in this article, we propose an end-to-end framework to learn a pair of disentangled identity and pose codes for a given data set using one model of generative adversarial network. The framework consists of two encoders, decoder and discriminator.

Зено Б. (науч. рук. Матвеев Ю.Н.) Learning disentangled representation of identity and head pose using generative adversarial networks // Сборник тезисов докладов конгресса молодых ученых. Электронное издание. – СПб: Университет ИТМО, [2019]. URL: https://kmu.itmo.ru/digests/article/1857