search

prstrive / UniMVSNet

[CVPR 2022] Rethinking Depth Estimation for Multi-View Stereo: A Unified Representation
MIT License
231 stars 12 forks source link

About the setting of the 3-step conf values in a new dataset #30

Open dontknowwhatstoname opened 9 months ago

dontknowwhatstoname commented 9 months ago

Hello. Could you please tell me how you confirm the 3-step conf values of different classifications for the 'Tanks and Temples' dataset? Do I need use another neural network to learn it? Or just need get the conf distribution and take some experiments?

prstrive commented 5 months ago

We set the conf values according to our experiments on our baseline model. And it should help if you have a robust network or design a more adaptive strategy like D2HC-RMVSNet on the Tank dataset.