How srey-pov/real-time-object-detection-tracking-and-counting works
This project detects cans and plastic bottles from images, a webcam, or recorded video. Its main processing paths are image inference, live webcam detection, and video tracking/counting; the video path counts ByteTrack identities crossing a virtual line and saves annotated video and a summary. The README also describes comparing five YOLO models, while the shown runtime scripts use YOLO11s. The graph focuses on implemented workflows; comparative evaluation is documented, but no evaluation implementation appears in the tree or excerpts.
Main entry points: detect_image.py Image detection · count_video.py Tracking and counting · models Trained YOLO models · real_time.py Live detection
Components 9
Detection workflows
- Image detection
object_detection/src/detect_image.py - Live detection
object_detection/src/real_time.py
Video counting
- Tracking and counting
object_detection/src/count_video.py
Models and results
- Trained YOLO models
object_detection/models - Annotated images
object_detection/output - Counted video and summary
object_detection/output/counting
Other components
- Image input
- Webcam
- Video input
Connections 8
- Image input → Image detection: reads image
object_detection/src/detect_image.py - Image detection → Trained YOLO models: loads YOLO11s
object_detection/src/detect_image.py - Image detection → Annotated images: writes image
object_detection/src/detect_image.py - Webcam → Live detection: supplies frames
object_detection/src/real_time.py - Live detection → Trained YOLO models: loads YOLO11s
object_detection/src/real_time.py - Video input → Tracking and counting: supplies frames
object_detection/src/count_video.py - Tracking and counting → Trained YOLO models: loads and tracks
object_detection/src/count_video.py - Tracking and counting → Counted video and summary: writes video and summary
object_detection/src/count_video.py