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

Smoke detection method and system based on foreground and background analysis and electronic equipment

A technology of foreground background and detection method, applied in the field of smoke detection, can solve the problems of false negatives and false positives, and achieve the effect of reducing false negatives and false positives, and improving sensitivity and accuracy.

Pending Publication Date: 2020-12-29
WENZHOU UNIVERSITY
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] Aiming at the deficiencies of the prior art, the present invention discloses a smoke detection method, system and electronic equipment based on foreground and background analysis, which are used to solve the problem that the existing smoke detection method has a relatively high statistical accuracy rate for comprehensive sample sets including various scenes. ; But for each scenario, there are still false negatives and false negatives, which bring troublesome problems in real-world applications

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
  • Smoke detection method and system based on foreground and background analysis and electronic equipment
  • Smoke detection method and system based on foreground and background analysis and electronic equipment
  • Smoke detection method and system based on foreground and background analysis and electronic equipment

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0037] In this embodiment, deep learning is still used for smoke detection, but confrontation generation technology is added, including 3 sub-networks smoke detection network D, smoke removal network G, smoke analysis network S, and 4 difference functions smoke motion loss L S , background difference L B , scene restoration difference L R and background legacy L DS ,Such as figure 1 shown. Among them, the total loss of the network is to extract frame I from the video stream from the camera t , the foreground image is generated by the smoke detection network D Background image generation via smoke removal network G foreground image and background Image Synthesis of New Scene Images the foreground image of the current frame Generate smoke motion loss L through smoke analysis network S S .

[0038] Background image for the current frame Background image with previous frame The difference is Accumulate and leave L for the background B ,which is:

[0039] ...

Embodiment 2

[0050] This embodiment discloses a smoke detection network D, which uses a codec network to generate a foreground image from the current frame It The network parameter is θ D , this process can be described as:

[0051]

[0052] The encoder consists of four convolutional layers that transform the image It into high-dimensional features of size 1 / 24, 512 channels. The decoder uses 4 symmetric deconvolution layers to convert from high-dimensional features to the original size of the smoke channel image Each layer has a kernel size of 5, a stride of 2, and uses leaky ReLU and batch normalization operations. The corresponding layers of the decoder and the encoder are connected by a jump layer to preserve high-level information to ensure that the upsampled pixels can resolve the smoke with high quality. Network structure such as figure 2 As shown, the smoke is gradient, and the generated smoke channel image does not have a large gradient, so the smoke detection loss inclu...

Embodiment 3

[0056] This embodiment discloses a smoke removal network G, which uses a codec network to generate t generate background image The network parameter is θ G , the process is described as

[0057]

[0058] The encoder contains 8 convolutional layers to convert the input into a high-dimensional feature of 512 channels, and the decoder uses 8 symmetrical deconvolution layers to convert the high-dimensional feature into the background image of the original size If the encoder keeps reducing the resolution, and then the decoder keeps increasing the resolution, only in this way there will be problems of gradient disappearance and information imbalance of each layer. In fact, the background content of the surveillance camera rarely changes, and the jump connection can make this part of the information directly used for generation without going through the calculation process of compression and decompression to avoid the loss of information. Network structure such as image 3 ...

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 relates to the technical field of smoke detection, in particular to a smoke detection method and system based on foreground and background analysis and electronic equipment, and the method comprises the following steps: S1, actually deploying an application, and obtaining a video stream from a monitoring camera; S2, using a smoke detection network D to generate a foreground image foreach frame of image; S3, extracting smoke features in the smoke through a convolutional neural network C; S4, sending the smoke features in S3 and the smoke features of the first n frames of images to a recurrent neural network R, and generating smoke probability ps; and S5, judging whether the smoke probability ps is greater than a set threshold value or not, further determining scene smoke, andsending out a fire alarm. According to the invention, a scene is decomposed into a background and a smoke foreground, so that the whole process of smoke generation and development can be separated, and a smoke alarm is given out when the early stage is weak, so as to early warn a fire in advance. The method has the advantages that the sensitivity and accuracy of the algorithm are improved, smokesignals are provided as early as possible, and missing report and false report are reduced.

Description

technical field [0001] The invention relates to the technical field of smoke detection, in particular to a smoke detection method, system and electronic equipment based on foreground and background analysis. Background technique [0002] At present, there are roughly two types of methods for smoke detection based on deep learning: (1) The method of extracting static features based on convolutional neural network can avoid the dependence on artificial features in traditional methods, and can automatically obtain high-level features that are difficult to obtain by traditional techniques. (2) The method of extracting dynamic features based on the cyclic neural network can extract the development and changes of the fire from smoldering to erupting, can eliminate the static approximate smoke, and can achieve relatively high accuracy. Rate. There is also a combination of these two methods, using a convolutional neural network to extract the static features of each frame, and then...

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
IPC IPC(8): G06K9/00G06K9/62G06N3/04G06T7/194G08B17/10G08B17/12
CPCG06T7/194G08B17/10G08B17/125G06T2207/20081G06T2207/20084G06V20/40G06V20/52G06N3/045G06F18/214
Inventor 罗胜
Owner WENZHOU UNIVERSITY
Features
  • Generate Ideas
  • Intellectual Property
  • Life Sciences
  • Materials
  • Tech Scout
Why Patsnap Eureka
  • Unparalleled Data Quality
  • Higher Quality Content
  • 60% Fewer Hallucinations
Social media
Patsnap Eureka Blog
Learn More

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

© 2025 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com