Update of MediaPipe, a video and audio processing framework using machine learning

Google company presented framework update MediaPipe, which offers a set of ready-made functions for applying machine learning methods to video and audio processing in real time. For example, MediaPipe can be used to detect faces, track the movement of fingers and hands, change hairstyles, detect the presence of objects and track their movement in the frame. Project code
spreads licensed under Apache 2.0. The models are processed using the TensorFlow and TFLite machine learning frameworks.

Source: opennet.ru