The method of fire detection at an early stage and measures of it’s elimination it in public buildings are described. An artificial intelligence system that combines heat, smoke sensors and image processing from cameras is presented. The technique has passed model studies and has shown its effectiveness. Special attention is paid to data analysis algorithms that can reduce the probability of false positives and increase the reaction speed. The key stages of fire detection, threat classification and decision-making are considered. The results of testing in real conditions are also presented, demonstrating the high accuracy of the system. The prospects of integrating the technology with existing fire protection systems and its possible application in large public and administrative buildings are discussed.