Implementation of "YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors"

This implementation is based on yolov7.

Store your image sequence inside seg/data/images directory, also store the ground truth masks inside seg/data/gt directory. Images and masks are sorted by filenames, so keep the filename numerical (Ideally 5 digit starting with 00000.(jpg|png))

Then go inside seg/ and run the following:

python segment/predict.py 

The output will be stored in runs/ directory.

you can run a docker container with the following command:

docker build -t yolov7 . && \
docker run -v host_images_dir:/app/seg/data/images -v host_ground_truth_dir:/app/seg/data/gt -v host_output_dir:/app/seg/runs  yolov7

For example, this exact command in local machine is run to generate the outputs.

docker build -t yolov7 . && \
docker run -v C:\Users\ge79pih\tmo_data\tmo\tmo_dataset\street:/app/seg/data/images -v C:\Users\ge79pih\tmo_data\tmo\tmo_gt\street:/app/seg/data/gt -v C:\Users\ge79pih\tmo_data\tmo\tmo_output:/app/seg/runs  yolov7