[Bug] Faided to run demo.py using pretrained models.

[fediel]

Prerequisite

• I have searched Issues and Discussions but cannot get the expected help.
• I have read the FAQ documentation but cannot get the expected help.
• The bug has not been fixed in the latest version (dev-1.x) or latest version (dev-1.0).

Task

I'm using the official example scripts/configs for the officially supported tasks/models/datasets.

Branch

main branch https://github.com/open-mmlab/mmdetection3d

Environment

pip install torch==1.10.1+cu111 torchvision==0.11.2+cu111 torchaudio==0.10.1 -f https://download.pytorch.org/whl/cu111/torch_stable.html

Reproduces the problem - code sample

with torch.no_grad():
results = model.test_step(collate_data)

is_occupancy = ('pred_occupancy' in results[0])
if is_occupancy:
classes = ['empty'] + classes # 0 = empty for occupancy

filtered_results = []
if not is_occupancy:
for i in range(len(results)):
boxes, labels = nms_filter(results[i].pred_instances_3d)
filtered_results.append((boxes, labels))

Reproduces the problem - command or script

python3 demo.py ../configs/detection/cont-det3d_8xb1_embodiedscan-3d-284class-9dof.py ../models/cont-3ddet.pth --root-dir openscan --scene office --device cuda:0

Reproduces the problem - error message

Traceback (most recent call last):
File "demo.py", line 288, in
main(args)
File "demo.py", line 219, in main
boxes, labels = nms_filter(results[i].pred_instances_3d)
File "demo.py", line 105, in nms_filter
iou = boxes.overlaps(boxes, boxes, eps=1e-5)
File "/home/iro/EmbodiedScan/embodiedscan/structures/bbox_3d/euler_box3d.py", line 134, in overlaps
_, iou3d = box3d_overlap(corners1, corners2, eps=eps)
File "/home/iro/.conda/envs/embodiedscan/lib/python3.8/site-packages/pytorch3d/ops/iou_box3d.py", line 161, in box3d_overlap
_check_nonzero(boxes1, eps)
File "/home/iro/.conda/envs/embodiedscan/lib/python3.8/site-packages/pytorch3d/ops/iou_box3d.py", line 87, in _check_nonzero
raise ValueError(msg)
ValueError: Planes have zero areas

Additional information

mv-det3d is available, but the results from different scenarios are integrated into one image.

[mxh1999]

Please see #40 for help.

[xjj1999]

请问为什么mv-det3d输出results的长度是1?我也碰到了这个问题