dvlab-research / UVTR

Unifying Voxel-based Representation with Transformer for 3D Object Detection (NeurIPS 2022)
227 stars 17 forks source link

excuese me,create data error #15

Closed eezhang123 closed 1 year ago

eezhang123 commented 1 year ago

environment: mmcv-full 1.3.8 mmdet 2.14.0 mmdet3d 0.17.3
mmsegmentation 0.14.1 torch1.7.1 cuda 11.0

UVTR/mmdetection3d/mmdet3d/core/bbox/structures/utils.py", line 139, in points_cam2img 4, device=proj_mat.device, dtype=proj_mat.dtype) AttributeError: 'numpy.ndarray' object has no attribute 'device'

eezhang123 commented 1 year ago

train scene: 700, val scene: 150 [>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>] 34149/34149, 1.2 task/s, elapsed: 28247s, ETA: 0s train sample: 28130, val sample: 6019

====== [ ] 0/28130, elapsed: 0s, ETA:Traceback (most recent call last): File "extra_tools/create_data.py", line 86, in max_sweeps=args.max_sweeps) File "extra_tools/create_data.py", line 40, in nuscenes_data_prep root_path, info_train_path, version=version)

yanwei-li commented 1 year ago

Thanks for this report! I think a naive solution is directly commenting these lines. Because the 2D annotation conversion is actually not used in our experiments. I update the code. Please try it again.