Not yet reported.
git clone https://github.com/ifzhang/ByteTrack.git
cd ByteTrack
pip3 install -r requirements.txt
python3 setup.py develop
2. Install pycocotools.
pip3 install cython; pip3 install 'git+https://github.com/cocodataset/cocoapi.git#subdirectory=PythonAPI'
pip3 install cython_bbox
Download Caltech-UCSD Birds and save it in the datasets
directory. The structure of the folder must look like this:
datasets
|——————CaltechBirds
| └——————bounding_boxes.txt
| └——————classes.txt
| └——————image_class_labels.txt
| └——————images.txt
| └——————train_test_split.txt
| └——————images/
Then, you need to turn the dataset to COCO format:
cd <ByteTrack_HOME>
python tools/convert_birds_to_coco.py