Model Zoo

This repo is a collection of several deep learning models implemented using Pytorch according to corresponding papers.

Each folder contains the model, train and test scripts.

A list of all the models:

(under construction)

U-net implementation
U-net implementation (Non-cropped version)

Some issues need to emphasize:

Choice of Upsampling or Transposed convolution

Upsampling can be seen as expanding the size by repeating the values to fill the extra space
transposed convolution (de-convolution) is acutally a convolution and have parameters to learn
In all the implementations, I used transposed convolution unless the paper has clearly pointed out the way they used to increase the dimensions.