How quzanh1130/object_tracking_trackline works
This is a command-line object-tracking application: a user supplies video or camera input, model weights, a device, and a display mode; the runtime detects selected objects, assigns tracks, and displays bounding boxes and optionally path lines. The sampled script shows the runtime orchestration and visualization state. Model and tracker internals are dependencies, not present in the supplied tree, so their internal stages are not expanded.
Main entry points: object_tracking.py CLI and Runtime · yolov9-c-converted.pt YOLO Weights · classes.names Class Names
Components 11
Inputs
- Video or Camera
- YOLO Weights
weight/yolov9-c-converted.pt - Class Names
data_ext/classes.names - OpenCV Capture
Detection Tracking
- YOLO Detector
- DeepSORT Tracker
Visualization
- Trackline State
object_tracking.py - Frame Renderer
object_tracking.py - OpenCV Display
Other components
- User
- CLI and Runtime
object_tracking.py
Connections 12
- User → CLI and Runtime: configures
- User → Video or Camera: provides
- User → YOLO Weights: selects
- CLI and Runtime → OpenCV Capture: opens source
- OpenCV Capture → CLI and Runtime: returns frames
- CLI and Runtime → YOLO Weights: loads
- CLI and Runtime → Class Names: reads
- CLI and Runtime → YOLO Detector: runs inference
- CLI and Runtime → DeepSORT Tracker: updates tracks
- CLI and Runtime → Trackline State: updates locations
- CLI and Runtime → Frame Renderer: draws overlays
- CLI and Runtime → OpenCV Display: shows frames