The goal of project is to implement the main ideas of research papers into pytorch code. The code is thus not meant to review all results given in the paper, but rather to get a feeling for the underlying algorithms.
I code every project in one comprehensive Jupyter notebook, mainly because it lets you run them in Google Colab with GPU support for free but also I feel that notebook are great ways to share knowledge.
Some paper implementations are still in progress.
Here is the current list of papers:
- DCGAN -> Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks
- pix2pix -> Image-to-Image Translation with Conditional Adversarial Networks
- cycleGAN -> Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks
- deepfillv2 -> Free-Form Image Inpainting with Gated Convolution
- SC-FEGAN -> SC-FEGAN: Face Editing Generative Adversarial Network with User's Sketch and Color
- fast_style -> Perceptual Losses for Real-Time Style Transfer and Super-Resolution
- neural_style_transfer (Combination of the following papers):
All models that have been training with the published code are available on my google drive.