AR and deep neural networks collide to provide ModiFace

ModiFace has announced a new live video based hair tracking and hair color simulation technology utilizing a deep neural network architecture. The new deep learning architecture utilizes a set of collaborative neural networks to detect hair in each video frame and to adjust the coloration of hair in a photo-realistic way. The implementation also utilizes the latest machine learning advances within iOS 11, including CoreML, to provide the smoothest video transformation experience.

“We have been working on deep learning architectures for a long time now, and recent advances in both the neural network architectures, basic hardware level optimizations, as well as the availability of significant training data, have made photo-realistic video hair tracking and coloration possible,” said Parham Aarabi, CEO of ModiFace and Professor at the University of Toronto.

The new patented neural network architecture was trained on 220,000 carefully annotated hair images - the largest such database in the world. The basic inspiration behind the collaborative neural network architecture was recently published in the IEEE Transactions on Neural Networks and Learning Systems. The new technology is available within ModiFace’s popular “Hair Color” app, available on the App Store for iPhone and iPad.

“We are extremely excited to offer live hair coloration and style alteration to our brand partners. The accuracy and realism of this new technology will forever change how hair color is tried on, explored, and purchased by consumers. We expect the first brands to leverage this technology to launch within the next few months,” said Parham Aarabi.

The latest deep learning advances by ModiFace are a direct result of a recent $4M investment by the company in AR and AI technologies at the University of Toronto, with various deep learning University of Toronto researchers joining ModiFace to pursue the development and advancement of AI technologies related to face-based image processing.