search

yinjunbo / SSDA3D

Thie repo provides the official implementation of our AAAI-2023 paper “SSDA3D: Semi-supervised Domain Adaptation for 3D Object Detection from Point Cloud”.
Apache License 2.0
45 stars 2 forks source link

About the intra_domain_point_mixup #3

Closed AGI-player closed 1 year ago

AGI-player commented 1 year ago

Hello, thanks for your great work. I found some differences between the code and paper. According to the paper, "before merging the 3D boxes, collision detection will be performed". But in the intra_domain_point_mixup function, the "gt_boxes" were merged directly. This may cause some problems.

YannnWanggg commented 1 year ago

Hello, thanks for your great work. I found some differences between the code and paper. According to the paper, "before merging the 3D boxes, collision detection will be performed". But in the intra_domain_point_mixup function, the "gt_boxes" were merged directly. This may cause some problems.

Thank you for your attention to the paper detail. I have updated the codes. Now you can specify the cfgs of stage 2 to decide whether to perform collision detection. Actually, we conduct both experiments, and the difference is not very obvious because collisions do not happen very often.

AGI-player commented 1 year ago

Hello, thanks for your great work. I found some differences between the code and paper. According to the paper, "before merging the 3D boxes, collision detection will be performed". But in the intra_domain_point_mixup function, the "gt_boxes" were merged directly. This may cause some problems.

Thank you for your attention to the paper detail. I have updated the codes. Now you can specify the cfgs of stage 2 to decide whether to perform collision detection. Actually, we conduct both experiments, and the difference is not very obvious because collisions do not happen very often.

ok, thanks.