Artificial intelligence

Abstract: The article is devoted to the development and application of computer vision methods to identify potential fire risk factors in rooms and offices and fire prevention. We offer a method for detecting a fire using a sequence of color images from a surveillance camera. The system of fire detection in the room was developed and the algorithm of checking it for accuracy. This work is aimed at detecting fires using cameras with different characteristics and, first of all, with poor performance in order to reduce the cost of such a system. Initially, the system performs processing and preparation of data, and then analyzing the data, detects fire. YCbCr was selected as a color model to reduce conversion costs. Also, given the limited resources, it was decided to develop a simplified quick algorithm to calculate the base scene to detect objects that appeared with the linear execution time. We use MNI for motion detection in an image. To reduce false positives, it was decided to use the information about gradients as the flame has characteristic contours. As a classifier, we used a linear combination of hog values in directions. We used a neural network to confirm the fact of ignition detected by the CCTV system.

Keywords: artificial intelligence, computer vision, recognition, fire danger, fire points

References: