search

yewzijian / 3DFeatNet

3DFeat-Net: Weakly Supervised Local 3D Features for Point Cloud Registration
MIT License
220 stars 46 forks source link

how to use trained model on my data with format .ply or .pcd? #9

Closed ltj95 closed 4 years ago

ltj95 commented 5 years ago

Hello! I want to use cloudcompare to convert my data for the format .bin,but when I run inference.py on my data with format .bin, error occur like

微信图片_20190701104412

How to sovle the problem?

yewzijian commented 5 years ago

Hi, if you get this error it means that the input file is not in the expected format. I'm not sure what cloudcompare saves to, but my code assumes the point cloud is stored as Nx6 float32's, with each row containing x,y,z,nx,ny,nz (the nx, ny, nz are normals which are unused in this code). For other formats, it should be easy to modify the load_point_cloud function to your needs.

ltj95 commented 5 years ago

I convert a pointcloud with Nx6 float32's to .bin file,it's the same error. I also use .txt file ,it has no result.My raw pointcloud data content like this 微信图片_20190701135019

yewzijian commented 5 years ago

Check if the file size of your .bin file is exactly Nx6x4 bytes for a point cloud of N points?