error for segmentation models in training and test

[heli223]

when i run the commands for segmentation with any model (camera-only, lidar-only, bevfusion), it will occur errors. 1. For training, its error is about File "/home/bevfusion-main/mmdet3d/datasets/pipelines/transforms_3d.py", line 218, in call data["radar"].rotate(-theta) AttributeError: 'dict' object has no attribute 'rotate'

2. For test, its error is about File "/home/bevfusion-main/mmdet3d/datasets/pipelines/loading.py", line 294, in call location = data["location"] KeyError: 'location'

can you help me with this? thanks!

[Yunge6666]

I have the same 'location' error. Have you solved this problem?

[heli223]

I have the same 'location' error. Have you solved this problem? I used an environment build by my friend, and it is okay now. Maybe I think it is because the versions of different packages. I don't know which packages are the reasons since the most important packages are the same.

[971022jing]

when i run the commands for segmentation with any model (camera-only, lidar-only, bevfusion), it will occur errors. 1. For training, its error is about File "/home/bevfusion-main/mmdet3d/datasets/pipelines/transforms_3d.py", line 218, in call data["radar"].rotate(-theta) AttributeError: 'dict' object has no attribute 'rotate'

2. For test, its error is about File "/home/bevfusion-main/mmdet3d/datasets/pipelines/loading.py", line 294, in call location = data["location"] KeyError: 'location'

can you help me with this? thanks!

hello，have you solved the problems?

[heli223]

when i run the commands for segmentation with any model (camera-only, lidar-only, bevfusion), it will occur errors. 1. For training, its error is about File "/home/bevfusion-main/mmdet3d/datasets/pipelines/transforms_3d.py", line 218, in call data["radar"].rotate(-theta) AttributeError: 'dict' object has no attribute 'rotate' 2. For test, its error is about File "/home/bevfusion-main/mmdet3d/datasets/pipelines/loading.py", line 294, in call location = data["location"] KeyError: 'location' can you help me with this? thanks!

hello，have you solved the problems?

No, I did not solve it. I used an old version of the code at https://gitee.com/linClubs/bevfusion. Maybe you can have a look.

[EpicGilgamesh]

@heli223 you had just downloaded this older version and everything is working fine? I'm fighting rn for over a month with BEVFusion, because I need it for my engineer thesis and after fixing one thing there is always another... I see that it has newer commits that the one here, you downloaded whole new repo or just multiple files? Also if you could list used packages pls

[heli223]

you had just downloaded this older version and everything is working fine? I'm fighting rn for over a month with BEVFusion, because I need it for my engineer thesis and after fixing one thing there is always another... I see that it has newer commits that the one here, you downloaded whole new repo or just multiple files? Also if you could list used packages pls

@EpicGilgamesh I download the whole repo at https://gitee.com/linClubs/bevfusion and use it according to the following link: https://zhuanlan.zhihu.com/p/649200166

[CaveSpiderLZJ]

To solve the first error, you can just comment these lines in mmdet3d -> datasets -> pipelines -> transforms_3d.py -> GlobalRotScaleTrans -> __call__ like this:

if "points" in data:
    data["points"].rotate(-theta)
    data["points"].translate(translation)
    data["points"].scale(scale)

# if "radar" in data:
#     data["radar"].rotate(-theta)
#     data["radar"].translate(translation)
#     data["radar"].scale(scale)

gt_boxes = data["gt_bboxes_3d"]
rotation = rotation @ gt_boxes.rotate(theta).numpy()
gt_boxes.translate(translation)
gt_boxes.scale(scale)
data["gt_bboxes_3d"] = gt_boxes