code for generating Figure 5.

Question

code for generating Figure 5.

lixiang-ucas opened this issue 5 years ago · comments

Hi,
Can you share the code for calculating the geodesic error maps (Fig. 5)? Thanks.

ThibaultGROUEIX · Answer 1 · Mon Feb 18 2019 21:55:40 GMT+0800 (China Standard Time)

Sure ! It is available here : https://github.com/zorah/KernelMatching/tree/master/tools/geodesic_cpp
Good luck,
Thibault

lixiang · Answer 2 · Tue Feb 19 2019 15:43:38 GMT+0800 (China Standard Time)

Thanks. @ThibaultGROUEIX

lixiang · Answer 3 · Fri Feb 22 2019 23:21:09 GMT+0800 (China Standard Time)

Hi,
I found it somewhat hard to call the geodesic distance function in KernelMatching. Would it be possible for you to share part of your code for generating the geodesic distance given a source point, a target point, and the reference mesh? Thank you so much for your patience.

ThibaultGROUEIX · Answer 4 · Tue Mar 05 2019 21:32:46 GMT+0800 (China Standard Time)

Hi @lixiang-ucas, it is actually pretty messy, i would need to clean it up and i won't have time to do so until the ICCV deadline... Let me know if you urgently need it.
Best regards,
Thibault

lixiang · Answer 5 · Tue Mar 05 2019 21:40:41 GMT+0800 (China Standard Time)

Thank you for your attention. I've figured it out.

lixiang · Answer 6 · Thu Mar 07 2019 16:29:26 GMT+0800 (China Standard Time)

Hi Thibault, I followed the instructions in Kernel Matching repo and successfully generate the desired geodesic distance for FAUST and TOSCA dataset, but failed with SCAPE dataset. The meshes in SCAPE dataset contains holes, which caused errors when calculating the geodesic distances. Can you give me some suggestions for how to figure it out? Thanks.

ThibaultGROUEIX · Answer 7 · Thu Mar 07 2019 16:54:39 GMT+0800 (China Standard Time)

hi @lixiang-ucas
why not use nearest neighbors between the scans with holes and the registered scans ? you should be able to compute geodesic distance on the registered scans i think.