mask head usage?

[Scitator]

Hi!

Firstly, thank u a lot. It looks like the best mask-rcnn implementation. (cause it does not use this magic from config import cfg)

Nevertheless, something is a bit strange for me. In the model init you define self.mask_head (cause it needed for segmentation task), but in the forward pass you dont use it at all. How is that?

Thanks!

[ignacio-rocco]

Hi, glad you liked the code :)

Please see the demo notebooks for the usage of the mask head. Essentially there is a results postprocess stage between the forward pass and running the mask head.

Ideally this post-processing stage (which involves NMS and other things) should also be converted into a torch.nn.Module, but this is not the case for now.