The neural network has learned to recognize the faces of people in masks and veils.
An international team of scientists has created a convolutional neural network that, with 99.95% accuracy, recognizes the appearance, gender and age of people with closed faces: in masks, niqabs and veils. Also, in 80.9% of cases, the algorithm for one eye determines whether a person is smiling. Further, scientists plan to test it with more diverse images - for example, taken from different angles, in different lighting, and so on. The technology can be used in the field of security, medicine, and education. It will probably work with low quality images as well.