facebookresearch / DepthContrast

DepthContrast self-supervised learning for 3D
Other
267 stars 35 forks source link

Per point feature vector? #26

Closed katadam closed 3 years ago

katadam commented 3 years ago

Is there a way to extract per point (vertex) features performing a forward pass as a descriptor for each point, given that this is a U-net architecture?

zaiweizhang commented 3 years ago

Of course. The original backbone usually gives per point feature. We just did max-pooling to extract the global feature vector.