Rahul
Takale

Designed and studied an AI-powered CCTV surveillance solution for crowd management and public safety during large-scale religious gatherings. Explored the use of Python, OpenCV, object detection, people counting, and crowd density estimation techniques for analyzing CCTV video streams. My contribution included designing the system workflow, studying computer vision models for crowd monitoring, analyzing real-time alert mechanisms, and evaluating how AI can assist authorities in identifying overcrowded zones and unusual movement patterns. A key challenge was understanding scalable surveillance requirements for handling large volumes of video data, which was addressed through independent research on computer vision, video analytics, and smart monitoring systems.

Developed an AI and IoT-powered aeroponics system for sustainable urban farming that delivers nutrient-rich mist directly to plant roots, achieving up to 95% water savings compared to traditional agriculture. Built a web-based monitoring platform to visualize real-time sensor data, including temperature, humidity, nutrient levels, and water conditions. Integrated an LLM-powered plant assistant using Ollama to provide intelligent recommendations on plant health, nutrient management, and crop maintenance. My contribution included designing the system architecture, developing the monitoring dashboard, integrating IoT sensor data pipelines, and implementing the AI assistant for decision support. The project was recognized among the Top 5 teams in an international innovation challenge featuring participants from South Africa, South Korea, USA, and Brazil.

Developed a computer vision-based fitness posture detection system using Python, OpenCV, and MediaPipe Pose to analyze body movements during exercises. Implemented pose landmark detection and angle calculation techniques to track joint positions and evaluate exercise form in real time. Designed logic for posture validation, movement tracking, and repetition counting to provide corrective feedback during workouts. My contribution involved integrating pose estimation models, processing live video streams, and developing algorithms to identify incorrect posture patterns. A key challenge was achieving accurate posture analysis across different body positions, which was addressed through extensive testing and optimization of landmark-based calculations.

CampusCare is a full-stack campus complaint management platform built using React.js, Spring Boot, and MySQL. The system enables students to submit and track complaints with image evidence while providing administrators with a centralized dashboard for complaint management, status updates, analytics, and report generation. Implemented JWT-based authentication, role-based access control, complaint lifecycle tracking, CSV export, and interactive dashboards to improve transparency and operational efficiency across campus services.