A flame identification method, system, medium and equipment

A flame identification, flame technology, applied in the field of flame identification

Active Publication Date: 2021-05-04
SOUTH CHINA NORMAL UNIVERSITY
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

This type of method has high feasibility and good development prospects, but it still has a large room for development in technical realization, for example, how to quickly and effectively distinguish real flames from suspected flames (such as flame-like lights, reflective lights, etc.) mirrors, etc.); how to achieve accurate identification when the pixels of the monitoring equipment are not high; how to ensure different lighting conditions and different burning materials (flames with different characteristics) due to different seasons, different weather, different indoor environments, etc. accuracy rate etc.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A flame identification method, system, medium and equipment
  • A flame identification method, system, medium and equipment
  • A flame identification method, system, medium and equipment

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0103] This embodiment discloses a flame identification method, comprising the following steps:

[0104] Step 1. Model building steps, such as figure 1 Shown:

[0105] Step 1-1, obtain a training sample set, the training sample set includes a plurality of labeled training samples, the training samples are flame images or non-flame images, and the label of each training sample is flame or no flame;

[0106] Step 1-2, each training sample is used as input, and the label of each training sample is used as output to train the convolutional neural network to obtain a trained convolutional neural network, after removing the fully connected layer in the trained convolutional neural network As a deep feature extraction model;

[0107] Steps 1-3, input each training sample into the depth feature extraction model to extract depth features; extract manual features from training samples; use the depth features and manual features extracted from each training sample as input, and label e...

Embodiment 2

[0192] This embodiment discloses a flame recognition system, including a model building module and a flame recognition module;

[0193] Model building blocks include:

[0194] The training sample acquisition module is used to acquire a training sample set, the training sample set includes a plurality of labeled training samples, the training samples are flame images or non-flame images, and the label of each training sample is flame or no flame.

[0195] The deep feature extraction model building block is used to use each training sample as input and the label of each training sample as output to train the convolutional neural network to obtain a trained convolutional neural network. After the connection layer is removed, it is used as a deep feature extraction model; in this embodiment, the convolutional neural network can use the residual network Residual Network, and the ResNet18 network obtained after removing the fully connected layer in the trained convolutional neural n...

Embodiment 3

[0218] This embodiment discloses a storage medium, which stores a program, and is characterized in that, when the program is executed by a processor, the flame identification method in Embodiment 1 is implemented, specifically as follows:

[0219] Step 1. Model building steps:

[0220] Step 1-1, obtain a training sample set, the training sample set includes a plurality of labeled training samples, the training samples are flame images or non-flame images, and the label of each training sample is flame or no flame;

[0221] Step 1-2, each training sample is used as input, and the label of each training sample is used as output to train the convolutional neural network to obtain a trained convolutional neural network, after removing the fully connected layer in the trained convolutional neural network As a deep feature extraction model;

[0222] Steps 1-3, input each training sample into the depth feature extraction model to extract depth features; extract manual features from ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a flame identification method, system, medium and equipment. Firstly, the model is constructed, and the depth feature extraction model for image depth feature extraction is obtained through training samples, and then the depth features and manual features of the training samples are used as Input training to obtain the flame recognition model; when the flame recognition is to be performed on the image, the deep features in the image are extracted through the deep feature extraction model, and the manual features in the image are also extracted, and finally the deep features and manual features of the image are simultaneously extracted. Input it into the flame recognition model for flame recognition, and the flame recognition model outputs the flame recognition result. The invention combines the manual feature value and the depth feature of the image to identify whether the flame phenomenon appears in the image, and can more accurately and quickly identify the advantages of the flame in the image through the combination of the manual feature and the depth feature.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a flame recognition method, system, medium and equipment combining depth features and manual features. Background technique [0002] With the continuous development of society and the introduction of concepts such as safe cities in my country, people pay more attention to disaster prevention and management. Among the disasters that occur in cities and towns, the danger of fire is self-evident. Although the technology in the field of fire prevention and control has made great progress, timely and accurate detection and fire early warning are still the focus of research. How to effectively extract flame image features, improve the flame recognition rate, and reduce false positives and false negatives is still an important research direction. [0003] In terms of flame detection, the detection range of traditional temperature-sensing and light-sensing fire detectors is re...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Patents(China)
IPC IPC(8): G06K9/62G06K9/46G06K9/00
CPCG06V20/52G06V10/462G06F18/241G06F18/214
Inventor 马琼雄唐钢张宇航罗智明蔡钰波王叶宁陈更生
Owner SOUTH CHINA NORMAL UNIVERSITY
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap