Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

TFRecords preprocessing #3

Closed
bluestyle97 opened this issue Sep 2, 2021 · 2 comments
Closed

TFRecords preprocessing #3

bluestyle97 opened this issue Sep 2, 2021 · 2 comments

Comments

@bluestyle97
Copy link

Hi, I want to generate tfrecords training data from another mesh dataset (which has already been watertight), but there seems to be no instruction on this step. I have read this istruction, but the code seems to compute sdf values only and have no help with my question. I found there are fields in the tfrecords data that I don't know what they means and where they come from, such as "sdf_samples_depth6", "sdf_samples_depth7", "depth6_coord" and "depth7_coord". Can you give a more detailed instruction on how to generate the right-format tfrecords data? Thanks a lot!

@Andy97
Copy link
Owner

Andy97 commented Sep 6, 2021

Hi, I want to generate tfrecords training data from another mesh dataset (which has already been watertight), but there seems to be no instruction on this step. I have read this istruction, but the code seems to compute sdf values only and have no help with my question. I found there are fields in the tfrecords data that I don't know what they means and where they come from, such as "sdf_samples_depth6", "sdf_samples_depth7", "depth6_coord" and "depth7_coord". Can you give a more detailed instruction on how to generate the right-format tfrecords data? Thanks a lot!

Hi, as mentioned in the paper, we have 2 set of sdf samples(sampled from 64x64x64 grid points and 128x128x128 grid points, please refer to Appendix A of paper), which were used for sdf training in a coarse-to-fine manner progressively(e.g. using 64^3 samples for first 30 epochs and 128^3 samples for remaining epochs). The "depth6_coord" and "depth7_coord" is the 3d coordinates of these sample points, while "sdf_samples_depth6", "sdf_samples_depth7" contain the corresponding ground truth sdf values and gradients. For details, please refer to code for data loading.

For memory efficiency, each 3-dimensional vector "depth6_coord" and "depth7_coord" is compressed to an int(from 0x0x0 to 257x257x257). (Please refer to here).

The field "octree_points_idx" contains the ground truth octree shape, which has also been compressed to an integer.

@Andy97 Andy97 closed this as completed Sep 23, 2021
@Andy97
Copy link
Owner

Andy97 commented Sep 30, 2021

Script for generating tfrecords is released, please see here.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants