search

yewzijian / RPMNet

RPM-Net: Robust Point Matching using Learned Features (CVPR2020)
MIT License
344 stars 59 forks source link

how to evaluate other point cloud ? #35

Open Huleyun opened 1 year ago

Huleyun commented 1 year ago

Hi, Thanks for your masterpiece. I am trying to use this network to obtain the rigid transformation parameters between two related point clouds, but I don't know how to realize this thing. Could you please give me some cues?

Any requirements for point clouds? It looks like that the point cloud should be pcl::pointXYZINormal. And they should be same size like (16,717,3)

yewzijian commented 1 year ago

The network already outputs the rigid transformation, e.g. see this line in the inference code.

The provided weights are trained on XYZ+normals so you'll need normals for the point clouds. The network is flexible, so point clouds do not need to have exactly 717 points, nor mini-batched in 16 point clouds at one go. Note however that the ModelNet trained weights may or may not generalize to other point clouds.

Huleyun commented 1 year ago

Oh, thanks for your kindly reply. I will try to generate the normals of point clouds, and then put them into this model. If it is possible, I will retrain this model, but actually our own datasets may be not enough.