search

Rorisis / Co-Occ

[IEEE RA-L] Co-Occ: Coupling Explicit Feature Fusion with Volume Rendering Regularization for Multi-Modal 3D Semantic Occupancy Prediction
https://ieeexplore.ieee.org/document/10517470
Apache License 2.0
44 stars 5 forks source link

雷达分支需要图像文件吗 #2

Closed vanelyang closed 6 months ago

vanelyang commented 6 months ago

使用 nusc_lidar.pth 预训练模型做评估，配置文件使用 coocc_lidar.py，报错信息提示缺少图片文件，请问使用雷达分支也需要图片吗？

Rorisis commented 6 months ago

Thanks for your questions. We apologize for uploading the incorrect version. We have made modifications to the codes in the LiDAR branch for your reference.

vanelyang commented 6 months ago

Thanks for your questions. We apologize for uploading the incorrect version. We have made modifications to the codes in the LiDAR branch for your reference.

Thanks very much.

algo-scope commented 3 months ago

Hello, I use bash tools/dist_test.sh projects/configs/coocc_nusc/coocc_lidar.py nusc_lidar.pth 1 --pred-save pred_result_lidar to evaluate lidar version, but I got this error:

  File "Co-Occ-main/projects/mmdet3d_plugin/coocc/apis/test.py", line 217, in custom_multi_gpu_test
    save_output_nuscenes(data['gt_depths'],
KeyError: 'gt_depths'

Do you have any idea?

============================== update: I can run by python tools/test.py projects/configs/coocc_nusc/coocc_lidar.py nusc_lidar.pth --eval=bbox

algo-scope commented 3 months ago

Thanks for your questions. We apologize for uploading the incorrect version. We have made modifications to the codes in the LiDAR branch for your reference.

Sorry but I cannot find the LIDAR branch