Smart Sangraha

Smart Sangraha is an AI-driven crowd management system tailored for high-density Indian public spaces like temples and festivals. Inspiration The project draws from tragic stampedes at events like the 2024 Hathras incident and Maha Kumbh overcrowding challenges, aiming to prevent such disasters through proactive tech in India's crowded cultural hotspots. What it Does Real-time CCTV analysis with YOLOv8 detects crowd density, tracks flows at entry/exit points, and predicts stampede risks via ML models, while IoT sensors monitor environmental factors like temperature for alerts to authorities. How We Built It Integrated OpenCV for video processing, Flask/React for dashboards, and Arduino-based sensors ; trained custom YOLO models on Indian crowd datasets Challenges We Ran Into Handling occlusions in dense crowds (95%+ accuracy goal), and fusing thermal/environmental data without false positives during night/festive lighting. Accomplishments Achieved sub-30-second risk predictions, seamless hardware-software integration for Kochi Metro pilot, and open-source GitHub repo with live demo dashboard, setting a benchmark for cost-effective public safety What We Learned Edge AI optimizes for real-time needs over cloud reliance; diverse Indian datasets are crucial for model robustness; iterative GitHub workflows with Copilot accelerated debugging in late-night sessions What's Next Scale to multi-camera drone feeds, partner with KMRL for metro deployment, and integrate predictive analytics for 2027 Kumbh-like events.