RizwanMunawar/yolov7-object-tracking GitHub

YOLOv7 Object Tracking 🚀

💥 Ultralytics YOLOv8 support added python detect.py --weights yolov8n.pt

🚀 YOLOv9, YOLOv10, YOLO11, YOLO12, YOLO13 support coming soon :)

How to Run the Code

Clone the repository:

git clone https://github.com/RizwanMunawar/yolov7-object-tracking.git

Navigate to the cloned folder:
```
cd yolov7-object-tracking
```

Create a virtual environment (Recommended to avoid conflicts):

For Anaconda:

conda create -n yolov7objtracking python=3.12
conda activate yolov7objtracking

For Linux:

python3 -m venv yolov7objtracking
source yolov7objtracking/bin/activate

For Windows:

python3 -m venv yolov7objtracking
cd yolov7objtracking/Scripts
activate

For MacOS:

python3 -m venv yolov7objtracking
source yolov7objtracking/bin/activate

Update pip and install dependencies:

pip install --upgrade pip
pip install -r requirements.txt

Run the script:

Select the appropriate command based on your requirements. Pretrained yolov7 weights will be downloaded automatically if needed.

Detection only:

python detect.py --weights yolov7.pt

# If you want to use your own videos.
python detect.py --weights yolov7.pt --source "your video.mp4"

# For Inference with YOLOv8
python detect.py --weights yolov8n.pt

Object tracking:

python detect_and_track.py --weights yolov7.pt

# If you want to use your own videos.
python detect_and_track.py --weights yolov7.pt --source "your video.mp4"

Webcam:

python detect_and_track.py --weights yolov7.pt --source 0

External Camera:

python detect_and_track.py --weights yolov7.pt --source 1

IP Camera Stream:

python detect_and_track.py --source "your IP Camera Stream URL" --device 0

Specific class tracking (e.g., person):

python detect_and_track.py --weights yolov7.pt --source "your video.mp4" --classes 0

Colored tracks:

python detect_and_track.py --weights yolov7.pt --source "your video.mp4" --colored-trk

Save track centroids, IDs, and bounding box coordinates:

python detect_and_track.py --weights yolov7.pt --source "your video.mp4" --save-txt --save-bbox-dim

Output files will be saved in runs/detect/obj-tracking with the original filename.

Arguments details 🚀

Argument	Type	Default	Description
`--weights`	`str`	`yolov7.pt`	Path(s) to model weights (`.pt` file).
`--download`	`flag`	`False`	Download model weights automatically.
`--source`	`str`	`None`	Source for inference (file, folder, or `0` for webcam).
`--img-size`	`int`	`640`	Inference image size in pixels.
`--conf-thres`	`float`	`0.25`	Object confidence threshold.
`--iou-thres`	`float`	`0.45`	Intersection over Union (IoU) threshold for NMS.
`--device`	`str`	`''`	CUDA device (e.g., `0` or `0,1,2,3`) or `cpu`.
`--view-img`	`flag`	`False`	Display results during inference.
`--save-txt`	`flag`	`False`	Save results to `.txt` files.
`--save-conf`	`flag`	`False`	Save confidence scores in `.txt` labels.
`--nosave`	`flag`	`False`	Do not save images or videos.
`--classes`	`list[int]`	`None`	Filter results by class (e.g., `--classes 0` or `--classes 0 2 3`).
`--agnostic-nms`	`flag`	`False`	Use class-agnostic Non-Maximum Suppression (NMS).
`--augment`	`flag`	`False`	Enable augmented inference.
`--update`	`flag`	`False`	Update all models.
`--project`	`str`	`runs/detect`	Directory to save results (`project/name`).
`--name`	`str`	`runs/detect/object_tracking`	Name of the results folder inside the project directory.
`--exist-ok`	`flag`	`False`	Allow existing project/name without incrementing.
`--no-trace`	`flag`	`False`	Do not trace the model during export.
`--colored-trk`	`flag`	`False`	Assign a unique color to each track for visualization.
`--save-bbox-dim`	`flag`	`False`	Save bounding box dimensions in `.txt` tracks.
`--save-with-object-id`	`flag`	`False`	Save results with object ID in `.txt` files.