Surveillance video straw burning inspection method based on deep neural network

Abstract

The invention discloses a surveillance video straw burning inspection method based on a deep neural network, which belongs to the technical field of information. A smoke / fire recognition and processing module based on a neural network classifier is built based on a deep learning neural network frame. The smoke / fire recognition and processing module adopts a method of intercepting each ten frames of video as a picture to increase a data set and amplify a data set feature label, a residual network is optimized, and the accuracy of recognizing an abnormal event in an image and a video is thus improved. According to judgment results towards smoke and fire in certain areas by the neural network classifier, the smoke / fire recognition and processing module gives an alarm to an inspection person and common users automatically, and monitoring on a straw burning event is further realized.

Description

technical field [0001] The invention belongs to the field of information technology, in particular to a surveillance video inspection method for straw burning based on a deep neural network, which is mainly used for environmental monitoring. Background technique [0002] With the development of economy and science and technology, the problem of environmental pollution is becoming more and more serious, especially the air pollution and smog phenomenon all over China. However, under the mainstream of "environmental protection", the phenomenon of straw burning is still repeatedly banned. In the traditional method, the patrol personnel generally go to the relevant area to deal with the straw burning site according to the video taken by the surveillance camera. However, when there are no abnormal events, a lot of manpower and material resources will be wasted. Therefore, it is very necessary to improve the inspection efficiency and intelligently extract key information. [0003]...

AI Technical Summary

Problems solved by technology

[0007] In order to solve the inefficiency of traditional monitoring of straw burning and the misjudgment of pyrotechnic events due to the interference of other things in the scene, the present invention provides a monitoring video straw burning inspection method based on deep neural network, and constructs a pyrotechnic recognition processing module , using the neural network classifier to automatically identify straw burning events in the monitoring video, and actively send SMS alarms, which realizes the efficiency and interactivity of information, and greatly saves manpower and material resources

Images