esgomezm
commented
6 months ago Hi @cfusterbarcelo
Thank you for your contribution. Here is a suggestion for the README file based on your text:
Conditional generative model for Image-to-Image Translation
## Description
This model consists on a conditional cycle-Generative Adversarial Network (cycleGAN), popularly known as pix2pix, to transform images from one domain into another (e.g., to predict a DAPI nuclei staining fluorescence channel from a brightfield image of cells). The conditional cycleGAN allows for a supervised training of the network. Thus, this model was trained using a pair set of images. This model was trained using a ZeroCostDL4Mic notebook developed upon the work “Image-to-Image Translation with Conditional Adversarial Networks" by Isola et al. (https://arxiv.org/abs/1611.07004).
## Disclaimer
This model was trained to predict images from a specific image domain and imaged living matter. Further details about how to train a similar model are given in https://github.com/HenriquesLab/DeepLearning_Collab/wiki
## Reference and Citation
If you use this model for your research, please refer to the ZeroCostDL4Mic paper and the original Pix2Pix model paper:
- von Chamier, L., Laine, R.F., Jukkala, J. et al. Democratising deep learning for microscopy with ZeroCostDL4Mic. Nat Commun 12, 2276 (2021). https://doi.org/10.1038/s41467-021-22518-0
- Isola, P., Zhu, J. Y., Zhou, T., & Efros, A. A. Image-to-Image Translation with Conditional Adversarial Networks (2016). arXiv preprint arXiv:1611.07004
Additionally, the pix2pix model has been updated with a few debugs. Please, update your branch with the respective changes and also remove the pycache generated by GH actions so we can merge your changes into the repository.
This PR introduces an enhanced README.md template for models exported using the Pix2Pix notebook. The updated README provides a comprehensive overview, including the model's background, its specific application, and relevant citations. This enhancement aims to standardize and enrich the documentation accompanying models exported to the BioImage Model Zoo format, facilitating better understanding and usage by end-users.