A fire early warning method based on machine learning to monitor smoke in video images

Abstract

The invention discloses a fire early warning method based on machine learning to monitor smoke in video images, which is characterized in that it comprises the following steps: Step 1) collecting and marking picture data sets of various smoke scenes, wherein non-fire early warning smoke scenes belong to It is class A, and the fire early warning smoke scene is classified as B class; Step 2) context target detection layer non-fire early warning smoke scene training: step 3) context target detection layer fire early warning smoke scene training, repeat step 2), the training picture is Class B fire early warning smoke picture; step 4) detection of suspected fire smoke picture. The beneficial effect achieved by the present invention is to solve the problem that the traditional machine learning method classifier cannot accurately distinguish whether the detected smoke is caused by fire. The invention utilizes the context object detection method to judge the context relationship of the area where the smoke is located, and lowers the false alarm and missing alarm rate on the premise of increasing the fire early warning rate.

Description

technical field [0001] The invention relates to a fire early warning method for monitoring video image smoke based on machine learning, and belongs to the technical field of video image processing. Background technique [0002] As we all know, when a fire is in the smoldering stage at the beginning of the fire or when the flame is small, smoke has already been produced, and the smoke has the characteristics of fast information transmission in a large space. With the development of computer vision, digital image processing, machine learning and other technologies, the laying of artificial intelligence cameras, video-based fire detection and early warning technology has gradually been researched and progressed. Fire detection and early warning technology based on video images is a new fire detection and early warning method based on digital image processing and analysis. Fire detection based on digital image processing has low cost, high accuracy and large amount of informatio...

AI Technical Summary

Problems solved by technology

[0004] In order to solve the deficiencies in the prior art, the object of the present invention is to provide a fire warning method based on machine learning to monitor smoke in video images, which solves the problem that the traditional machine learning method classifier cannot accurately distinguish whether the detected smoke is caused by fire.

Images