evaluation on testing data

[MrSliMs]

Your great work has helped me a lot, but I don’t know how to run your work on the test data. I wish you could write a guide if it’s convenient for you

[hailanyi]

Take the CasTrack as an example:

1. You need create a KITTI account.
2. You need prepare the test set detections of CasA, and organize the files as follows.

   data
   └── casa
      ├── testing
      |      ├──0000
      |      |    ├──000001.txt
      |      |    ├──....txt
      |      |    └──000153.txt
      |      ├──...
      |      └──0020

3. Modify the casa config file to adapt the test set:

dataset_path: "L:/data/kitti/tracking/testing" # training -> testing
detections_path: "data/casa/testing" # training -> testing

tracking_seqs:  [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28] # test tracking seq

5. Modify the config file path in kitti_3DMOT.py line 164, and run the kitti_3DMOT.py:

    parser.add_argument('--cfg_file', type=str, default="config/global/casa_mot.yaml",

6. The tracking results are saved into evaluation/results/sha_key/data/, you need zip all the .txt files into a single .zip file. The file can be submitted to KITTI website based on your KITTI account.