search

haofeixu / aanet

[CVPR'20] AANet: Adaptive Aggregation Network for Efficient Stereo Matching
Apache License 2.0
521 stars 100 forks source link

Some questions about inference on PSMnet #49

Closed zcong17huang closed 3 years ago

zcong17huang commented 3 years ago

Because of the version problems, I haven't been able to run your code. I'm trying to read your code to get a better understanding of your work. I'm a little confused when i try to understand the construction of PSMNet-AA. You use the PSMNetFeature to extract feartures to construct cost volumes. Although you make a volume list, it actually has only one cost volume. So when you use AdaptiveAggregation, default num_scales = 3, how could you use CSA In this case? I think there would be an error when running assert len(self.branches) == len(x), because len(self.branches) = num_scales = 3, but len(x) = 1. So where did I get it wrong? Looking forward to your reply and you pointing out my mistake. Extremely grateful！

haofeixu commented 3 years ago

We have constructed an additional feature pyramid for PSMNet feature.