mjdietzx/SimGAN
Keras implementation of Apple's SimGAN for refining synthetic eye-gaze images using adversarial training.
Not currently ranked — collecting fresh signals.
star history
This repository implements Apple’s SimGAN paper, a generative adversarial network approach for simulated unsupervised learning. The model refines synthetic images to better match real-world distributions without paired training data. The implementation uses Keras with TensorFlow backend and focuses on gaze estimation applications using the SynthesEyes and MPIIGaze datasets.
Frequently asked
- What is mjdietzx/SimGAN?
- Keras implementation of Apple's SimGAN for refining synthetic eye-gaze images using adversarial training.
- Is SimGAN open source?
- Yes — mjdietzx/SimGAN is open source, released under the MIT license.
- What language is SimGAN written in?
- mjdietzx/SimGAN is primarily written in Python.
- How popular is SimGAN?
- mjdietzx/SimGAN has 417 stars on GitHub.
- Where can I find SimGAN?
- mjdietzx/SimGAN is on GitHub at https://github.com/mjdietzx/SimGAN.