360 Degree detection issue
Hi,
First of all, thanks for the work, the program works really well for me. I am using the pretrained network on my own LiDAR datasets. I noticed that the point cloud visualization and detection only works for the front half.
I changed the test.py to
| def point_range_filter(pts, point_range=[-69.12, -39.68, -3, 69.12, 39.68, 1]): |
|---|
and now it visualizes the whole point cloud, but the detection still only applies to the front half. Can you help with that?
Current progress:

Thanks in advance!
Hi @CrazyMage24,
Try to update the point_cloud_range parameter in the following 3 code blocks,
- https://github.com/zhulf0804/PointPillars/blob/9c98d51742ff21209e1a587cf6c6f680f1cc5224/model/pointpillars.py#L220-L226
- https://github.com/zhulf0804/PointPillars/blob/9c98d51742ff21209e1a587cf6c6f680f1cc5224/test.py#L15
- https://github.com/zhulf0804/PointPillars/blob/9c98d51742ff21209e1a587cf6c6f680f1cc5224/test.py#L38
Please let me know if there are any problems, thanks.
Hi again @zhulf0804,
When I update these 3 parameters to [-69.12, -39.68, -3, 69.12, 39.68, 1]
this happens

But if I update the anchor parameters as well in the pointpillars.py line 246-248, then it goes back to the previous state. Thanks again!
Hello @CrazyMage24,
Could you please send the tested Lidar data to my mailbox ([email protected]) if possible?
I'll debug it in the next few days.
Thanks.
Hi,
I have a Drive folder with the bin files https://drive.google.com/drive/folders/1HAAgm9dSs5PbshG-mb4CV-bG7DgoOwyK?usp=sharing
Thank you!
Hi @CrazyMage24,
Maybe there are 4 code blocks to be updated for point cloud range.
In addition to the above 3 code blocks, the 4th code block needed to be updated is https://github.com/zhulf0804/PointPillars/blob/9c98d51742ff21209e1a587cf6c6f680f1cc5224/model/pointpillars.py#L245-L248
Please let me know if there are any problems, thank you.
Update
Prediction on 000022.bin you provided in the Drive folder.
