search

qinzheng93 / GraphSCNet

[CVPR2023] Deep Graph-based Spatial Consistency for Robust Non-rigid Point Cloud Registration
MIT License
84 stars 4 forks source link

How to generate correspondences for my own data. #6

Open FDC-WuWeb opened 1 year ago

FDC-WuWeb commented 1 year ago

This is a very impressive job, and thank you for sharing the source code. I am currently using my own data to train the model. Could you please guide me on how to use GeoTransformer to generate correspondences, save them, and use them for training GraphSCNet?

ShuaikangXie commented 1 year ago

Have you solved your problem? Could you guide me to obtain the correspondence of the data？

FDC-WuWeb commented 1 year ago

Have you solved your problem? Could you guide me to obtain the correspondence of the data？

After completing the Geo model testing, I found the npz files in the output/feature directory. I conducted model testing with my own data and am currently attempting to use the generated npz files for training GraphSCNet. I am not certain yet if these npz files can be used.

FDC-WuWeb commented 10 months ago

I found that the npz file generated by the test.py in the Geotransformer's open-source code does not correspond to the correspondence provided in your GraphSCNet project, for example, it lacks the item "s2t_flow", "corr_scene_flows", and "corr_labels". How can I use Geotransformer to generate correspondence for my own data for training GraphSCNet?

FDC-WuWeb commented 10 months ago

Have you solved your problem?

Have you solved your problem? I found that the npz file generated by the test.py in the Geotransformer's open-source code does not correspond to the correspondence provided in your GraphSCNet project, for example, it lacks the item "s2t_flow", "corr_scene_flows", and "corr_labels". How can I use Geotransformer to generate correspondence for my own data for training GraphSCNet?

lemon-essence commented 7 months ago

Have you solved your problem?

Have you solved your problem? I found that the npz file generated by the test.py in the Geotransformer's open-source code does not correspond to the correspondence provided in your GraphSCNet project, for example, it lacks the item "s2t_flow", "corr_scene_flows", and "corr_labels". How can I use Geotransformer to generate correspondence for my own data for training GraphSCNet?

Whether you solved this problem.I also had this problem