Pytorch resnet with zero loss

Question

I am training a Pytorch Resnet model for image segmentation. I have two classes and I am training with RGB images and corresponding binary masks. While my accuracy values ar