Artificial intelligence-based fire monitoring method and device and storage medium

Abstract

The invention relates to the technical field of artificial intelligence, and discloses an artificial intelligence-based fire monitoring method, which comprises the following steps: performing region division processing on an acquired sample image, and acquiring each sub-region corresponding to the sample image; obtaining an RGB mean value of each sub-region, and determining a target risk region in all the sub-regions based on the RGB mean value and a preset risk threshold value; preprocessing a sample image containing the target risk area, and forming training data based on the preprocessed sample image; training a preset neural network model based on the training data until the neural network model converges to a preset range, and forming a fire monitoring model; and based on a real-time monitoring video and the fire monitoring model, carrying out real-time fire monitoring on a monitoring area of the monitoring video. The application scene and accuracy of fire monitoring can be improved.

Description

technical field [0001] The present invention relates to the technical field of artificial intelligence, in particular to a method, device, electronic equipment and computer-readable storage medium for fire monitoring based on artificial intelligence. Background technique [0002] At present, in the fire prevention systems of hospitals, schools, supermarkets, etc., smoke alarms are mostly used to identify smoke, but smoke alarms have obvious limitations in some cases, such as mountains, open warehouses, etc. In open-air or well-ventilated scenes, it is difficult to touch the alarm due to the smoke, or the smoke alarm cannot be triggered due to the low smoke concentration in the air circulation, which will cause the system to fail to report and cause the loss of people's lives and property; secondly, the smoke alarm The device is a hardware device, which is relatively difficult to construct, and the construction and maintenance costs are relatively high. If it is in a forest o...

AI Technical Summary

Problems solved by technology

[0002] At present, in the fire prevention systems of hospitals, schools, supermarkets, etc., smoke alarms are mostly used to identify smoke, but smoke alarms have obvious limitations in some cases, such as mountains, open warehouses, etc. In open-air or well-ventilated scenes, it is difficult to touch the alarm due to the smoke, or the smoke alarm cannot be triggered due to the low smoke concentration in the air circulation, which will cause the system to fail to report and cause the loss of people's lives and property; secondly, the smoke alarm The device is a hardware device, which is relatively difficult to construct, and the construction and maintenance costs are also relatively high. If it is in a forest or farmland, the monitoring range is too large, and it is difficult to cover most areas.

[0003] However, the existing intelligent fire detection model needs to identify each frame of the video, which not only requires a lot of calculation pressure, but also may encounter various factors when a fire occurs, such as lighting, background reflections, and object colors that are too close to the flame color , the distance between the fire situation and the camera equipment, etc., may affect the recognition results of the model, resulting in false positives

However, if the corresponding neural network model is trained according to different scenarios, ideal test results will not be obtained due to over-generalization.

[0004] It can be seen that the existing fire detection methods need to be perfected and improved in terms of applicable scenarios and detection accuracy.

Images