Smoke detection algorithm based on video analysis

Abstract

The invention relates to a smoke detection algorithm based on video analysis. The algorithm comprises the steps of making a training set, firstly, obtaining a smoke-containing picture from an existingforest smoke video; manually labeling a smoke-containing area, manufacturing a label image, training a full lap machine network model, obtaining the suspected smoking area in the single frame image by using the trained network model, carrying out the dynamic growth analysis on the suspected smoking area, calculating a formula of the area sum of the suspected smoking area corresponding to a continuous video sequence, and calculating a formula of the relative growth of the suspected smoking area. According to the present invention, the forest smoke detection based on the rotary lens is realized, the forest fire is alarmed, the effect of early prediction of the forest fire is achieved, the economic loss caused by the fire is reduced, and a better prospect is brought.

Description

technical field [0001] The invention belongs to the field of deep learning and artificial intelligence, in particular to an image recognition algorithm used for early warning of forest fires. Background technique [0002] The causes and locations of fires are diverse, which hinders fire early warning and fire fighting work. Over the years, both at home and abroad are committed to the research of fire prevention and control, and various attempts have been made. Various types of fire detectors at this stage mainly include heat-sensitive detectors, light-sensitive detectors and smoke-sensitive detectors. These traditional detectors are cheap and highly accurate, but generally have some intractable flaws. For example, due to the relatively long time required for the occurrence of smoke propagation and temperature rise, traditional sensors will inevitably have a response delay. In addition, traditional sensors need to be installed near the fire point and exposed to a large amou...

AI Technical Summary

Problems solved by technology

These traditional detectors are cheap and highly accurate, but generally have some intractable defects

For example, due to the relatively long time required for the occurrence of smoke propagation and temperature rise, traditional sensors will inevitably have a response delay. In addition, traditional sensors need to be installed near the fire point and exposed to a large amount of dust for a long time. Sensors are prone to failure or failure, and are especially not suitable for fire detection in high fire hazard places such as tall spaces or outdoor scenes

Images