Smoke detection method, system, medium and equipment

Abstract

The invention belongs to the technical field of visual identification, and particularly provides a smoke detection method. The invention aims to solve the problems that in the prior art, smoke detection labeling difficulty is large, and many computing resources are occupied. In order to achieve the purpose, the smoke detection method comprises the following steps: marking a small batch of smoke image samples, training a first detection model, detecting the large batch of smoke image samples by using the first detection model to generate pseudo labels, and enabling the first detection model to apply the labeled small-batch smoke image samples and the large-batch smoke image samples with the pseudo labels, and guiding a second detection model to train to obtain the smoke detection model by applying a knowledge distillation algorithm. Through the semi-supervised learning and knowledge distillation algorithm, the workload of marking the smoke image samples is reduced, and meanwhile, a teacher network with high accuracy is applied to train a student network, so that the occupation of computing resources can be reduced under the condition of ensuring the accuracy.

Description

technical field [0001] The invention belongs to the technical field of visual recognition, and in particular relates to a smoke detection method, system, medium and equipment. Background technique [0002] Fire is one of the disasters that are prone to occur in people's production and life, and smoke is one of the important early features of fire. By detecting smoke, the occurrence of fire can be predicted in time, so that measures can be taken in time to reduce the impact on people's lives and property. caused losses. At the same time, smog is also one of the important indicators of environmental protection testing, and the detection of smog can assist in the assessment of air quality. [0003] Traditional smoke detection methods are mainly based on smoke alarm devices such as smoke sensors. However, this kind of smoke sensor needs to be detected after the smoke arrives, and the spread of smoke in a large space will be affected by height, air flow, etc., making it difficu...

AI Technical Summary

Problems solved by technology

[0005] In order to solve the above-mentioned problems in the prior art, that is, in order to solve the difficulty of labeling a large number of smoke images when using deep learning technology for smoke detection and the large consumption of computing resources when using computers and other equipment to execute corresponding algorithms Problem, the present invention provides a smoke detection method, including:

Images