Shoplifting Detector

Upload a surveillance video (max 30 seconds) to detect theft in real time

Shoplifting Detector

Pose estimation pipeline that labels each person in a surveillance video as Suspicious or Normal, frame by frame.

Python FastAPI YOLO11 XGBoost OpenCV

What this proves

  • Real-time pose estimation using YOLO11 keypoint detection
  • XGBoost classifier trained on UCF-Crime shoplifting dataset
  • Frame-by-frame Suspicious / Normal labeling with bounding boxes
  • RTSP live stream support for real surveillance cameras
  • Telegram alert integration on theft detection
🎥

Drop video here or click to browse

MP4 · AVI · MOV · MKV  ·  max 30 seconds

Processing… this may take up to 60 seconds

Frames
Suspicious
Process time