Unlock instant, AI-driven research and patent intelligence for your innovation.

An Attention Target Recognition Method Based on Convolutional Neural Network

A convolutional neural network and target recognition technology, applied in the field of attention target recognition based on convolutional neural network, can solve the problems of the disappearance of network gradients and the inability of the network to learn features, and achieve the effect of improving accuracy

Active Publication Date: 2021-11-09
TIANJIN UNIV
View PDF8 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, when the number of layers of the model reaches a certain number, it will cause the problem that the gradient of the network disappears during the backpropagation learning process, and the network will not be able to effectively learn features.

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
  • An Attention Target Recognition Method Based on Convolutional Neural Network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0022] A convolutional neural network-based attention target recognition method of the present invention will be described in detail below with reference to the embodiments and the accompanying drawings.

[0023] A convolutional neural network-based attention target recognition method of the present invention is mainly aimed at two aspects: one is the feature enhancement operation, which increases the weight of the target object pixel in the feature map by multiplying and weighting the feature map; It is through the attention mechanism that the network can quickly focus on the target object in the feature map, and the combination of these two aspects of feature operations makes the feature map to be detected more capable of representing the target object.

[0024] Such as figure 1 Shown, a kind of attention target recognition method based on convolutional neural network of the present invention comprises the following steps:

[0025] 1) Obtain the features of the image throug...

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

An attention target recognition method based on a convolutional neural network of the present invention, through two major processes of feature enhancement and feature attention on the feature maps of different levels of the convolutional neural network, so as to obtain features that are more capable of characterizing the target object , and then perform target detection operations on the obtained feature maps, which not only retains the small target information in the shallow feature map, but also retains the large target information in the deep feature map, making the feature map more capable of representing the target object. As a single-stage object detector, it greatly improves the accuracy of object detection while ensuring the efficiency of object detection.

Description

technical field [0001] The invention relates to an attention target recognition method. In particular, it involves an attentional object recognition method based on convolutional neural networks. Background technique [0002] Target detection is a very challenging task in the field of computer vision. In recent years, convolutional neural networks have been applied to target detection tasks and have achieved remarkable results, which has aroused researchers' interest in the convolutional neural network model. Research interest, the number of layers of the model is also continuously deepened. However, when the number of layers of the model reaches a certain number, it will cause the problem that the gradient of the network disappears during the backpropagation learning process, and the network will not be able to effectively learn features. After the deep residual network was proposed, it solved the problem of gradient disappearance well, so that the neural network model ca...

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): G06K9/46G06N3/04G06N3/08
CPCG06N3/084G06V10/40G06V2201/07G06N3/048G06N3/045
Inventor 冀中孔乾坤李晟嘉
Owner TIANJIN UNIV