🦅 Birds of Play
Advanced Motion Detection with DBSCAN Clustering
🎥 Real-time Motion Detection Demo
Watch DBSCAN clustering in action with live bounding box visualization
🎬 Upload Your Bird Video
Upload a video of birds to run the complete DBSCAN motion detection pipeline!
or click to browse your computer
Latest Pipeline Test Results
Full pipeline test completed successfully with DBSCAN implementation!
📹 Motion Detection
65 frames processed from test video
All frames had motion regions detected
🔗 DBSCAN Clustering
3 motion objects → 1 consolidated region
244×259 pixels containing 17 objects
🎯 YOLO11 Detection
74 regions extracted and analyzed
25 high-confidence bird detections
🧠 ML Pipeline
4 bird species clusters identified
80% test accuracy on trained model
🔍 DBSCAN Clustering
Advanced clustering algorithm that groups motion objects based on overlap-aware distance metrics. Perfect for handling overlapping bounding boxes like small birds inside larger motion regions.
📊 No Size Constraints
Regions are created purely based on spatial clustering without artificial size limitations. This allows for more natural grouping of motion objects.
🎯 Smart Distance Calculation
Combines bounding box overlap ratio with edge-to-edge distance for intelligent motion object grouping. Weighted combination ensures optimal clustering results.
🔄 Full Pipeline Integration
Seamlessly integrated with MongoDB storage, YOLO11 detection, machine learning clustering, and web visualization interfaces.
🚀 Deployed on Vercel | ⚡ Powered by Modern C++ & Machine Learning
Latest commit: ff85cd8 - DBSCAN clustering implementation