OpenDriveLab / ViDAR

[CVPR 2024 Highlight] Visual Point Cloud Forecasting
https://arxiv.org/abs/2312.17655
Apache License 2.0
278 stars 18 forks source link

About the mini train.pkl and val.pkl #20

Closed jiangxb98 closed 7 months ago

jiangxb98 commented 7 months ago

Could you please provide the mini divided val and train pkl files in the log provided by OpenScene? I found that the metric (CD) were much worse than the baseline provided by the official pth when I trained the baseline. What I understand is that part of the data in the official training set is divided into our validation set, resulting in good metric (CD) using the official pth file.

tomztyang commented 7 months ago

Attached. mini_train.txt mini_val.txt