Where is curvature-aware point sampling code

[sheshap]

Dear Authors,

Please provide the complete code. Thanks in advance.

[X-Bruce-Y]

I fail to find the "curvature-aware point sampling" function or module too amongst the uploaded scripts.

[alfredtorres]

Sorry for your Question about the Curvature sampling code. For the training efficiency, I intergrate the code in the sampling_gpu.cu https://github.com/alfredtorres/3DFacePointCloudNet/blob/9a857ed9619ddaebe99bfb6bd75462e4f3055e75/pointnet2/_ext-src/src/sampling_gpu.cu#L100 https://github.com/alfredtorres/3DFacePointCloudNet/blob/9a857ed9619ddaebe99bfb6bd75462e4f3055e75/pointnet2/_ext-src/src/sampling_gpu.cu#L106 So, if you want to use the curvature-aware sampling, please use our dataset function refer to GPMM_Normal_Curvature_dataset.py @Seikegn @sheshap

[X-Bruce-Y]

Hi @alfredtorres , thank you for responding. I'm not familiar with CUDA programming, but if my understanding was right, is it that the "Curvature sampling code" part of sampling_gpu.cu works on a dataset already containing curvature info? Then I wonder if there is a recommended way for fast and accurate estimation of curvature that can be incorporated into a python workflow?

I believe you are the first author of "Data-Free Point Cloud Network for 3D Face Recognition" and reside in China. As I'm also working on 3D facial data (for biological applications), it would be great if I could add your Wechat (my ID is Seikegn) for counselling and possible future collaborations.

Thank you!

[alfredtorres]

Hello, @Seikegn. we use PCL's curvature estimation in C++ environment. I have try to re-implement in python but with low speed. So we do not provide the python script. I am glad to have your Wechat. :D

[Jcxloyal]

Hello friends, I use the GenerateTrainData.m file to generate face scans, I got many .bc files, I have no idea how to open or read them, could you please give me some advice on how can I virtualize the face scans.

[aliagha135]

@Jcxloyal did you figure it out