstract: The rapid growth of digital documents across domains such as research, business, and education has created significant challenges in efficient information extraction. Traditional methods relying on manual reading or keyword-based search lack contextual understanding and are time-consuming. This paper presents PaperSense, an AI-powered document analyzer that enables users to upload documents and interact with them using natural language queries. The system integrates the Google Gemini 2.5 Flash model for contextual understanding, summarization, and question answering. Built using Python and Streamlit, the system follows a modular architecture separating frontend and backend processing. Experimental evaluation demonstrates improved efficiency in document comprehension and faster information retrieval. The system provides a scalable and user-friendly solution for intelligent document analysis.
Lab2Life: Medical Report Summarization Using Artificial Intelligence
Authors: Assistant Professor C.P.Lachke, Madhura Shete, Roshan Bhuruk, Jitendra Paramar, Asmita Ghalme
Abstract: Medical reports are really difficult to patients for understand because of complex language and numerical data. This paper presents Lab2Life AI-based web application that simplifies medical reports into understandable summaries. The system use OCR to extract text from reports and NLP to generate clear explanations. It also provides multilingual support and a doctor verification feature to ensure accuracy. Lab2Life helps users better understand their health informatMedical reports are really difficult to patients for understand because of complex language and numerical data. This paper presents Lab2Life AI-based web application that simplifies medical reports into understandable summaries. The system use OCR to extract text from reports and NLP to generate clear explanations. It also provides multilingual support and a doctor verification feature to ensure accuracy. Lab2Life helps users better understand their health information and improves communication between patients and healthcare providers.ion and improves communication between patients and healthcare providers.
Current Trends in Hair Growth Research and Evidence-Based Hair Care Solutions
Authors: Dr.G.V.Khandekar, Manasvi Gaikwad, Vaishnavi Baheti, Sukanya Nagare, Harshada Sidankar
Abstract: Hair is a crucial aspect of personal hygiene and self esteem, but there are people who need help with that information, personalized solutions that are easy to access. The paper discusses the design and development of Haircare Application which people can use all ages, men, women and children . The application combines features like a tailored hair care routine, hairstyle recommendations, home remedies, AI-based approach for your hairs analysis, salon appointments, doctor consultations and progress check-ins from the doctors in practice as well as an on-demand marketplace to explore haircare products. Features It uses AI technology to analyze hair conditions and suggest appropriate treatments and routines. Users will have the ability to keep track of a schedule calendar, progress with hair from day-to-day basis, and receive expert you-need to suit your needs, exactly when needed. The application endeavors to merge technology with daily haircare necessities, establishing a digital one-stop solution that elevates the customer experience and encourages convenient healthy habits while acting as an interactive bridge between individuals, experts and products. It addresses accessibility as well as promotes informed decision-making in haircare management.
NASA API Implementation
Authors: Prof. A.C.Sawant, Ghule Seema, Ghodke Kiran, Yadav Sayali, Sonwane Mayank
Abstract: This paper presents the integration of publicly available APIs provided by NASA to develop a web-based application for accessing and visualizing space-related data. The study focuses on APIs such as Astronomy Picture of the Day (APOD), Mars Rover Photos, and Near-Earth Object (NEO) data. The objective is to demonstrate how RESTful API integration can be implemented using modern web technologies including HTML, CSS, and JavaScript. The system retrieves real-time data from NASA servers, processes JSON responses, and displays meaningful information to users in a structured format. The application improves user engagement by providing interactive and dynamic content related to space exploration. The results indicate that NASA APIs are reliable, efficient, and suitable for educational and research-based applications. Challenges such as API rate limits, data handling, and network dependency are also addressed. The paper concludes that API integration plays a vital role in building scalable and real-time application.
Aion-Healthcare
Authors: Om Lohar, Pranav Shelar, Rudra Saraswate, Mohit Jamadar, Prof. C. P. Lachake, Prof. S. P. Gunjal
Abstract: AION is an AI-driven healthcare ecosystem designed to bridge the gap between demanding occupational routines and personal well-being. Unlike generic fitness apps, AION leverages machine learning to deliver occupation-specific hyper-personalization, tailoring diet, exercise, and mental health interventions to the unique physical and psychological stressors of diverse roles—from sedentary IT professionals to labor-intensive agricultural workers. The platform integrates a multi-layered support system featuring automated medication tracking, real-time progress monitoring, and proactive stress-reduction modules. By synthesizing behavioral data with AI-driven insights, AION provides a holistic, adaptive framework that optimizes long-term health outcomes and occupational performance.
