Fire and smoke prediction method and system based on transfer learning, and medium

Abstract

The invention relates to a fire and smoke prediction method and system based on transfer learning and a medium, and the method comprises the steps: obtaining a plurality of scene image samples containing fire and smoke, and making a data set according to the scene image samples; on the basis of a transfer learning method, training a pre-acquired pre-training model by adopting the data set, and constructing a trained target prediction model; and predicting a to-be-tested scene image according to the target prediction model to obtain a prediction category of the to-be-tested scene image. According to the fire and smoke prediction method, a large number of training samples are not needed; a good learning effect can be obtained; training time is greatly shortened. The training cost is greatlyreduced, the higher-precision classification effect on the prediction category of the fire disaster and the smoke is achieved, the prediction precision and prediction efficiency of the target prediction model on the prediction category of the fire disaster and the smoke are improved, and therefore the fire disaster and smoke alarm situation can be found conveniently and timely, and the life and property safety of people is prevented from being threatened.

Description

technical field [0001] The present invention relates to the technical field of fire and smoke prediction, in particular to a fire and smoke prediction method, system and medium based on migration learning. Background technique [0002] Nowadays, fire and smog are already a major threat in domestic life and industrial production. Especially in industrial production, it is extremely important to detect fires and dangerous smoke in a timely and accurate manner, and to grasp the fire and smoke alarm conditions. Therefore, industrial production has urgently needed solutions that can detect fire and smoke accurately, reliably and quickly. [0003] In industrial production, surveillance cameras are mainly used to monitor the production environment. When a fire occurs, the environment can be photographed through surveillance cameras, and the captured images can be analyzed and classified through artificial neural networks. Deep learning is the study of simulating the human brain ...

AI Technical Summary

Problems solved by technology

[0004] In the traditional method, it is usually necessary to iteratively train the artificial neural network with a large amount of data, which has problems such as large demand for training samples, long training time, high training cost, and difficulty in guaranteeing the training effect.

Images