Tai Wang

It seems some boxes outside of the image are not filtered and the projection is from boxes behind the image plane. Do you locate the bug successfully? If not, we...

Thanks for your report. Actually, we find another tricky bug that may cause some boxes not to be correctly filtered [here](https://github.com/open-mmlab/mmdetection3d/blob/master/tools/data_converter/kitti_converter.py#L532). We should input a custom argument for image size...

> > > > > It seems some boxes outside of the image are not filtered and the projection is from boxes behind the image plane. Do you locate the...

@ZCMax Please have a look at this issue. We may make a plan to adjust related codes and make a regression for validation.

Sorry for the little late reply. Awesome attempt! It's really interesting. As you mentioned, we have made a refactoring in terms of the hack of yaw angle and dimensions for...

If you have involved all the other refactoring in that PR, the overall implementation of mmdet3d should be correct. Then how about your mAOE in the evaluation? Does it also...

I mean the mAOE instead of mAP because the mAP of nuScenes does not consider the orientation of boxes (only consider whether the distance of centers is under certain thresholds).

Yes, your evaluation results show that your predicted orientations have something wrong. The mAP is defined based on the distance thresholds instead of 3D-IoU on nuScenes, so it can not...

You can also see in the paper's statistics that we take the 288 categories with the most objects (>10 instances per category) for the detection benchmark. The remaining categories have...

Hi Luke, glad to see you coming to try our dataset and repo! Thanks for your suggestion, and we will consider releasing the full annotation with long-tailed categories in the...