System and method for identifying fire in video based on convolutional neural network

Abstract

The invention provides a system and method for identifying a fire in a video based on a convolutional neural network, and relates to the technical field of deep learning. The system comprise a user login and registration module, a user information management module, a video fire identification module and an early warning information notification module, wherein the video fire identification moduleis a core part. The method comprises the following steps: firstly, carrying out frame extraction processing on a target video by using OpenCV; carrying out preprocessing such as random parameter change and image enhancement on the image; obtaining input data of the neural network; then, creating a convolutional neural network model LeNet-5 by means of TensorFlow; and reading the image data in thetraining set to train the model, performing persistence on the model with the best performance on the test set, finally inputting the preprocessed image data into the trained model for identification, and analyzing and displaying a result.

Description

technical field [0001] The invention relates to the field of deep learning technology, in particular to a convolutional neural network-based fire recognition system and method in videos. Background technique [0002] In recent years, there have been many major fire accidents in China, causing great casualties and property losses, which makes fire early warning and fire safety more and more important in today's era. With the continuous development of modern science and technology, image recognition technology has gradually expanded to fields such as industrial automation, natural resource analysis, physiological lesion research, and disaster climate discovery. People have begun to try to apply image recognition technology to fire early warning research. [0003] At present, the research on fire early warning is mainly divided into three aspects: hardware, software and the combination of software and hardware. In terms of hardware, most of the existing research is to improve ...

AI Technical Summary

Problems solved by technology

[0006] In terms of hardware research, although the recognition rate is very high at the initial stage of hardware equipment installation, when the equipment is aging or in disrepair for a long time, its accuracy rate is often greatly reduced

The research on the combination of software and hardware, although it performs well in some special scenarios, is difficult to be widely promoted due to the need for special equipment, resulting in high costs.

Images