I am doing an image semantic segmentation task and I want to use pretrained models such as
SegNet. I generally understand the idea of segmentation and its structure but do not know clearly how to implement it in code. I also have some pictures which I should use for training and validation. My question is, what are the steps of this process, from downloading models to training on a particular dataset?
Maybe you know some nice examples which can lead me through all of it.
There are only few Steps,
First Make a Dataloader for semantic segmentation
Second, choose to write your network or perform transfer learning using pretrained network from torchvision models.
Follow any simple github example of training using the same hyperparameters