search

Owen-Liuyuxuan / visualDet3D

Official Repo for Ground-aware Monocular 3D Object Detection for Autonomous Driving / YOLOStereo3D: A Step Back to 2D for Efficient Stereo 3D Detection
https://owen-liuyuxuan.github.io/papers_reading_sharing.github.io/3dDetection/GroundAwareConvultion/
Apache License 2.0
365 stars 77 forks source link

关于截断目标的问题 #58

Closed wzan0001 closed 2 years ago

wzan0001 commented 2 years ago

网络无法检测出被截断的目标3D框，那么能否检测出2D框，对截断目标的2D框检测是否有影响呢？

Owen-Liuyuxuan commented 2 years ago

目前的代码会为了提高3D检测的点数，把截断目标剔除在训练数据中,因而网络无法检测出被截断的目标. 取消这一步就可以正常训练截断目标,这个可以设置这个超参,在config中修改,重跑imdb_precompute就好了 https://github.com/Owen-Liuyuxuan/visualDet3D/blob/77a3925c401cf8f4705d0b3079c79c78bb66dab2/scripts/imdb_precompute_3d.py#L77

wzan0001 commented 2 years ago

Thanks！！