This paper describes an implementation of a novel real-time recognition system which is capable to identify important information from a single user such as gender, age, emotions and hand gestures. The key of this recognition system is the classification process. This is carried out by using several convolutional neural networks that were designed to achieve a high accuracy rate and acceptable response time making use of low computational resources. As a result, this recognition system could be useful in numerous applications like human-computer interaction, person identification, security control and others.