Browse
Recent Submissions
Item Remote Code Execution Environments(2026-02-11) Abhinandan Sodhi 1NH22CS006; Ankith Anand 1NH22CS020; Saqib Hussain Dar 1NH22EC147Remote Code Execution (RCE) environments have become a foundational component of modern cloud computing platforms, online programming judges, AI-assisted development tools, continuous integration pipelines, and distributed evaluation systems. These environments allow users to execute code remotely without local setup, enabling scalability, automation, and accessibility. However, executing untrusted or user-submitted code introduces serious security, performance, and operational challenges. Vulnerabilities such as container escape, unauthorized network access, and supply-chain attacks pose significant risks to traditional RCE systems. Existing approaches based solely on containers, virtual machines, or serverless functions fail to provide an optimal balance between strong isolation, execution efficiency, and large-scale scalability.Item X-AI Enabled Hybrid Approach for Detection of Cyber Terrorism(2026-02-11) Adoni Mahammad Rehan 1NH22CS011; Avula Reddy Bharath Yadav 1NH22CS034; Rayampalli Prasen Akhil Babu 1NH22EC131; Pittu Mahendranath Reddy 1NH22EC114Cyber terrorism has become a significant threat in today’s connected world. Attackers use social media, networks, and online platforms to spread extremist content and carry out harmful activities. Traditional cybersecurity methods often struggle to keep up with the fast-changing behaviors and tactics of cyber terrorists. This project introduces a hybrid approach powered by X-AI that combines artificial intelligence with clear decision making to improve the detection and prevention of cyber terrorism. The proposed system uses several machine learning and deep learning algorithms, including BERT, LSTM, GRU, Naïve Bayes, and Random Forest. BERT captures the context of tweets and text data, while LSTM and GRU analyze sequence patterns. Ensemble models like Random Forest provide strong classification. Explainable AI tools such as SHAP and LIME offer transparency in predictions, helping analysts understand why a tweet or behavior is flagged as harmful.Item AI Based Nanorobot Assisted Endoscopic Imaging For Early Detection Of Gastric Disorders(2026-02-09) Chandresh M 1NH22CS054; Patan Izaz Khan 1NH22CS281; Agnes Arul 1NH22EC006; Akshayaasri S 1NH22EC009This project introduces a comprehensive intelligent system designed for the early identification of gastric disorders through AI-driven CT image analysis, paired with a prototype for drug delivery inspired by nanorobotics. The early detection of stomach tumors can be quite difficult due to the subtlety of symptoms and the intricacies involved in manually interpreting CT images. To bridge this gap, the proposed model utilizes a Convolutional Neural Network (CNN) created with Python 3.13 and TensorFlow, which automatically classifies CT images of the stomach and accurately identifies potential tumor regions. The system employs sophisticated image preprocessing methods, including noise reduction, contrast enhancement, and normalization, to enhance diagnostic reliability.Item Depression Detection from Audio Using Deep Neural Network(2026-02-09) Vidula S 1NH22CS243; Venkatesh Kumar MU 1NH22CS242; Harish S 1NH22EC059; Hari Sai Nikhil S 1NH22EC056The Tri-Modal Depression Detection and Counselling System introduces a novel approach to mental health monitoring by seamlessly integrating three critical modalities of human emotional expression speech, facial signals, and textual language into a single, browser-based platform. Similar to this, unimodal emotion recognition systems that rely solely on speech, facial expressions, or textual analysis are prone to difficulties such as occlusion, ambient noise, poor language, or insufficient contextual knowledge, resulting in unpredictable and unreliable predictions. To address these limitations, the proposed tri-modal system builds on advances in affective computing and artificial intelligence to integrate spoken emotion detection, facial affect analysis, and AI-driven textual therapy, resulting in a more comprehensive and robust picture of user emotionality.Item Smart Traffic Management System Using IOT and Machine Learning(2026-02-09) Soundarya R Raikar 1NH22CS213; Srushti 1NH22CS218; Abhijith U B 1NH22EC002; Ankitha B G 1NH22EC017Urban traffic congestion continues to be one of the most critical problems in fast growing cities, leading to longer travel durations, increased fuel consumption, higher pollution levels, and daily stress for commuters. Conventional fixed cycle traffic lights are unable to respond to constantly changing road conditions, which results in poor traffic movement and frequent bottlenecks. To address these shortcomings, this project presents a smart traffic management solution that combines Internet of Things (IoT) technology, embedded hardware, machine learning (ML), and cloud computing to optimize signal control, improve safety, and reduce congestion in real time. The system collects live traffic data using IR sensors, ultrasonic sensors, and camera modules to detect vehicle count, presence, and classification.Item Deep Learning-Based Facial Emotion Recognition Using EEG and CNN(2026-02-09) N Rakshitha 1NH22CS261; Shatakshi Pattanaik 1NH22EE099; Lakshmi Thrisha Pabolu 1NH22CS276; Sakshi SM 1NH22EE097Facial emotion recognition plays a vital role in human–computer interaction and mental-health monitoring, but image-based methods alone often struggle with issues like lighting variations, occlusions, and intentionally masked expressions. This work presents a multimodal deep-learning approach that combines facial images with EEG signals to achieve more reliable emotion detection. Facial features are extracted using a Convolutional Neural Network (CNN), while EEG signals undergo preprocessing and time frequency analysis to capture brain activity patterns linked to emotional states. The system fuses visual and EEG-based features to enhance classification accuracy and robustness.Item Path Pilot-Real Time Road Navigation System(2026-02-09) Geena Annu Alexander 1NH22CS078; Poornima V 1NH23CS412; Gowri Manojkumar 1NH22EE048; Iqra Bashir 1NH22EE053The rapid growth in road traffic, urban complexity, and dynamic travel patterns has in-creased the demand for intelligent navigation systems capable of responding instantly to real-world conditions. With the advancement of artificial intelligence, deep learning, computer vision, and simulation technologies, modern navigation systems are evolving toward perception-based, adaptive, and context-aware platforms. The present work, Path Pilot – Real-Time Road Navigation System, aims to design and implement an intelligent navigation framework that integrates real-time object detection, lane analysis, traffic interpretation, and optimized routing for safe and efficient mobility. Path Pilot combines three major technological pillars: computer vision–based perception, AI-driven decision making, and cloud/GPS-based navigation.Item Implementation of Ramp-CNN-Based Radar Camera Fusion Framework For Enhanced Missile Detection(2026-02-09) Nikhil LK 1NH23CS410; Shraddha G 1NH23CS417; Sanjay S 1NH23EC415; Shrinivas R M 1NH23EC416This project presents an advanced Radar–Camera Fusion framework for highly accurate and reliable missile detection using modern deep learning techniques. Traditional single-sensor systems such as radar-only or camera-only approaches often fail under challenging conditions involving noise, clutter, poor visibility, or rapidly moving targets. To overcome these limitations, the proposed system integrates FMCW radar processing with real-time computer vision, combining their complementary strengths. Radar signals are transformed using 3D Fast Fourier Transform (3D-FFT) to generate Range Azimuth (RA), Range Velocity (RV), and Velocity Azimuth (VA) heatmaps, which are processed using Convolutional Autoencoders and a RAMP-CNN architecture to extract discriminative spatial–temporal features. In parallel, the camera subsystem enhances visual frames through preprocessing operations and employs YOLOv8 for high-precision object detection. A feature-level fusion module merges radar and camera features into a unified representation, enabling superior classification and localization of missile targets.Item AI-Backed Risk Intelligence for Decentralized Credit Rating(2026-02-09) Likith P Reddy 1NH22CS118; Manoj M 1NH22CS132This projectpresentsanAI-drivencreditscorepredictionsystemintegratedwith block chain technology to ensure data transparency, privacy, and security. The system analyses key financial attributes such as income, savings, liabilities, and expenditure patterns to predict an individual’s credit score using advanced machine learning models Including Random Forest(RF), XGBoost, and LSTM. A user-friendly web application, developed using HTML, CSS, and Flask, allows users to register, login, and submit their financial details.Item Namma Farmer an AI Drivern Agricultural Platform For Farmer Empowerment(2026-02-09) Mohammed Ummar Khan 1NH22CS137; Sasank Reddy Ν 1NH22CS142; Razik Shariff 1NH22CS178; Dileep Kumar V 1NH22CS240Agriculture is evolving, calling for technology that not only provides information but also guides farmers toward better decisions. Namma Farmer is a farmer-friendly digital platform designed to deliver real-time insights, trusted knowledge, and trading support through a single, easy-to-use system. The platform brings together advisory services, marketplace features, and data analytics to simplify daily agricultural activities and reduce dependency on multiple disconnected applications. It aims to bridge the information gap between rural farmers and modern agricultural