Question about the pmf estimating module
Closed this issue · 4 comments
In the article, it is shown that a learnable context model is adopted in your method. However, I only find a rate loss for the learnable context model in your method to upate the parameters. So how to ensure that the learnable context model estimate the pmf correctly? In my opinion, the rate loss only push the entropy of the pmf to be lower but not let the pmf estimation to be accurater. Is there anything wrong?
Thank you for your attention!
We acknowledge that this context model has its shortcomings - the shape of the down-sampled skeleton may be too complex, which makes it difficult to train&fit an accurate pmf for each individual patch.
We strongly suggest you check out our latest work, Pointsoup, which uses a cross-scale context modeling (with the same rate loss) to exploit the skeleton prior, and further provides more accurate and effective PMF estimation for local details (patches).
Thank you for your reply.
Hi, could you please share your test code for tmc13? Thank you very much.
Hi there. Some projects have already organized the test code of TMC13, you can directly refer to:
https://github.com/yydlmzyz/Test-GPCC
https://github.com/3dpcc/GRNet
These projects provide excellent scripts as well as executable files for TMC13. If you need our test code, you can also contact me via email :)