which features are fed into the matching network?

[shaayaansayed]

Hey guys, fantastic work.

I have a question about the paper. You feed the output of ROIAlign into the matching network. I'm having trouble understanding figure 4. How is the input for the matching network of a single image an NxNx256 tensor? N is the number of garment classes, correct? The output of ROIAlign is either 7x7x256 or 14x14x256 (depending on if you take the bbox stream or mask stream). How are you getting NxN?

Thanks!

[geyuying]

N is the size of feature map of a ROI. Given a ROI, a fixed NxNxC feature map is extracted after ROIAlign to represent features of that ROI and is then fed to the matching network.

[xwjabc]

Still curious about the RoI features fed into the matching net.
In the mask head (link), it has the procedure:
backbone -> RoI Pooling -> 4x conv (feature extractor) -> 1x deconv + 1 conv (predictor)
So the RoI features fed into the match net should be the features after RoI Pooling. Am I correct?