Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Method for depth learn pattern recognition of fire image

A technology of learning mode and image depth, which is applied in character and pattern recognition, instruments, computer parts, etc., can solve the problems of small sample overfitting, parameter redundancy, false negative, etc., to eliminate overfitting and improve accuracy The effect of high rate and fast network speed

Inactive Publication Date: 2018-12-21
YANSHAN UNIV
View PDF1 Cites 36 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] In summary, there is no effective solution to the problems of low accuracy, false positives, missed negatives, parameter redundancy and small sample overfitting in the existing technology of fire image recognition.

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
  • Method for depth learn pattern recognition of fire image
  • Method for depth learn pattern recognition of fire image
  • Method for depth learn pattern recognition of fire image

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0047] The present invention will be further described below in conjunction with accompanying drawing:

[0048] Feature representation refers to the activation value of an image in a certain layer of CNN, and the size of feature representation should be slowly reduced in CNN. High-dimensional features are easier to process, and training on high-dimensional features is faster and easier to converge. Spatial aggregation is performed on low-dimensional embedding space, and the loss is not very large. The explanation for this is that there is a strong correlation between adjacent neurons, and the information is redundant.

[0049] Balanced network depth and width. If the width and depth are appropriate, the network can have a relatively balanced computing budget when applied to distributed systems.

[0050] figure 1 It is a flowchart of the present invention, comprising the following steps:

[0051] Step 1, input fire images for preprocessing, as training samples and test sampl...

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 fire image depth learning pattern recognition method, comprising the following steps: preprocessing a test sample and a training sample of the fire image; designing an improved GoogleNet-based deep learning network, which extracts features automatically through training to eliminate artificial traces; deep features are used to train the classifier and identify the fire test samples. The invention applies the image multi-scale convolution fusion network to the fire image pattern recognition technology, remarkably improves the training efficiency of the network, effectively solves the problems encountered in the fire recognition method, such as the accuracy and real-time property are not ideal, the network structure is complex, and the training time is long, the stability and the robustness are poor, and the like. The recognition accuracy of the trained network is 99.2% in three types of fire images.

Description

technical field [0001] The invention relates to the field of fire detection, in particular to a fire image deep learning pattern recognition method. Background technique [0002] When a fire breaks out, it will cause great harm to people's lives and property. If it can be found in the early stage of the fire and make an alarm in time, it will have important practical significance for reducing personal and property losses and gaining time for rescue. Collect fire data at the fire scene through sensors, cameras and other detection equipment, input the collected data into the fire recognition model, judge the occurrence of fire, and feed back the recognition results to the alarm control system to give an alarm to the fire situation. To achieve timely and accurate identification and alarm when it occurs, a reliable fire identification method with high identification accuracy is needed. [0003] Fire recognition is the core of the whole fire detection technology, and its recogni...

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 Applications(China)
IPC IPC(8): G06K9/62
CPCG06F18/241G06F18/214
Inventor 张秀玲侯代标董逍鹏
Owner YANSHAN UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products