practices, making advanced guidance accessible to everyone.Item Lip Reading and Speech Conversion(2026-02-09) Himanth Reddy 1NH22CS155; Punith NB 1NH22CS170; Sreyas S 1NH22CS215; Srinivas 1NH22CS216Lip movements contain rich visual information that can be interpreted to understand speech without relying on audio signals. This project introduces a comprehensive Lip Reading and Speech Conversion System that transforms silent video input into meaningful linguistic output. The system is designed to support users in challenging acoustic environments and individuals with hearing difficulties by providing an alternate, vision-based pathway for speech interpretation. The proposed system processes an uploaded video through several machine-learning components, beginning with frame extraction and facial landmark identification. A dedicated deep-learning model isolates and analyzes lip-region dynamics to capture temporal patterns associated with spoken words.Item AI-Rescue Force(2026-02-09) Nivetha Madhavan 1NH22CS150; Harshitha M Anil Kumar 1NH22CS257; Mahi Srivastava 1NH22CS127; Vishal Shetty 1NH23CS422With their growing frequency and unpredictable nature, disasters represent a serious threat to infrastructure and human life, necessitating quicker, more intelligent, and more coordinated responses. In order to automatically identify different disasters like floods, wildfires, and earthquakes from satellite and sensor data, this research suggests a multi-disaster detection and alarm system powered by deep learning, specifically Convolutional Neural Networks (CNNs). The suggested method provides a scalable, cross-disaster platform that not only classifies occurrences but also assesses their severity based on confidence scores, occurrence frequency, and location spread, in contrast to conventional systems that concentrate on a single hazard type and lack real-time responsiveness.Item RescueReach – Emergency Assistance Locator(2026-02-09) Koushik Kumar M S 1NH22CS115; Shashiraj 1NH22CS199; Suman S 1NH22CS222; Tharun Kumar K 1NH22CS230The rapid growth of urban populations and mobile connectivity has fundamentally changed how emergency assistance systems must operate. While traditional emergency response methods rely heavily on phone calls and manual coordination, these approaches often lead to delays, miscommunication, and uncertainty during critical situations. This creates a strong need for modern, technology-driven platforms that can improve response efficiency while ensuring user safety and reliability. This project, RescueReach – Emergency Assistance Locator, is developed to address these challenges. It aims to provide a real-time, scalable, and user-centric solution for emergency reporting and response coordination.Item TALKBRIDGE- A Real-Time Speech Translation Network(2026-02-09) N Vignesh Reddy 1NH22CS139; S Charitha Devi 1NH22CS182; Trupti L 1NH22CS232; Varshini P 1NH22CS239In today’s globally connected society, people interact across regions, cultures, and languages more than ever before. However, linguistic differences still interrupt smooth communication in education, business, travel, and daily conversations. When speakers do not share a common language, communication often becomes slow and unclear, leading to misunderstandings and reducing opportunities for effective collaboration. To overcome these challenges, TALKBRIDGE is designed as a real-time speech translation system that allows individuals to communicate naturally even when they speak different languages.Item A Hybrid CNN and Gradient Boosting Framework for Spatio-Temporal Prediction Of Urban Crime Spots(2026-02-09) Akshara Nair 1NH22CS014; Devarshi Khilariwal 1NH22CS065; Sohani Nagappa 1NH22CS256; Ashmith Maddala 1NH22CS273Urban crime continues to pose significant challenges to public safety and policy enforcement, necessitating predictive systems that are both accurate and actionable. This research proposes a hybrid framework integrating Convolutional Neural Networks (CNNs) with Gradient Boasting techniques specifically CatBoost and XGBoost for the spatio-temporel prediction of urban crime hotspots. By leveraging historical crime dats enriched with spatial, temporal, and socio-demographic features, the system generates accurate forecasts of high-risk zones. CatBoost nancies categorical features such as crime types and locations, while XGBoost captures trends from numerical variables like population density and socioeconomic indicators. These models are ensemble-combined to improve predictive reliability. Additionally, CNNs are applied to heatmap-transformed data to extract spatial and temporal crime patterns.Item PRIME WHEELS – A Smart Taxi Management Solution(2026-02-09) R Seshusai 1NH22CS172; Purvaj Dev M 1NH22CS171; T Vinay Kumar 1NH23CS421; Gurushant J Matti 1NH23CS405Prime Wheels is developed as a smart taxi management system to handle common