Instance segmentation

[WaterKnight (David Lacalle Castillo)]

For getting Torchvision Mask-RCNN work with learner class I think that the proper way is subclassing Learner.

I have added a topic where I am explaining my concerns and problems that I am struggling with.

Error in a Learner Subclass

We are trying to make instance segmentation in this other post However, I am facing the issue that Torchvision Mask-RCNN forward method need image and target in training and just image in validation. In addition, it outputs losses in training and detection in inference. During training, the model expects both the input tensors, as well as a targets (list of dictionary), containing: boxes ( FloatTensor[N, 4] ): the ground-truth boxes in [x1, y1, x2, y2] format, with values of x betwee…