This repository is dedicated to the implementation of the Pix2Pix Conditional Generative Adversarial Network (GAN) for image-to-image translation tasks. The generator and discriminator were trained from scratch for two datasets: maps to satellite views (and vice versa) and faces to comics translation.
- Maps Dataset: Available on Kaggle, this dataset allows the model to translate between maps and satellite views.
- Face to Comics Dataset: Also available on Kaggle, this dataset is used for translating human faces to comic styles.
conf.py: Configuration parameters and settings.datasets.py: Dataset loading and preprocessing script.discriminator.py: Discriminator network architecture.export tensorboard results.ipynb: Jupyter notebook for exporting TensorBoard results.generator.py: Generator network architecture.paper.pdf: Original Pix2Pix paper.train.py: Training script for the Pix2Pix model.utils.py: Utility functions.- A report written for a Master's degree course will also be added to this repository.
The Pix2Pix model is based on the paper titled "Image-to-Image Translation with Conditional Adversarial Networks" by Isola et al., 2017, which is included in this repository.
Please cite the original publishers of the datasets and the Pix2Pix paper as follows:
@inproceedings{isola2017image,
title={Image-to-Image Translation with Conditional Adversarial Networks},
author={Isola, Phillip and Zhu, Jun-Yan and Zhou, Tinghui and Efros, Alexei A},
booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
pages={1125--1134},
year={2017}
}- Maps Dataset citation: Alin Cijov, "Pix2Pix Maps Dataset", Kaggle, https://www.kaggle.com/datasets/alincijov/pix2pix-maps.
- Face to Comics Dataset citation: Defileroff, "Comic Faces Paired Synthetic Dataset", Kaggle, https://www.kaggle.com/datasets/defileroff/comic-faces-paired-synthetic.
To train the Pix2Pix model, execute the train.py script with the appropriate dataset and configuration settings in conf.py. Ensure the datasets are downloaded and structured as expected by datasets.py.
This project has been developed for educational purposes within a Master's degree program. Contributions to enhance the model and its applications are welcome.
This code is distributed under the same terms as the datasets and the Pix2Pix paper. Adherence to the respective licenses is required when using and distributing this code.