XuyangBai

Hello, I am trying to train your model on CIFAR10 by loading cifar picture and modify some hyperparameter in `trainDataSpaceNetwork` and `generateFigure` ```python def loadCIFAR(): with open("cifar-10-batches-py/data_batch_1", 'rb') as f:...

Hi, Thanks for your sharing. In your project website you mentioned that the ply file for geometric registration fragments `Each fragment is a 3D point cloud of a surface, integrated...

Objects formed by reflection effect.

2

Hi, Thanks for sharing this great dataset. For scenes with large glass, there usually exist some LiDAR points behind the glass maybe due to the reflection, and sometimes the points...

Program stop at some iteration

6

Hi @tmbdev Thanks for sharing the excellent libaray. When I use webdataset to build my dataset, my training program stops at some random iterations. The GPU utility is still high...

Varifocal loss is increasing.

2

Hi, thanks for sharing this excellent work. When I add an IoU prediction branch with varifocal loss to my anchor-free detection model, the mAP does increase, but I find the...

Why performs so well on KITTI ?

1

Hi @lijx10 Could you please provide some insight on why USIP performs so well on KITTI? Even only 4 keypoints are detected, USIP achieves 34% relative repeatability, which is really...

Code Release.

15

Hi, I really appreciate your work on point cloud keypoint detection. Is there any estimated date for the release? Thank you very much.

Thanks a lot for your great work and code, do you have a plan to add support for rigid transformation estimation? Best, Xuyang

How the feature re-use is performed

1

Hi thanks for sharing the excellent work. I wonder how the feature re-use is performed. As mentioned in your paper, feature reuse brings a huge gain, but I can not...

Hi, thanks for your sharing. I find for each decoder layer, you use the `cluster_xyz` as the initial location instead of the updated `base_xyz` https://github.com/zeliu98/Group-Free-3D/blob/ef8b7bb5c3bf5b49b957624595dc6a642b6d0036/models/detector.py#L221-L227 My question is since each...