In connection with the recent terrorist attacks in Paris and Brussels, the need to create special technical means for identifying criminals and terrorists in a crowd has once again become obvious. In some European countries and the USA, such systems are already in operation.
It just became known about the success of the specialists of the Tomsk Polytechnic University (TPU). Their efforts have developed a unique algorithm for identifying faces in a crowd, recorded by surveillance cameras. It allows you to accurately determine the number of those present, determine their age, gender, and calculate those who are on the wanted list.
According to the project manager, Professor Vladimir Spitsyn, the method they proposed is based on machine learning technologies that are actively used in many areas. In this case, we are talking about a special image classification method called a convolutional neural network.
Video equipment equipped with such an algorithm is capable of working with 4K Ultra HD resolution in automatic mode and promptly monitoring the situation, analyzing it, including in the event of emergency situations. The efficiency of the algorithm for processing video information is tenfold higher than that of existing foreign analogs.