Gesture Controlled Music Player Using Hand Gestures
Authors: Professor S.P.Gunjal, Pranali Yemale, Jeel Makwana, Tanushree Kshirsagar
Abstract: This paper presents an AI-powered music player that enhances user interaction by integrating gesture recognition and voice-based control with conventional playback mechanisms. The system is developed as a full-stack web application using React for the frontend, Flask for the backend, and MongoDB for data management. It incorporates secure user authentication through JWT and Bcrypt to ensure safe access. The application supports dual music modes, enabling users to play locally uploaded songs stored on Cloudinary or stream music via Spotify integration. A key contribution of this work is the implementation of real-time hand gesture recognition using MediaPipe, allowing users to control playback functions such as play, pause, track navigation, and volume adjustment without physical contact. Additionally, a voice assistant based on the Web Speech API facilitates natural language commands for seamless interaction. The system intelligently switches between local and online sources, ensuring uninterrupted playback. By combining computer vision, speech recognition, and API-based streaming, the proposed solution offers an innovative and user-friendly approach to interactive media systems, highlighting the potential of multimodal interfaces in modern applications.
Voice Based Mail System for Visually Impaired Using AI/ML
Authors: Prof. A. C. Sawant, Anish Mavkar, Omkar Dalvi, Pranav Gonte, Vivek Yewale
Abstract: In today’s digital era, email communication plays a vital role in personal and professional interactions. However, traditional email systems are not easily accessible to visually impaired users due to their dependency on graphical interfaces and manual input methods. The proposed system, Voice Based Mail System, provides a solution by enabling users to perform all email-related operations using voice commands. The system utilizes Speech-to-Text (STT) and Text-to-Speech (TTS) technologies to allow seamless interaction without the need for a keyboard or screen. It integrates Artificial Intelligence (AI), Natural Language Processing (NLP), and facial recognition techniques to improve command understanding, enhance security, and provide a user-friendly experience. The system supports functionalities such as composing emails, reading inbox messages, and managing email operations entirely through voice interaction.
E-Governance Based Gram Panchayat Management System Using Blockchain Technology
Authors: Prof. S. P. Gunjal, Sakshi Takhik, Shraddha Totre, Pratiksha Pawar, Shubham Ahire
Abstract: The E-Governance Based Gram Panchayat Management System Using Blockchain Technology aims to enhance transparency, security, and efficiency in rural governance by digitizing Panchayat operations. Traditional systems often face challenges such as data manipulation, lack of accountability, and delays in service delivery. This proposed system integrates blockchain technology to provide a decentralized and tamper-proof platform for managing records, transactions, and citizen services. It enables functionalities such as certificate issuance, grievance handling, and scheme management through a secure and user-friendly interface. Smart contracts are used to automate processes, ensuring faster and reliable service delivery. Overall, the system improves trust, reduces corruption, and promotes effective and transparent administration at the Gram Panchayat level.
GradeMate: A MERN-Stack Platform for Secure Online Examinations with Real-Time Monitoring and AI-Assisted Grading
Authors: Associate Professor S. P. Gunjal, Ritesh Kashid, Krishna Gangurde, Aarush Prabhudesai
Abstract: The widespread adoption of remote learning has intensified the demand for reliable, scalable, and intelligent online examination platforms. This paper presents GradeMate, a full-stack web application built on the MERN (MongoDB, Express.js, React.js, Node.js) technology stack, designed to streamline the complete lifecycle of academic assessments. GradeMate provides distinct authenticated portals for administrators, teachers, and students, enabling role-specific workflows from quiz creation and batch management to real-time examination proctoring and automated result generation. A key contribution of GradeMate is its integration of the OpenAI API for AI-assisted grading of subjective answers, significantly reducing the evaluative burden on educators while maintaining consistent scoring standards. Real-time communication is achieved via Socket.IO, allowing live monitoring of active examination sessions. The platform employs a Glassmorphism-based responsive UI framework and enforces security through JSON Web Tokens, BcryptJS password hashing, Helmet middleware, and Express Rate Limiting. Empirical evaluation demonstrates that GradeMate reduces manual grading effort by an estimated 60% and supports concurrent examinations across multiple academic batches with negligible latency. The architecture, design decisions, feature set, and security considerations of GradeMate are discussed in detail.
Al Powered Presentation Builder
Authors: Assistant Prof S.P.Gunjal, Harshada Chawla, Rohit Satav, Tanmay Kakulte, Nayan Salunke
Abstract: In the modern digital era, presentations have become an essential tool for communication in education, business, and professional environments. However, creating an effective presentation requires not only subject knowledge but also skills in content organization, design, and time management, which many users find challenging. Traditional presentation tools demand significant manual effort to structure information, select layouts, and maintain visual consistency. To overcome these limitations, the concept of an AI Powered Presentation Builder is introduced, which utilizes Artificial Intelligence and Natural Language Processing to automatically generate well-structured, visually appealing presentations from simple user inputs. This approach aims to simplify the presentation creation process, enhance productivity, and make high-quality presentation design accessible to all users.
Farmer’s Smart Assistant System
Authors: Prof.S.P.Gunjal, Shivraj Kakde, Rohan Kale, Atish Kharde, Rushikesh Mate
Abstract: Farmer Helper App is a mobile-based solution developed to simplify identification of agrochemical products. Application integrates Artificial Intelligence (AI) with Optical Character Recognition (OCR) to allow farmers to capture images of product labels using smartphones. Captured data is processed to extract essential information including usage instructions, recommended dosage, suitable crops, and safety precautions. System architecture consists of a Python FastAPI backend combined with a Flutter-based frontend, ensuring efficient performance and user-friendly interaction. Offline functionality is incorporated, enabling usability in rural regions with limited internet connectivity. This solution provides a cost-effective and reliable approach that supports farmers in making informed and safer agricultural decisions.
