A video analysis system for object tracking and zone intrusion detection with configurable parameters.
- 🎯 Object tracking with history visualization
- 🚨 Zone intrusion detection and alarm triggering
- ⚙️ Fully configurable through YAML settings
- 📹 Video input/output processing
- 📊 Custom channel region configuration
model_path: 'checkpoints/best.pt'
video_path: 'video/test.mp4'
output_path: "outputs/output_custom_tracker_best/result.mp4"
channel_region:
- [409, 1113]
- [870, 1080]
- [1745, 1500]
- [630, 1520]
track_history_length: 30
alarm_threshold_seconds: 5
save_debug_frames: False
fps: null
channel_region
: Polygon coordinates defining surveillance areaalarm_threshold_seconds
: Duration threshold for triggering alertstrack_history_length
: Number of frames to visualize tracking pathmodel_path
: Pretrained object detection model
- Install dependencies:
pip install -r requirements.txt
-
Configure settings in
config/config-v2.yaml
-
Run main script:
python main.py
MIT License