simulated-unsupervised-tensorflow and SimGAN
These are competing implementations of the same paper (SimGAN), both aiming to reproduce Apple's approach for bridging the domain gap between synthetic and real images through adversarial training, with the carpedm20 version using TensorFlow while the mjdietzx version's framework choice differs.
About simulated-unsupervised-tensorflow
carpedm20/simulated-unsupervised-tensorflow
TensorFlow implementation of "Learning from Simulated and Unsupervised Images through Adversarial Training"
This project helps machine learning engineers improve the realism of synthetic images for computer vision tasks, particularly eye tracking. You provide simulated images and real-world images, and it outputs refined synthetic images that look more like real ones. This is useful for researchers and developers building systems that rely on large datasets of realistic images, especially when real data is scarce or expensive to collect.
About SimGAN
mjdietzx/SimGAN
Implementation of Apple's Learning from Simulated and Unsupervised Images through Adversarial Training
This project helps computer vision researchers adapt models trained on synthetic eye images to perform well on real-world eye images. It takes a dataset of computer-generated eye images and a dataset of real human eye images as input. The output is an improved model for tasks like gaze estimation, making it more robust for use with actual human data. This tool is for researchers and engineers working on gaze tracking or similar applications who need to bridge the gap between simulated and real image data.
Scores updated daily from GitHub, PyPI, and npm data. How scores work