Sony T
Scientists have developed an Artificial Intelligence system that can help to identify people on video, detecting their gender and age more accurately and quickly.
The development has already become with the basis for the offline detection system in the Android mobile apps, according to researchers from the Higher School of Economics in Russia.
Modern neural networks detect the gender on the videos with a 90% of accuracy. The situation with the prediction of age is much more complicated.
Traditional neural networks consider discrete values of the age. In each of the video frames, the network estimates the probability of the person in the image being of a certain age.
For example, if in 30 percent of the frames the top prediction of the network is a person, age as 21 years, and in 10 percent as 60 years, its conclusion will be as follows: with a probability of 30 percent this person is 21, and with a probability of 10 percent, he or she is 60.
Due to various conditions of observations or even slight head rotation, prediction of the same person, age in the different video frames varies in the range of 5 years, minus or plus.
Researchers found a way to optimize natural networks operations. They implemented a novel method to aggregate confidence levels which are produced and conducted by the neural networks for each frame.
The software systems of the facial recognition analysis usually include the several separate neural networks. One of them identifies the person, and another one determines the gender, etc.
An effective neural network with several outputs has been developed. It solves several tasks at a time predict gender and age and produces a set of 1000 numbers that uniquely attribute each person and allows them to be distinguished from other people.
