Method and device for early warning fire in highway tunnel based on video recognition

Abstract

The invention discloses a method and a device for early warning fire in a highway tunnel based on video recognition. The method comprises the following steps: acquiring video data acquired by a multi-channel video acquisition device of a highway tunnel to be monitored; according to a preset image deep learning model, judging whether each path of video data contains a fire source with smoke and / or flame characteristics; and when the at least two paths of video data are judged to contain the mobile fire source, judging whether the fire sources in the at least two paths of video data are the same fire source or not according to a preset fire source fusion model, and obtaining fire early warning information containing fire behavior grade and fire source quantity information. According to the embodiment of the invention, early warning is carried out according to the fire early warning information, so that by adopting the embodiment of the invention, a plurality of same mobile fire sources can be prevented from being mistakenly considered as a plurality of fire events, multiple times of early warning are carried out, a large amount of redundant information is prevented from being generated, the early warning efficiency is improved, and the early warning accuracy is also improved.

Description

technical field [0001] The invention relates to the technical field of fire early warning, in particular to a video recognition-based fire early warning method and device in a highway tunnel. Background technique [0002] Due to the tight and narrow structure of the tunnel, the consequences of a fire are often very serious. Tunnel fires occur rapidly, the flame temperature is high; the smoke spreads quickly, there are many toxic gases, and the visibility is low, which seriously endangers the lives and property safety of personnel in the tunnel. Tunnel fires are highly random, and are often accompanied by vehicle congestion and difficulty in transporting rescue supplies. [0003] The following fire early warning systems and corresponding early warning methods adopted in the prior art have different technical defects, as follows: [0004] An existing fire early warning system is a linear fire detection and early warning system based on the temperature detection of the temper...

AI Technical Summary

Problems solved by technology

[0004] An existing fire early warning system is a linear fire detection and early warning system based on the temperature detection of thermal gratings. The technical defects of this kind of fire early warning system are: the early warning time is too long, the sensitivity is low, therefore, it is impossible to monitor smoke It is impossible to determine the size of the fire

[0005] Another kind of fire early warning system is a point-test fire early warning system based on dual-wavelength spectral analysis. The technical defects of this type of fire early warning system are: easy to be interfered by smoke, high false alarm rate, and unable to detect the size of the flame

[0006] There is also a fire alarm system, as disclosed in CN110517441A based on the deep learning framework embedded flame or smoke video alarm system and method, the fire detection system and method cannot judge whether the fire is a fire when multiple cameras recognize the same fire at the same time. The same fire source, applied to scenes such as highway tunnels, multiple cameras to identify the same moving vehicle on fire will be misidentified as multiple fire incidents, resulting in a large amount of redundant information, which brings great troubles to users

Images