udaykusupati / Normal-Assisted-Stereo

[CVPR 2020] Normal Assisted Stereo Depth Estimation
https://udaykusupati.github.io/NAS
MIT License
111 stars 20 forks source link

Details of Table1 in paper #15

Closed KyuminHwang closed 4 years ago

KyuminHwang commented 4 years ago

@udaykusupati

I has questions about Table 1.

MVS    A.Rel  A.diff Sq.Rel   a=1    a=2    a=3
Paper  0.0679 0.1677 0.0555 0.9054 0.9644 0.9879

When testing this quantitative measure, Did you trained MVDNet using DeMoN dataset and Consistency module trained using MVS Datasets only?

Similarly, others dataset also trained MVDNet using DeMoN dataset and Consistency module trained specific datasets only ? Or, trained MVDNet and Consistency module through DeMoN and evaluated each datasets ?

Originally posted by @makeastir in https://github.com/udaykusupati/Normal-Assisted-Stereo/issues/12#issuecomment-704201862

udaykusupati commented 4 years ago

Table 1 is evaluation of MVDNet alone and there is no consistency module used. We train MVDNet jointly on all DeMoN datasets and provide evaluations on each of the individual sub-datasets. For consistency module, we train the entire pipeline using a single dataset (SUN3D or Scannet)