A flame detection method based on fire-mcanet

By improving the Fire-MCANet model and combining data augmentation and the MCA module, the problems of slow flame detection speed and insufficient accuracy were solved, achieving faster and more accurate flame detection.

CN116844016BActive Publication Date: 2026-07-21HUAIYIN INSTITUTE OF TECHNOLOGY
View PDF 4 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
HUAIYIN INSTITUTE OF TECHNOLOGY
Filing Date
2023-05-24
Publication Date
2026-07-21

AI Technical Summary

Technical Problem

Existing flame detection methods suffer from problems such as high computational cost, large number of parameters, susceptibility to overfitting, and lack of attention mechanisms, resulting in slow detection speed and insufficient accuracy.

Method used

We adopt the Fire-MCANet model and construct a new MCA Block network structure by replacing the convolutional module with data augmentation, CA attention mechanism and MCA module, which reduces the amount of computation and parameters, while improving the feature extraction capability.

Benefits of technology

It improves the speed and accuracy of flame detection, reduces the overall computational load and number of parameters of the model, and enhances the accuracy of flame detection.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116844016B_ABST
    Figure CN116844016B_ABST
Patent Text Reader

Abstract

The application discloses a flame detection method based on Fire-MCANet, which can be applied to flame detection problems in various environments. The detection algorithm is mainly based on pictures after data enhancement processing, increases the CA attention mechanism, and combines the MCA module and the MCA Block module to create a new MCANet network model. First, the collected picture data is cleaned, and all the cleaned pictures are subjected to data enhancement. Then, the flame images subjected to data enhancement are imported into the Fire-MCANet model for training, so that a target model is obtained to realize flame detection. The method can effectively judge whether the current image contains flame and make accuracy prediction, improve the accuracy of flame detection, and increase the use value of the deep learning network in detection.
Need to check novelty before this filing date? Find Prior Art