[bug] TypeError: can't convert np.ndarray of type numpy.object_

[HankKung]

Hi I'm encountering the following error when training vip3d with exactly the same scripts and configs:

File "/media/user/data/ViP3D/plugin/vip3d/models/attention_dert3d.py", line 440, in feature_sampling
lidar2img = reference_points.new_tensor(lidar2img) # (B, N, 4, 4)
TypeError: can't convert np.ndarray of type numpy.object_. The only supported types are: float64, float32, float16, complex64, complex128, int64, int32, int16, int8, uint8, and bool.

Does anyone have idea how to fix this bug?

Many thanks!!!

[GentleSmile]

Have you modified the batch_size or something else?

[HankKung]

My bad I modified the fix_feats=True, samples_per_gpu=2 in /vip3d_resnet50_3frame.py
I run it with 4 GPUs and the batch_size remains 8:
bash tools/dist_train.sh plugin/vip3d/configs/vip3d_resnet50_3frame.py 4 --work-dir=vip3d_resnet50_3frame.1

[GentleSmile]

For larger batch_size, you can refer to this issue.