About Correlation layer and can you share me the code without deformable conv
yanzhou-li opened this issue · 3 comments
yanzhou-li commented
i come back again. Drive into the theme,I want to know if I use the concat
and convolution
instead of correlation
,the net can work? if it works, how can I do that,thanks!
@haofeixu @Ilya-Muromets
haofeixu commented
@yanzhou-li , yes, our implementation supports both cost volume construction methods, you can check the --feature_similarity
argument in train.py
.
haofeixu commented
BTW, if you don't want to use the deformable conv, you can set the --num_deform_blocks
to 0 in train.py
.
yanzhou-li commented
@haofeixu i got it. thank you