pvtokmakov

Hi, nope, sorry, we used Docker. The corresponding Docker file is released in this repo.

Hi, sorry for the late reply :) We haven't done a thorough investigation of the best approach for supervising the visibility head. Heatmap just seemed like the most natural thing...

Hi, thanks for you interest in our work! 1. We avoid false positives by additionally training a box visibility classifier, and only outputting visible detections at test time. 2. CrowdHuman...

Hi, can you please be more specific?

Hi, thanks for your interest in our work! Passing --debug 4 flag to the evaluation command should save the videos with visualized predictions into the exp/tracking/experiment_name/debug directory. Hope this helps.

Hi, MOT is annotated with amodal bounding boxes, that is, if a part of a person falls outside of the frame the model still needs to predict a box for...

Hi, thanks for your interest in our work! Unfortunately, we did not save the code for that experiment, but we basically just used the original CenterTrack codebase and added our...

Hi, thanks for your interest in our work! Training on PD takes 4-6 days depending on your GPU model and PyTorch version.

Hi, I'm sorry, but we don't provide a demo script at the moment. I'll try to add it in the future, but not in the next few weeks.

Hi Brian, sorry for the delay! I'll try to find time to write a generic demo script in before the end of the month. Have been quite busy with other...