Air conditioner outdoor unit portrait intelligent detection method based on visual attention and multi-scale convolutional neural network

A convolutional neural network and visual attention technology, applied in biological neural network models, neural architectures, computer components, etc., can solve the problems of low information processing efficiency, increase the difficulty of calculation and analysis, etc., and achieve the effect of improving performance

Inactive Publication Date: 2021-04-16
SHANDONG UNIV
View PDF6 Cites 7 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Although the multi-scale network has advantages in feature extraction, it still has certain limitations in portrait detection: when the multi-scale network extracts features, irrelevant information and attention information are trained at the same time, and the weights are allocated equivalently, which will increase the computational analysis. Difficulty, low information processing efficiency

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
  • Air conditioner outdoor unit portrait intelligent detection method based on visual attention and multi-scale convolutional neural network
  • Air conditioner outdoor unit portrait intelligent detection method based on visual attention and multi-scale convolutional neural network
  • Air conditioner outdoor unit portrait intelligent detection method based on visual attention and multi-scale convolutional neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0041] An intelligent detection method for portraits of air-conditioning outdoor units based on visual attention and multi-scale convolutional neural networks, comprising the following steps:

[0042] (1) Data preprocessing: Manually classify the image samples of the air-conditioning external units. The areas of concern in the image samples are the icon and the color of the connecting pipe nozzle. According to whether there is an icon in the image sample and whether the icon matches the model of the external unit, Whether there is a connecting pipe, whether the color of the nozzle of the connecting pipe matches the model of the external unit, and generate two kinds of labels, correct and incorrect. The image of the external unit of the air conditioner whose color of the nozzle of the connecting pipe matches the model of the external unit of the air conditioner, otherwise, it is a sample with the wrong label.

[0043] (2) Read the sample image preprocessed in step (1), input it...

Embodiment 2

[0049] According to the intelligent detection method for the portrait of an air conditioner outdoor unit based on visual attention and multi-scale convolutional neural network described in Embodiment 1, the difference lies in:

[0050] Such as figure 2 As shown, the visual attention network is stacked by multiple residual attention modules, and each residual attention module contains two branches: the backbone branch and the mask branch; the backbone branch is the basic residual network structure, and the image Perform feature extraction to generate a feature map with the same size as the original image; the mask branch is a structure combining top-down and bottom-up, first through the residual module and the downsampling layer, gradually extract high-level features and increase the residual For the receptive field of the difference module, the downsampling is completed by pooling, and then through the upsampling layer with the same number of downsampling layers, the size of ...

Embodiment 3

[0057] According to the intelligent detection method for the portrait of an air conditioner outdoor unit based on visual attention and multi-scale convolutional neural network described in Embodiment 2, the difference is that:

[0058] The multi-scale network is a three-branch model including different convolutional layers, which can effectively extract feature abstractions with different levels, and use the convolutional layers to adjust the scale. Smaller resolutions can display more local features. While higher resolution can display more global features, the combination of the two can effectively improve network performance. In the three-branch model, a residual connection is made between different layers with the same scale feature map; the residual connection between different layers helps the features in the multi-scale network to perform identity mapping in the forward process, when The output of the shallow network has reached the optimum, and the layer behind the dee...

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 an air conditioner outdoor unit portrait intelligent detection method based on visual attention and a multi-scale convolutional neural network, and the method comprises the following steps: (1) data preprocessing: carrying out the manual classification of an air conditioner outdoor unit portrait sample, and generating a correct label and a wrong label; (2) reading the preprocessed sample image, inputting the preprocessed sample image into a visual attention network, and generating an attention distribution diagram; (3) inputting into a multi-scale network for training to obtain a deep fusion feature vector; (4) taking the deep fusion feature vector as the input of a softmax classifier model for training; (5) inputting the verification sample set into a softmax classifier model to verify the classification precision to obtain a trained softmax classifier model; and (6) inputting the test sample set into the trained softmax classifier model to obtain a correct or wrong classification result of the test sample set. And the conduction gradient is helped in the reverse process, so that a deeper model can be successfully trained, and the performance of the network is improved.

Description

technical field [0001] The invention relates to the technical field of intelligent detection of portraits of external air conditioners, in particular to an intelligent detection method for portraits of external air conditioners based on visual attention and multi-scale networks. Background technique [0002] Due to the difference in the model of the air conditioner's external unit, the affixed icons will be different, and the connecting pipes used will also be different. Using manual detection is time-consuming and labor-intensive. In the context of the industrial Internet, we hope to apply neural networks to image detection to replace the traditional method of relying on manual detection, so as to be able to quickly and in real time judge whether the supporting connecting pipes and icons of the product are accurate. At the same time, the results are fed back to the factory in real time, and the air conditioner external unit inspection is completed with high efficiency and l...

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): G06K9/40G06K9/62G06N3/04
Inventor 袁东风狄子钧梁聪
Owner SHANDONG UNIV
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