Here is how the repo is organized per folder:
detection: Contains the models as well as the provided starter code. The trained models are saved here, as well as the inference results for test and validation sets.
evaluation: Contains the script used to evaluate the accuracy. We used camera 5 for evaluation. Here, "camera_0005_gt.txt" is the ground truth, and "camera_0005.txt" is the tracking result.
reid: Includes the feature extraction code provided by the TA as well as the model weights.
tracking: Contains the tracking code provided by the TA.
Within each file, each one has a Readme.txt file indicating what was changed.