New SDK release pushes the boundaries of our Face AR technology, making its features accessible on more platforms. We’ve also improved the face detection technology implementing the neural-based approach and added the support of Windows and Web for a range of neural networks.
Neural network-based face detection
The poorly optimized face detection deteriorates the end-user experience, which results in face filter lagging and frequent losses. With fast and stable face detection optimized for different conditions, e.g. low lighting, head angles or face occlusion, the end-user experience runs smoothly. The application saves the computational resources and deliveries better speed and performance.
By implementing the neural network-based face detection approach we improved the speed of face detection and achieved wide angles support. As a result, the quality of the face filter experience significantly improved. See the video below to estimate the difference.
FaceAR 0.23.0: Neural Network-based Face Detection
Neural networks support for Windows and Web
We’ve been working on making the Face AR features available beyond mobile devices. So we’ve added our neural networks support for Windows and Web for you could implement the best augmented reality user experiences right where you need them, be it your website or photo booth. The full list of neural networks with Windows and Web support include: