search

SxJyJay / MSMDFusion

[CVPR 2023] MSMDFusion: Fusing LiDAR and Camera at Multiple Scales with Multi-Depth Seeds for 3D Object Detection
Apache License 2.0
167 stars 10 forks source link

MSMDFusion-base val performance。 #8

Closed AndyYuan96 closed 1 year ago

AndyYuan96 commented 1 year ago

Hi，I just wonder the MSMDFusion-base‘s val performance， as I don’t see the result in lastest version paper，and does you keep using same input image resolution as transfusion？ thanks。

SxJyJay commented 1 year ago

The MSMDFusion-base achieves mAP and NDS (on nuScenes val) in the range of 69.3-69.5 and 71.9-72.1, respectively. The precise performance results on the validation set will be reported after retraining the refactored code. We don't enlarge image resolution in MSMDFusion-base due to memory constraints.

AndyYuan96 commented 1 year ago

The MSMDFusion-base achieves mAP and NDS (on nuScenes val) in the range of 69.3-69.5 and 71.9-72.1, respectively. We don't enlarge image resolution in MSMDFusion-base due to memory constraints.

get，for test submission，I just wonder if you use same training config with val，only change to use train+val dataset to train。

SxJyJay commented 1 year ago

Yes.