search

deeplearning-wisc / stud

source code for CVPR'22 paper "Unknown-Aware Object Detection: Learning What You Don’t Know from Videos in the Wild"
Apache License 2.0
117 stars 14 forks source link

File path issue #20

Closed lzylzylzy123456 closed 1 year ago

lzylzylzy123456 commented 1 year ago

I tried to run the training process using the bdd100k dataset, but I couldn't get it to work. The code throws an error in builtin.py. Is "/nobackup-slow/dataset/my_xfdu/video/bdd/" a folder you define yourselves? How should I organize the data inside that folder? And how can I obtain bdd100k_mot_domain_splits_val.json and bdd100k_mot_domain_splits_train.json? I noticed many file paths like this in the code, and it seems that they all need to be modified? This doesn't seem to be clearly explained in your readme. How should I proceed?

d12306 commented 1 year ago

Hi, @lzylzylzy123456 , thanks for your interest in the paper.

please replace the /nobackup-slow/dataset/my_xfdu/video/bdd/ with the ./datasets as specified in the readme file

For bdd100k_mot_domain_splits_val.json and bdd100k_mot_domain_splits_train.json, please execute the following in the readme: image