This repo presents you the official code of "VISTA: Boosting 3D Object Detection via Dual Cross-VIew SpaTial Attention"
MIT License
127
stars
13
forks
source link
why I use your pretrained module to test on validation set, it can not reach your result, and I use your config to train module again, it also can not reach your result? #20
use your pretrained model: mAP 0.6097 mATE 0.3028 mASE 0.2483 mAOE 0.2644 mAVE 0.2113 NDS 0.6826
use your config train again mAP 0.6077 mATE 0.2933 mASE 0.2474 mAOE 0.2600 mAVE 1.4374 NDS 0.5753
why the train model mAVE and NDS is too bad
use your pretrained model: mAP 0.6097 mATE 0.3028 mASE 0.2483 mAOE 0.2644 mAVE 0.2113 NDS 0.6826 use your config train again mAP 0.6077 mATE 0.2933 mASE 0.2474 mAOE 0.2600 mAVE 1.4374 NDS 0.5753 why the train model mAVE and NDS is too bad