MedLink: A Healthcare Management System Using AI & ML Models
Authors: Associate Professor A.C. Sawant, Devendra Meshram
Abstract: In the modern digital era, presentations have become an essential tool for communication in education, business, and professional environments. However, creating an effective presentation requires not only subject knowledge but also skills in content organization, design, and time management, which many users find challenging. Traditional presentation tools demand significant manual effort to structure information, select layouts, and maintain visual consistency. To overcome these limitations, the concept of an AI Powered Presentation Builder is introduced, which utilizes Artificial Intelligence and Natural Language Processing to automatically generate well-structured, visually appealing presentations from simple user inputs. This approach aims to simplify the presentation creation process, enhance productivity, and make high-quality presentation design accessible to all users.
AutoDS: Simplified Data Analysis and Machine Learning Modelling
Authors: Assistant Professor A. C. Sawant, Satyam Shende, Jayesh Mahajan, Alfaj Tamboli
Abstract: The AutoDS project report presents a web application designed to simplify and automate machine learning (ML) model training and deployment for users with limited coding expertise. The platform supports various traditional ML and deep learning models, automates data preprocessing, exploratory data analysis (EDA), model training, evaluation, and report generation, making ML accessible across diverse domains. This review articulates the project's motivation, design, methodology, applications, advantages, limitations, and future scope, highlighting its contribution to democratizing ML workflows.
Coin Hub: A Full-Stack Cryptocurrency Information and Management Platform
Authors: Assistant Professor A. C. Sawant, Shirish Kanoje, Sunny Bhujbal, Yash Kanchan, Mehafuj Pathan
Abstract: The increasing adoption of cryptocurrencies has transformed digital finance, leading to the need for reliable, real-time information systems for users and investors. However, most existing applications provide either limited data visualization or lack personalization and security. This paper presents Coin Hub, a full- stack cryptocurrency information and management platform designed to deliver seamless access to live coin data, user specific dashboards, and a secure, responsive interface. The system is built using Java Spring Boot for backend development and React.js for frontend implementation, supported by a MySQL database for structured storage and RESTful APIs for smooth communication. The application provides real-time and intuitive visualization of cryptocurrency trends, offering a practical demonstration of end-to-end web engineering. Extensive testing through Postman, load-testing tools, and frontend validation confirmed the robustness, scalability, and synchronization, secure authentication and responsiveness of the system. Coin Hub demonstrates the integration of modern full-stack technologies for efficient information systems in the evolving world of digital finance.
Labo Link
Authors: Prof. C. P. Lachake, Karan Khumkar, Kunal Jadhav, Shivam Lahane, Sahil Khadse
Abstract: The LabourLink application is a web-based platform developed to connect users with skilled labourers in their nearby area. In many cases, people face difficulty in finding reliable workers such as electricians, plumbers, carpenters, and painters when needed. This project aims to solve that problem by providing a simple and efficient digital solution. The system allows users to search for professionals based on their location and service type. It uses location-based technology to display nearby labourers on a map and in a list format. Users can view profiles, check ratings, and contact or book the worker directly through the application. At the same time, labourers can register on the platform, showcase their skills, and receive job opportunities. The application is developed using modern technologies such as React.js for the frontend, Node.js and Express.js for the backend, and MongoDB for database management. It also includes features like filtering, real-time updates, and a labour cost calculator. Overall, LabourLink provides a user-friendly, reliable, and fast solution for connecting customers with skilled workers. It helps save time, improves accessibility to services, and supports employment opportunities in the local labour market.
Elevate AI”: Enhanced Learning Experience & Virtual Assistant Tool for Education
Authors: Associate Prof. Chandani Lachake, Rushikesh Wani, Shivraj Ghodake, Pritesh Chaudhari
Abstract: AI tools are widely used by students for studying, but most of them are general-purpose and don’t follow university-specific formats, syllabus, or marking schemes. Because of this, students may understand concepts but still struggle to score well in exams. To address this, we propose ELEVATE AI, a system designed for exam-oriented learning that generates structured answers using syllabus content, textbooks, previous year questions, and diagrams. It also allows students to practice writing answers and get feedback by comparing with ideal responses. With added voice interaction, the system can act like a tutor, making learning more interactive. Overall, it focuses not just on understanding, but also on how to write and present answers effectively in exams.
AI-Based Banana Disease Detection Using Deep Learning and Transfer Learning
Authors: Associate Prof Chandani Lachake, Shreyas Gogawale, Mahesh Sakhare
Abstract: This study proposes an automated system for the early identification of banana plant diseases using Deep Learning and Transfer Learning techniques. By leveraging pre-trained convolutional neural network (CNN) architectures—such as ResNet or MobileNet—the model effectively classifies common pathologies like Black Sigatoka, Panama Wilt, and Banana Bunchy Top Virus from leaf imagery. Transfer learning is utilized to overcome the limitations of small datasets, ensuring high feature extraction accuracy while significantly reducing training time. The integrated approach achieves superior classification performance compared to traditional manual inspection, providing a scalable solution for small-scale farmers.