problems seen in today's urban transportation, such as poor coordination, uneven service quality, and confusing pricing. In many cities, traditional taxi operations still depend on manual processes and limited communication, which leads to delays and dissatisfaction for both customers and drivers. This project proposes a single digital platform that brings together all key activities of a taxi service so that bookings, monitoring, and payments become easier to manage and more transparent for everyone involved.The application is organised into clear functional modules, including customer and aniver management, admin control, booking, billing, and feedback. A relational database is used to connect these modules and to store all records in a structured way.Item PronouncePro AI- Automatic Pronunciation Mistake Detection System(2026-02-09) M Bharathi 1NH22CS121; Shashank R 1NH22CS198; Sohan R Poojary 1NH22CS211; Prajwal Raj A R 1NH23CS413Pronunciation plays a crucial role in achieving clear and effective spoken communication. Many non‑native speakers, however, struggle with accurate pronunciation even when they possess good vocabulary and grammar skills. Traditional classroom teaching and widely used speech tools rarely provide real‑time, phoneme‑level feedback, and they are often limited to a single language or tied to paid platforms. As a result, learners find it difficult to identify exactly which sounds they mispronounce and how to correct them independently. This project, PronouncePro AI, presents an intelligent, browser‑based pronunciation learning and feedback system designed for modern language learners. The system captures speech through the Web Audio API, performs local preprocessing in the browser, and then uses a cloud‑based AI engine to carry out phoneme alignment, prosody analysis, and fluency scoring.Item Al-Integrated Focus Learn(2026-02-09) M Vashishta Varma 1NH22CS124; P Akshay Reddy 1NH22CS152; Tejas P 1NH22CS228; Jambu Jahnavi 1NH22CS260Al-Integrated Focus Learn is a distraction-free e-learning platform designed to enhance focus, personalization and engagement in digital learning. Unlike traditional video platforms, it organizes educational content into structured learning joumeys where users can add individual YouTube videos or import entire playlists that are automatically converted into chapters for systematic learning. To promote focused study, the platform uses a custom video player that removes ads, recommendations and unnecessary interface elements. An integrated note-taking panel allows learners to write, save, and downioad notes alongside videos, supporting active leaming and better retention. The platform incorporates Al-powered features, including automatic note generation and a 24/7 Al tutor that helps users clarify doubts instantly.Item HEARTIQ-Smart Heart Disease Prediction System(2026-02-09) Akash Verma 1NH22CS013; Aryan Ganesh 1NH22CS028; Aum Patel 1NH22CS158; Siddhant Paradkar 1NH22CS208The Smart Heart Disease Prediction System is an AI-driven health platform developed to assist in the early identification of cardiovascular risks using patient-reported symptoms and basic clinical data. Designed with usability, accessibility, and accuracy in mind, the system enables users to input parameters such as age, gender, blood pressure, cholesterol levels, and chest pain types. Leveraging machine learning algorithms, the backend analyzes this input to generate an immediate risk assessment, helping users and healthcare providers make informed decisions. The platform prioritizes data privacy and ensures secure processing through authenticated access and encrypted data handling. Built using Python, Flask, and a trained predictive model, the system provides a lightweight and responsive interface accessible across devices. Future upgrades include integration with wearable devices, continuous monitoring support, and personalized health recommendations. This system serves as a cost-effective tool to raise awareness, enhance preventive care, and support early medical intervention, especially in resource-limited settings.Item Edumorph -A Quiz Driven Adaptive Learning System Using Dynamic Learner Classification(2026-02-09) Bhumika Murthy 1NH22CS045; Bincy Mary P 1NH22CS047; Utkarsh Bhardwnaj 1NH22CS234; Vikrant Singh 1NH22CS245EduMorph is an adaptive learning system designed to overcome the limitations of traditional e-learning platforms that follow a fixed, one-size-fits-all instructional approach. The system employs diagnostic quizzes to evaluate learner performance through accuracy, response behaviour, and conceptual strengths, and uses K-Means clustering to automatically categorize students into proficiency groups. Based on these clusters, EduMorph generates a dynamic, real-time learning roadmap that delivers personalized notes, videos, and practice exercises, ensuring mastery-based progression.