search

jasjeetIM / Mask-RCNN

Implementation of Mask-RCNN in Caffe https://arxiv.org/pdf/1703.06870.pdf
232 stars 141 forks source link

meaning of no_loss_mask = -1? #8

Open InnovArul opened 6 years ago

InnovArul commented 6 years ago

Can anyone give a clarification on the intuition of using no_loss_mask = -1? https://github.com/jasjeetIM/Mask-RCNN/blob/master/functions/mask_rcnn/mask_get_minibatch.m#L131

As far as I can tell, cross-entropy loss expects target = [0, 1].

lilydedbb commented 6 years ago

@InnovArul In the prototxt file (https://github.com/jasjeetIM/Mask-RCNN/blob/master/models/mask_rcnn_prototxts/resnet_50layers_conv/train_val.prototxt#L3741), the model ignore the labels which equal -1 when computing the loss.