A Study on Modern Student Grievance Portals and Evidence-Based Approaches to Efficient Complaint Management
Authors: Prof. S. P. Gunjal, Sujal Landge, Samruddhi Suryawanshi, Shrusti Layane, Prasad Kusmude
Abstract: The Student Complaint Portal replaces the frustrations of outdated, manual grievance systems with a streamlined digital platform built for the modern campus. By moving away from physical paperwork and "black-hole" bureaucracy, the system allows students to submit concerns online, track their progress in real-time, and communicate directly with the right authorities. Through smart, role-based access, complaints are automatically routed to the appropriate departments, ensuring that issues are addressed promptly rather than sitting on the wrong desk. Beyond just fixing individual problems, the portal maintains a structured database that helps administrators analyze trends and identify systemic issues. Ultimately, this platform transforms the institutional culture into one that is more transparent, accountable, and—most importantly—genuinely responsive to the student experience.
Artificial Intelligence Enabled Virtual Try-On System for Enhancing User Interaction and Personalization in E-Commerce Applications
Authors: Assistant Professor Mrs Ashwini Sawant, Ritika Gupta, Prajwal Shinde, Veena Suse, Siddhi Sali
Abstract: Virtual try-on technology has transformed online fashion retail by enabling users to visualize garments digitally, enhancing shopping confidence and reducing return rates. Modern Virtual try-on systems leverage deep learning, computer vision, and pose estimation to simulate realistic garment-person interactions. Early GAN-based methods faced limitations in garment deformation and identity preservation, whereas recent diffusion-based models, notably IDM-VTON, demonstrate superior image fidelity and adaptability. Benchmark datasets like VITON-HD and DressCode support objective evaluation using perceptual and semantic metrics. Despite substantial progress, challenges remain in modeling fine details, handling occlusion, and extending to multimodal and 2D representations. This study reviews key VTON components, emerging trends, and future research directions aimed at achieving more accurate, inclusive, and deployable virtual try-on solutions
Genface: Forensic Face Sketch Construction and Recognition
Authors: Assistant Professor Siddhi Shinde, Rajeshwari Ambekar, Narendra Kakde
Abstract: Traditional forensic sketching is hindered by subjectivity and a lack of digital interoperability. This paper introduces GenFace, a JavaFX-based application that enables rapid composite sketch assembly via a modular drag-and-drop interface. The system’s primary innovation is the integration of an AWS Rekognition backend to facilitate Heterogeneous Face Recognition. By matching user-generated sketches against criminal databases in real-time, the system provides ranked similarity scores to investigators. This approach bridges the gap between eyewitness testimony and digital suspect identification, significantly enhancing investigative efficiency.
Plant Disease Detector Application
Authors: Asst. Professor C.P.Lachake, Tushar Yadav, Siddhisri Kannekanti, Shuti Raut, Pranjal Narsale
Abstract: Agricultural productivity plays a vital role in the economic stability of many developing countries. However, the rise of plant diseases causes significant losses to farmers and agricultural industries every year. Early detection and accurate identification of plant diseases are essential to prevent these losses. This paper proposes a Plant Disease Detector Application, a mobile-based system designed using the Flutter framework integrated with a Convolutional Neural Network (CNN) model built using TensorFlow and Keras. The app enables users to capture images of infected plant leaves and instantly identify the disease along with suggested preventive measures. The proposed model is trained on the PlantVillage dataset, achieving a testing accuracy of over 97%. The integration of TensorFlow Lite allows the model to run efficiently on smartphones, even without internet connectivity. This application demonstrates a scalable, cost-effective solution for farmers, researchers, and agricultural institutions to promote sustainable farming through technology-driven management
Development of AI/ML Based Solutions for Deep Fake Detection
Authors: Prof. Chandani Lachke, Anushka Singh, Snehal Bhosarkar, Pradnya Jambhale
Abstract: Deep learning has proven effective in a variety of tough issues, including computer vision, human-level control, and large data analytics. However, as deep learning technology advanced, software was developed that jeopardized national security, democracy, and privacy. Deepfake is a new technology that uses deep learning to create fake photos and videos that look very real. It's important to have tools that can automatically detect and check the quality of these AI-created images and videos. These systems help us quickly tell if a picture or video is real, edited, or fake, and they ensure that the quality is good and not misleading. An investigation of the strategies used to construct the most significant deepfakes, as well as the approaches proposed in the literature for detecting them. We provide a complete examination of the difficulties highlighted by deepfake technology, as well as recommendations for future and upcoming research opportunities. It also supports creating new and more reliable ways to handle deepfakes as they become more complex.
