About the abnormal mASE and mASE

[zhanghan0112]

Hi, thank you for your excellent work! I try to reproduce the results of "resnet-101 with fcos3d initialized backbone" in the config. The data is created by the "create_data.py" in this repo, the produced "train.pkl" and "val.pkl" are used for training. The mAP and mAAE, mAVE, mATE are similar to the results of the paper, but why mAOE and mASE are 0.7, 1.6, they are twice difference with the paper produced? Looking forward to your reply, thank you!

[yihongXU]

Hi,

Maybe it is related this issue #31 and please check if your mmdet3.0 version is consistent with the version of the generated pkl files.

Good luck!