search

junzhezhang / shape-inversion

[CVPR 2021] Unsupervised 3D Shape Completion through GAN Inversion
MIT License
131 stars 21 forks source link

Inter-Realsense PointCloud #6

Closed liu52020 closed 3 years ago

liu52020 commented 3 years ago

Hello! Can your code complete the stair-type point cloud data obtained by Inter-Realsense?

junzhezhang commented 3 years ago

Hi, thanks for your interest in our work. Our method is invariant to the sensors and point cloud domain, as long as the category of objects has some complete shapes (synthetic and unpaired) to pretrain the GAN. Hope it clarifies.

liu52020 commented 3 years ago

So is it understandable? When I train your code, I select CRN for dataset, chair for class_choice, and my own stair point cloud data (.pcd format) in dataset_path. I wonder if this is ok? Thank you for your answers!

junzhezhang commented 3 years ago

I believe you would need to modify the existing dataset class or write you own ones, as currently my dataset classes handle h5 and ply data only.

liu52020 commented 3 years ago

Thank you for your answers and suggestions, and I wish you success in your work!