Why the name_to_bottom_map in ASDNLossLayer is different from fast_rcnn_adv/train.prototxt?

[huyangc]

In the roi_data_layer.layer.ASDNLossLayer, the bottom[0] is 'mask_pred', but in the train.prototxt, which is 'conv10_mask'. It is the slip of ASDNLossLayer or train.prototxt? Or it is right and I am confused

[huyangc]

I figure out why you write it like that after reading the code and prototxt of ASDNPretrainLossLayer. I think you need to change it to avoid misunderstanding. Thanks for the thoughts and codes!