Case Study: ModiFace delivers browser-based AR makeup try-on with TensorFlow.js

How Modiface utilized TensorFlow.js in production for AR makeup try on in the browser

ModiFace, a developer of AI for the beauty industry acquired by L'Oreal, faced the challenge of creating an accessible and accurate augmented reality makeup try-on experience for web browsers. Their previous solution was too large, leading to slow load times and reduced accuracy on various face shapes, which limited inclusivity. They needed a lightweight, high-performance framework to deploy their new convolutional neural network model directly in the browser to overcome network latency and privacy concerns.

The solution was implemented using TensorFlow.js, which allowed ModiFace to run its state-of-the-art face tracking model client-side. By leveraging TensorFlow.js's WebGL acceleration, they created a highly accurate application under 3 MB that loads quickly, even on mobile networks. This resulted in a greatly improved user experience, increased consumer engagement, and higher conversion rates for L'Oreal's brands, all while ensuring user privacy as no images ever leave the device.

ModiFace

Jeff Houghton

Chief Operating Officer