mit-han-lab / bevfusion

[ICRA'23] BEVFusion: Multi-Task Multi-Sensor Fusion with Unified Bird's-Eye View Representation
https://bevfusion.mit.edu
Apache License 2.0
2.33k stars 420 forks source link

Problems about tasks, min_radius, nms_type and nms_scale #631

Closed zhujiagang closed 3 months ago

zhujiagang commented 3 months ago

Thanks for sharing your excellent code! I am not sure the meaning of tasks and other settings here https://github.com/mit-han-lab/bevfusion/blob/4d1ab76686ab53ccf8ead4a1bdf5b6d39c10b5ee/configs/nuscenes/det/centerhead/default.yaml#L30, https://github.com/mit-han-lab/bevfusion/blob/4d1ab76686ab53ccf8ead4a1bdf5b6d39c10b5ee/configs/nuscenes/det/centerhead/default.yaml#L30C1-L36C41 https://github.com/mit-han-lab/bevfusion/blob/4d1ab76686ab53ccf8ead4a1bdf5b6d39c10b5ee/configs/nuscenes/det/centerhead/default.yaml#L22, https://github.com/mit-han-lab/bevfusion/blob/4d1ab76686ab53ccf8ead4a1bdf5b6d39c10b5ee/configs/nuscenes/det/centerhead/lssfpn/camera/default.yaml#L21, https://github.com/mit-han-lab/bevfusion/blob/4d1ab76686ab53ccf8ead4a1bdf5b6d39c10b5ee/configs/nuscenes/det/centerhead/lssfpn/camera/default.yaml#L28. Are these classes arranged according to detection difficulties? Are other nms settings set according to the size of classes?

zhijian-liu commented 3 months ago

Thank you for your interest in our project. This repository is no longer actively maintained, so we will be closing this issue. Please refer to the amazing implementation at MMDetection3D. Thank you again!