Tai-Wang
RuntimeError: max(): Expected reduction dim to be specified for input.numel() == 0. In tools/test.py
Have you ever tried our pretrained models? Maybe your trained models are not good enough and produce no predictions, which causes the input.numel() == 0.
RuntimeError: max(): Expected reduction dim to be specified for input.numel() == 0. In tools/test.py
You can add a breakpoint in the show function and have a look at why the input.numel() == 0. I guess it might be compatible for no predictions during evaluation...
RuntimeError: max(): Expected reduction dim to be specified for input.numel() == 0. In tools/test.py
> One thing more, I think the pre-trained models must have been trained on spconv1.0. But I have spconv2.0 with my environment is it going to be some mismatch issue...
The unit test check fails. Please fix it accordingly.
> > The unit test check fails. Please fix it accordingly. > > Which one has problems in your environment? I checked carefully agian, but nothing wrong. I work in:...
I find there are many comments you resolved while did not fix actually. Is there any code you have not updated online? If you do not understand a specific comment...
> > I find there are many comments you resolved while did not fix actually. Is there any code you have not updated online? If you do not understand a...
We use the nuScenes LiDAR coordinate system to directly perform LiDAR-based 3D detection on nuScenes. This is indeed a confusing point, and we may consider adjusting it in the future.
Could you please describe your issue more clearly? What do the "don't work" and "don't process with nms" mean?
FCOS3D本身不支持，但应该可以通过微调一些地方也加上相应的模块。 bbox2d的结果应该是没有经过nms后处理的，因为本身主任务还是 3D 检测，加这个 branch 除了 regularize 3D 检测框之外只是为了适配 KITTI 上可以单独输出 2D 框的需求来进行评测，不过这个地方确实还可以优化来提升一下 bbox2d 结果。