Apple's new patent granted: training CNN algorithm to upgrade pseudo-self expression, which can map user micro-expression to enhance emotional expression.

CTOnews.com, October 20 (Xinhua)-- according to a list recently released by the United States and the Patent Office, Apple has obtained a new technology patent that can further enhance animated facial expressions (Animoji) and personal expressions (Memoji), which can more realistically restore user expressions, thus allowing users to express their emotions more strongly.

Users can create dynamic mimic expressions on compatible iPhone or iPad Pro, which will use your voice and simulate your facial expressions to express their personality and mood.

Apple said in its patent description that such systems tend to be computationally intensive, require high-performance general-purpose processors and graphics processors, and usually do not work properly on mobile devices such as smartphones or tablets.

More importantly, current simulated expression systems usually do not provide the ability to provide nuanced facial expressions or emotional states.

CTOnews.com learned from Apple patents that Apple wants to train and use expression convolution neural networks (CNN) to implement facial images and 3D mesh representations corresponding to facial images in the case of a small number of facial expression feature images, and the potential vectors of the images are compared with a plurality of previously processed potential vectors associated with known emotion types to estimate the emotion type of the image.