Fake Product Detection System Using Blockchain
Authors: Assistant Professor C.P. Lachake, Sujal Chavan, Sahil Vidhate, Pranav Kute., Jagdish Navpute
Abstract: The increasing presence of counterfeit goods across industries threatens consumer safety, brand trust, and supply chain transparency. Traditional anti-counterfeiting methods are prone to duplication and lack unified traceability. This paper presents an implemented blockchain-based system for fake product detection that ensures secure and tamper-evident product provenance. Each product is assigned a unique on-chain identity on a permissioned blockchain, with smart contracts managing lifecycle events such as manufacturing, transfer, and sale. Consumers can verify authenticity through QR/NFC-based access. Additionally, an image recognition module compares product visuals with verified references to detect anomalies. The system, developed using React and Solidity, is evaluated through simulated supply chain scenarios and image datasets. Results demonstrate low-latency verification, improved traceability, and robustness against common threats such as cloning and replay attacks. The approach provides a scalable and practical solution for enhancing trust in modern supply chains.
Employee Management System
Authors: Assistant Professor C.P.Lachake, Suraj Lanke, Ajit Zol, Pushkar Talele, Mahesh Sonawane
Abstract: Effective management of employee information is a critical requirement in modern organizations. Manual processing of attendance, leave records, payroll activities, and employee data often results in delays, redundancy, and limited data accuracy. This paper presents a web-based Employee Management System designed using ReactJS as the front-end interface and Spring Boot as the back-end service framework, with MySQL providing persistent relational storage. The system incorporates JSON Web Token (JWT) based authentication and role-based access control to ensure secure usage across administrative and employee user groups. The application provides centralized, real-time access to employee information and automates core Human Resource (HR) workflows, improving accuracy, transparency, and operational efficiency. The modular architecture of the system enables scalability and provides scope for future integration with biometric systems, cloud deployment, and predictive HR analytics.
AgriMesh: AgriRent and the Farmer Support Platform
Authors: Associate Professor A.C. Sawant, Suyog Kute, Ayush Kalbhor, Janmajay Dethe, Sonali Kshirsagar
Abstract: The increasing complexity of malware-based network attacks poses a serious challenge to modern communication systems. Traditional signature-based intrusion detection systems struggle to detect novel or evolving attack patterns. This paper presents a comprehensive machine learning framework for classifying malware network traffic using flow-based features. The study uses a dataset of 60,938 instances spanning six classes: normal traffic and five malware types (worm, rootkit, buffer overflow, ipsweep, and sqlattack). The dataset exhibits extreme class imbalance, with normal traffic comprising 99.43% of samples. To address this, we apply label encoding, one-hot encoding, stratified sampling, and feature standardization, and evaluate performance using accuracy, precision, recall, F1-score, confusion matrix, ROC analysis, and training time. Seven supervised classifiers from different learning paradigms—Logistic Regression, Decision Tree, K-Nearest Neighbors, Random Forest, Gradient Boosting, XGBoost, and LightGBM—are developed and compared. Experimental results show that ensemble methods significantly outperform single classifiers in detecting minority attack classes. XGBoost achieves the highest overall performance: 99.95% accuracy, 99.96% precision, 99.95% recall, and 99.95% F1-score, with a training time of 3.70 seconds. LightGBM provides the best trade-off between accuracy (99.79%) and speed (3.25 seconds), while Gradient Boosting requires substantially longer training (82.99 seconds). The findings confirm that ensemble learning is highly effective for malware traffic classification under severe class imbalance.
AI Based Gamified Educational Application
Authors: Prof. P. R. Chaugule, Swejal Vairagade, Dnyaneshwari Thopate, Aryan Bhagat
Abstract: Gamification in education has emerged as an effective approach to improve student engagement and learning outcomes. Traditional learning methods often fail to maintain student interest, resulting in reduced motivation and poor knowledge retention. This paper presents an AI-Based Gamified Educational Application named CodeVerse, designed to provide interactive learning through quizzes, leaderboards, rewards, and progress tracking. The system integrates AI-based question generation and performance analysis to personalize learning experiences. The application allows users to learn programming concepts through interactive gameplay. Features such as login authentication, topic-wise quizzes, score tracking, and leaderboard ranking improve user participation. The proposed system enhances learning efficiency, increases student engagement, and provides a competitive environment for learners.
Smart AI-Based Network Assistant Using Raspberry PI
Authors: Assistant Professor Chandani Lachake, Rupali Talekar, Aishwarya Salunke, Namrata Biradar, Kalyani Patekar
Abstract: With the rapid growth of modern network infrastructures, the need for intelligent, real-time monitoring and automated management systems has become increasingly important. Conventional network management approaches often rely on manual intervention and lack predictive capabilities, resulting in delayed fault detection and reduced system efficiency. This paper presents a Smart AI-Based Network Assistant Using Raspberry Pi, a plug-and-play solution designed for real-time monitoring, automation, and intelligent diagnosis of network infrastructure. The proposed system leverages the capabilities of Raspberry Pi integrated with artificial intelligence techniques to continuously analyze network parameters such as latency, packet loss, and bandwidth utilization. The system is capable of detecting anomalies, generating alerts, and performing automated corrective actions to minimize downtime and improve network reliability. Experimental results indicate that the proposed solution enhances system performance, reduces manual effort, and provides a cost-effective and scalable approach for network management. The architecture is compact, energy-efficient, and suitable for both small-scale and enterprise-level deployments.
COVID-19 Data Analysis and Forecasting Using Machine Learning and Time Series Models
Authors: Assistant Professor S. P. Gunjal, Tanaya Balasaheb Sandbhor, Nisha Sanjay Tekade, Sadichha Balshiram Talekar, Tejaswini Rajendra Pawar
Abstract: The COVID-19 pandemic has caused an unprecedented global health crisis, making accurate forecasting of case counts critically important for government planning and resource allocation. This paper presents a comprehensive data analysis and multi-model forecasting study on the Kaggle COVID-19 day-wise dataset spanning January 22 to July 27, 2020. We apply statistical time series models — ARIMA and Holt-Winters Exponential Smoothing — alongside machine learning approaches including Ridge Regression, Lasso, ElasticNet, and Random Forest Regressor. Data preprocessing includes Augmented Dickey-Fuller (ADF) stationarity testing, first-order differencing, 7-day rolling smoothing, and lag feature engineering. Models are evaluated using MAE, RMSE, and R² metrics with 80/20 temporal train-test split and 5-fold time-series cross-validation. Random Forest achieved the best performance with RMSE = 13,227 and R² = 0.9612. A 30-day future forecast with 95% confidence intervals is generated using the best-fit ARIMA(2,1,2) model. Results demonstrate that ensemble machine learning methods outperform classical statistical models for COVID-19 case prediction.
AI-Based Training and Placement Assistance
Authors: Prof. C. P. Lachake, Vivek Baraskar, Akshat Balgude, Sanket Konale, Kunal Bhagat
Abstract: In the present generation, focuses on developing a smart web-based system to support students in the placement process using Artificial Intelligence (AI) and Machine Learning (ML) techniques. The system analyzes student academic backgrounds, skills, and test performances to provide personalized placement recommendations and training guidance. It also matches eligible students with suitable job opportunities based on company criteria. This system includes a Skill Test module that evaluates the student’s technical and aptitude abilities through multiple-choice questions (MCQs). Based on the test scores and academic data, the system predicts whether a student is eligible for further placement rounds. The integration of AI ensures data-driven and fair decisions, reducing manual effort by placement officers and improving the efficiency of the recruitment process. The proposed solution bridges the gap between students and recruiters by providing an automated, intelligent, and reliable platform for training, assessment, and placement assistance.
Smart Hostel Automation and Management System for Educational Institutions
Authors: Prof. S.P. Gunjal, Phopale Tanaya Avinash, Yernale Shrikant Dhanraj, Chaudhari Mohit Ravindra, Ghadge Madhav Balaji
Abstract: The increasing demand for digitized infrastructure in educational institutions necessitates the development of efficient and scalable hostel management solutions. This paper presents the design, development, and evaluation of a Smart Hostel Management System, a full-stack web-based platform engineered using modern technologies such as React.js, Node.js, Express.js, and MongoDB. The proposed system aims to automate and optimize core hostel operations, including student registration, authentication, room allocation, complaint management, and attendance monitoring, through a centralized and interactive interface. Conventional hostel management approaches are predominantly manual and fragmented, leading to data redundancy, limited accessibility, and operational inefficiencies. In contrast, the proposed system leverages a database-driven architecture and RESTful APIs to ensure real-time data synchronization, secure user authentication, and efficient resource management. The system incorporates role-based access control, enabling distinct functionalities for administrators and students, thereby enhancing system usability and security. Additionally, dynamic room allocation and complaint tracking mechanisms improve responsiveness and transparency in hostel operations. The proposed solution is evaluated in terms of usability, scalability, and performance, demonstrating significant improvements over traditional methods. The findings indicate that the adoption of a web-based hostel management framework not only reduces administrative overhead but also enhances data integrity and user experience. The system provides a cost-effective and extensible solution adaptable to diverse institutional requirements, contributing to the advancement of smart campus infrastructure.
Retino AI: A DR Detection
Authors: Prof. A.C.Sawant, Dangi Ritisha, Jaykar Gayatri, Dudhate Kavita, Samuel Kumarkunta
Abstract: Diabetic Retinopathy (DR) is a serious eye disease caused by prolonged diabetes and is one of the leading causes of blindness worldwide. Early detection is crucial to prevent vision loss, but manual screening is time-consuming and requires expert ophthalmologists. This paper presents a deep learning-based approach for automatic detection and classification of diabetic retinopathy using retinal fundus images. Convolutional Neural Networks (CNN) are employed to extract features and classify images into different stages such as No DR, Mild, Moderate, Severe, and Proliferative DR. The proposed model is trained on publicly available datasets and optimized using preprocessing techniques such as image normalization, augmentation, and noise reduction. Experimental results show high accuracy and reliability, demonstrating the effectiveness of deep learning in medical image analysis. This system can assist healthcare professionals in early diagnosis and improve patient outcomes.
