A team of scientists from Russia and Uzbekistan has developed a neural network for automatically recognizing street violence in footage from surveillance cameras. This was reported by the press service of the South Ural State University (SUSU).
The problem was that street surveillance cameras provide a huge stream of data that overloaded the existing neural networks developed for surveillance cameras installed indoors. The resulting neural network recognizes more than 98% of cases of street violence recorded by street surveillance cameras.
The new model uses multi-valued logic, including the values "yes", "no", and "undefined". The third value contains unique mathematical nuances. This allows the neural network to efficiently process a large stream of data.
The development is designed to improve safety in populated areas and can be successfully integrated into the "Smart City" system. The invention will improve monitoring and response to cases of violence in public places.
Earlier, www1.ru reported that in Moscow they created a neural network for calculating holograms — 3D-CGH-Net.
Read materials on the topic:
Russia wants to train AI to create cancer vaccines
AI-based Russian Sign Language translator created in Russia
AI helped reduce mortality from vascular diseases in a Khabarovsk hospital