New AI System Can Detect Age and Gender to Identify People on Video

New AI System Can Detect Age and Gender to Identify People on Video

New artificial intelligence Future can detect the age and type of people in your videos

Due to the different conditions of observation or even slight rotation of the head, the prediction of the same person’s lifetime in different video frames varies in the range of 5 years, more or less.

Scientists have developed an artificial intelligence Future system (AI) that can identify people on video, and discover their age and their type faster and more accurately.

Development has already become the basis for remote detection systems in Android mobile phone applications, according to researchers at the Russian Higher School of Economics.

Modern neural networks discover the genre in videos at 90% accuracy. The situation with age prediction is more complex.

Atificial Intelligence Futre, Traditional neural networks look at separate values for age. In both video frames, the network estimates that the person in the picture is at a certain age.

For example, if the best prediction of the grid in 30% of frames is the age of the person 21 years, and in 10% to 60 years, the conclusion will be as follows: with a probability of 30% is 21, and with a probability of 10%, he or she is 60 .

Due to the different conditions of observation or even slight rotation of the head, the prediction of the same person’s lifetime in different video frames varies in the range of 5 years, more or less.

The researchers found a way to improve the processes of neural networks. They implemented a new method to add levels of confidence produced by the neural network of each frame.

Facial recognition software systems generally include several separate neural networks. Someone identifies someone, identifies the other gender, and so on.

An effective neural network has been developed with many outputs. Solves many tasks at once: predicts age and gender, and produces a set of 1000 numbers uniquely attributed to each person and allows them to distinguish from other people.

According to researchers, this solution will work even in low-performance smartphones.

Please follow and like us:

Leave a Reply

Close Menu