Agri Management System
Authors: Pranav Pawar, Rameshwar Jadhav, Nishita Jawale, Swapnali Kopnar, Prof . C . P . Lachake
Abstract: The Agri Management System is a digital platform designed to support farmers by integrating multiple agricultural services into a single application. The system consists of key modules such as Dairy Management, Farm Management, Veterinary Services, and Tractor Services. The Farm Management module provides updated information on the rates of pesticides and fertilizers from various shops, enabling farmers to make cost-effective decisions. The Veterinary Services module helps farmers locate nearby veterinary doctors and facilitates direct communication through calls or messaging. Additionally, the Tractor Services module connects farmers with tractor owners, displaying service availability and pricing details for different types of agricultural work. The Dairy Management module assists in handling dairy-related activities efficiently. Overall, the system aims to enhance productivity, reduce operational challenges, and promote better decision-making for farmers by providing real-time information and easy access to essential agricultural services.
Notes2Card: AI-Based Study Content Generation
Authors: Professor S.P.Gunjal, Professor C.P.Lachake, Dandale Vaishnavi, Gawai Yash, Pal Babita, Deshmukh Ashutosh
Abstract: The rapid growth of digital learning has increased the need for tools that can convert broad topics into structured and revision-friendly study material. This project presents Notes2Card, an AI-assisted web application that helps learners generate complete study resources from a user-provided topic and learning purpose. The system creates a course outline with chapter-wise structure, generates detailed chapter notes, and supports additional study formats such as flashcards and quizzes for active recall and self-assessment. Notes2Card is built using Next.js for the application layer, Clerk for user authentication, Neon PostgreSQL with Drizzle ORM for data persistence, Gemini API for AI-based content generation, Inngest for asynchronous background processing, and Stripe for subscription-based membership management. The platform follows a modular architecture with separate APIs for course creation, content retrieval, study-type generation, user status, and payment of workflows. It also includes robust AI error classification and status tracking to improve reliability during content generation. The implemented system supports both free-tier and paid-tier users, with server-side enforcement of usage limits for non-members. Experimental usage of the application demonstrates that the platform can reduce the manual effort required to prepare structured study material while improving accessibility and consistency of learning content. The project validates the practical use of AI and event-driven processing in building scalable educational tools.
Real-Time Elderly Monitoring System Using Deep Learning and Computer Vision
Authors: Sahil Jagtap
Abstract: With the growing elderly population, ensuring their safety, especially for those living alone, has become a critical concern. This paper presents a real-time elderly monitoring system that leverages computer vision and deep learning to detect falls and hazardous objects such as knives and guns. The system integrates YOLOv8 for fall detection and object recognition, OpenCV for real-time video frame processing, and Pushbullet API for instant caregiver notifications. A Flask-based user interface allows caregivers to monitor live annotated video feeds remotely, offering a non-intrusive alternative to wearable sensors that require user compliance.The proposed system achieves 95% accuracy in fall detection and 98% accuracy in hazardous object recognition, with an average alert delay of under 3 seconds. This approach ensures rapid emergency response, enhances elderly safety, and minimizes caregiver burden. Despite challenges such as privacy concerns and reliance on internet connectivity, the system provides an effective, scalable, and adaptable solution for use in homes, care facilities, and hospitals. Future enhancements include multi-camera integration, AI-based anomaly detection, and wearable device support for health monitoring to further improve elderly care and safety.
NetSentinel: A Machine Learning-Based Intrusion Detection Framework Using Random Forest
Authors: Associate Professor C.P. Lachake, Maitray Rangari, Sonawane Gaurav Vishwas, Ritik Palve, Shailesh Bhise
Abstract: This paper presents NetSentinel, a machine learning-based intrusion detection framework designed to identify malicious network traffic with high accuracy. The system utilizes a Random Forest classifier trained on the NSL-KDD dataset using an 80:20 train-test split. The proposed architecture integrates a Python-based machine learning pipeline with a Node.js backend and MongoDB database for efficient data handling and user interaction. Performance evaluation using metrics such as accuracy, precision, recall, and F1-score demonstrates strong detection capability. Additionally, confusion matrix and ROC curve analysis validate the robustness of the model. The system aims to provide a scalable, efficient, and reliable cybersecurity solution. Future enhancements include real-time traffic capture, streaming integration, and deployment in live network environments.
A Survey on Medical Diagnosis of Retinopathy and Detection Techniques
Authors: Aishwarya Ajit Padwalkar, Assistant Professor Dr. G. V. Khandekar, Assistant Professor Upendra Sinhg, Diksha Rawat
Abstract: Diabetic Retinopathy is a frequent complication of diabetes mellitus that leads to retinal lesions affecting vision. Without timely detection, it can result in permanent blindness. Sadly, Diabetic Retinopathy is irreversible, and medical interventions only help preserve existing vision. Early identification and treatment of DR play a crucial role in minimizing the risk of vision impairment. However, manual diagnosis of retinal fundus images by ophthalmologists is time-consuming, labor-intensive, costly, and susceptible to errors, unlike automated computer-aided diagnostic systems. This paper has summarized the type of retinopathy and various stages before getting blindness. Many of researcher proposed models that find the disease in early stages by analyzing medical images. Various techniques of image optimization, analysis and classification were discussed. Paper has summarized the image features that were used in different research article for identify the retinopathy image class.
