In this paper we propose an algorithm for recognition of the flame on the video sequence. The subject area is the oil and gas industry and its features in the framework of the problem. The current technical means for fire detection are described, the restrictions in their application are characterized. Various approaches to flame recognition on video sequences are mentioned, as well as the need for their joint use to create a reliable model for recognition of the flame on he pictures of video stream.

Keywords: artificial intelligence, fire safety, automation, color mask, Haar cascade, morphological erosion, oil and gas industry, flame recognition, fire automation, video sequence, each-frame analysis