02 ML experiment – Face segmentation
New Machine Learning experiment in nuke, in this time trying to do a face segmentation that allows us to select 20 different IDs in the body. At the moment is work in progress but I hope can release in the future.
Microsoft demonstrate that it is possible to perform face-related computer vision in the wild using synthetic data alone.
The community has long enjoyed the benefits of synthesizing training data with graphics, but the domain gap between real and synthetic data has remained a problem, especially for human faces. Researchers have tried to bridge this gap with data mixing, domain adaptation, and domain-adversarial training, but Microsoft show that it is possible to synthesize data with minimal domain gap, so that models trained on synthetic data generalize to real in-the-wild datasets.
They describe how to combine a procedurally-generated parametric 3D face model with a comprehensive library of hand-crafted assets to render training images with unprecedented realism and diversity. We train machine learning systems for face-related tasks such as landmark localization and face parsing, showing that synthetic data can both match real data in accuracy as well as open up new approaches where manual labelling would be impossible.
Also you can download the original data set in case you want to do your own test or train your own model.
Microsoft dataset of 100,000 synthetic faces with 2D landmark and per-pixel segmentation labels is available for non-commercial research purposes.