Agentic AI Systems: Architecture, Challenges, and Applications
Authors: Riya Singh, Assistant Professor Dr. G. V. Khandekar, Assistant Professor Upendra Sinhg, Diksha Rawat
Abstract: Agentic Artificial Intelligence (Agentic AI) represents a paradigm shift in the field of artificial intelligence, enabling systems to operate autonomously with minimal human intervention. Unlike traditional AI models that rely on predefined instructions or reactive responses, Agentic AI systems are designed to perceive their environment, reason about complex situations, plan actions, and execute tasks in a goal-oriented manner. This paper provides a comprehensive overview of Agentic AI systems, focusing on their architecture, key challenges, and diverse applications. The architectural framework of Agentic AI integrates components such as perception, memory, reasoning, planning, and action modules within a closed-loop system that supports continuous learning and adaptation. The study also explores the wide range of applications across domains including healthcare, finance, robotics, and smart systems, where Agentic AI enhances efficiency, decision-making, and automation. However, the development of such systems presents several challenges, including issues related to trust, explainability, scalability, and computational complexity. Ethical and legal concerns, such as bias, privacy, and accountability, further complicate their deployment. This paper aims to provide a structured understanding of Agentic AI while highlighting the need for robust frameworks and responsible implementation strategies to fully realize its potential.
RideLink – An Android-Based Intelligent Carpooling System for Sustainable Urban Mobility
Authors: Associate Professor S. P. Gunjal, Sahil Hase, Pratik Jadhav, Abhishek Pokharkar, Siddhi Horane
Abstract: The increasing number of private vehicles in urban areas has resulted in serious transportation challenges such as traffic congestion, fuel wastage, and rising pollution levels. This paper presents the complete design, development, and evaluation of RideLink — an Android-based intelligent carpooling system engineered to promote shared mobility and sustainable commuting. RideLink allows users to register as drivers or passengers and enables them to create, search, and book rides in real time. The system integrates Firebase Realtime Database for secure data handling, osmdroid (OpenStreetMap) for GPS-based route visualization, and a Haversine geospatial algorithm for intelligent ride matching based on route proximity. The platform supports real-time seat occupancy tracking, gender-aware booking, driver license verification through an admin approval workflow, luggage capacity management, automated SMS notifications to drivers upon booking, an SOS emergency button, and complete ride lifecycle management. Testing across physical Android devices running API levels 26 through 34 demonstrated sub-2-second ride search performance and accurate geolocation-based matching. All 14 planned functional test cases passed successfully. RideLink demonstrates how mobile and cloud technologies can deliver a practical, scalable, and eco-friendly transportation solution without paid infrastructure.
TraffNet: Smart Traffic Management System Using Python With AI
Authors: S.P.Gunjal, Kanifnath Gite, Rutuja Mokhar, Vishakha Nadgiri, Dhanshri Bachkar
Abstract: Traffic congestion remains one of the most pressing urban challenges globally, leading to increased travel time, fuel consumption, and pollution. This paper presents a software-only implementation of a Dynamic Traffic Light System (DTLS) powered by Edge Machine Learning (ML) techniques. Using an optimized YOLOv3-tiny model for real-time vehicle detection and intelligent traffic light timing algorithms, the system adapts dynamically to traffic conditions across multiple junctions. Simulation results on standard datasets demonstrate reduced vehicle wait times and enhanced emergency response capabilities, making the solution highly scalable for Intelligent Transportation Systems (ITS) and smart city integration.
AI Based Resume Screening
Authors: Vaishnavi Dhaygude, Rushikesh Biradar, Aditya Bhosale, Chinmay Kadu
Abstract: In the recruitment process, organizations receive a large number of resumes for a single job opening, making manual screening time-consuming and inefficient. The AI Resume Screening system is an effective solution to automate the evaluation of candidate profiles and improve the hiring process. Models are used for analysis and are designed using Natural Language Processing and Machine Learning techniques. In comparison to the traditional manual screening method, the final result demonstrates that the AI-based system reduces processing time while improving accuracy, consistency, and candidate shortlisting efficiency.
Electric Vehicle Technology
Authors: Professor Deepali Vaidya, Amardip Bandu Raipure
Abstract: Electric Vehicle Technology has emerged as one of the most promising solutions for sustainable transportation in the modern world. Electric Vehicles (EVs) use electric motors powered by rechargeable batteries instead of conventional internal combustion engines that rely on fossil fuels. The increasing concerns regarding environmental pollution, global warming, and depletion of fossil fuel reserves have accelerated the development and adoption of EVs worldwide. Electric vehicles offer numerous advantages, including reduced greenhouse gas emissions, lower operating costs, improved energy efficiency, and decreased dependence on petroleum products. Recent advancements in battery technology, charging infrastructure, and power electronics have significantly enhanced EV performance and reliability. This research paper discusses the concept, history, components, working principles, advantages, disadvantages, applications, challenges, and future prospects of Electric Vehicle Technology.
DOI: http://doi.org/