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

Improved YOLOv3 traffic sign detection method based on multi-scale feature layer

A multi-scale feature and traffic sign technology, applied in the field of computer vision, can solve the problem of not being able to enhance the detection ability of YOLOv3, and achieve the effect of alleviating the imbalance of sample categories and improving the detection accuracy

Pending Publication Date: 2021-03-19
YANSHAN UNIV
View PDF10 Cites 6 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Using this method does not enhance the ability of YOLOv3 to detect different target scales.

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
  • Improved YOLOv3 traffic sign detection method based on multi-scale feature layer
  • Improved YOLOv3 traffic sign detection method based on multi-scale feature layer
  • Improved YOLOv3 traffic sign detection method based on multi-scale feature layer

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0045] A traffic sign detection method based on an improved YOLOv3 multi-scale feature layer, which includes the following steps:

[0046] Step 1. Prepare the traffic sign data set and perform data enhancement, which specifically includes the following two steps:

[0047](1) Using the TT100K public data set, the training set contains a total of 6103 pictures, the test set contains a total of 3067 pictures, and the picture resolution is 2048*2048 high-definition pictures. Less, it is difficult for the network to learn its features. Therefore, this patent uses 45 traffic signs whose frequency exceeds 100 times for training. And, in order to learn enough features, the data set is divided into training set and test set with a ratio of 8:2. At this time, the label format of the data set is a json file. In order to be able to use it on the network, it needs to be manually converted to the VOC format.

[0048] (2) In the generative data enhancement method, the standard template of...

Embodiment 2

[0070] A traffic sign detection method based on an improved YOLOv3 multi-scale feature layer, which includes the following steps:

[0071] Step 1. Prepare the traffic sign data set and perform data enhancement, which specifically includes the following two steps:

[0072] (1) Using the TT100K public data set, the training set contains a total of 6103 pictures, the test set contains a total of 3067 pictures, and the picture resolution is 2048*2048 high-definition pictures. At this time, in order to learn enough features, the data The set adopts the ratio of 8:2 to divide the training set and the test set. At this time, the label format of the data set is a json file. In order to be able to use it on the network, it needs to be manually converted to the VOC format.

[0073] (2) In the generative data enhancement method, the standard template of the traffic sign with a small number of categories in the training set is first randomly enhanced according to the set probability, and...

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 an improved YOLOv3 traffic sign detection method based on a multi-scale feature layer, and the method comprises the following steps: 1, preparing a traffic sign data set, and carrying out the data enhancement; 2, establishing a YOLOv3 improved network model, improving a backbone network, replacing an original Darknet53 with a Densenet network, and optimizing the Densenet network; 3, training the YOLOv3 improved network model; 4, using the optimal training model to detect the traffic sign. The method is balanced in sample category, is high in detection capability for different target scales, and can detect smaller traffic signs.

Description

technical field [0001] The invention relates to the field of computer vision, in particular to an improved YOLOv3 traffic sign detection method based on multi-scale feature layers. Background technique [0002] With the development of science and technology and the improvement of people's living standards, more and more cars are produced. Cars seem to be a means of transportation for people to travel and travel. People's lives are more convenient, but it also brings traffic congestion and traffic accidents. . In order to reduce the occurrence of these things, the intelligent transportation system emerged. Traffic sign recognition is a very important part of the intelligent transportation system. It plays an important role in the process of people driving on the road. Its working principle is to use the camera in the The traffic sign pictures are collected in the outdoor complex scene, and then the corresponding area is extracted, the traffic sign is detected from the backgr...

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): G06K9/00G06K9/62G06N3/04G06N3/08
CPCG06N3/084G06V20/582G06N3/045G06F18/23213G06F18/2431G06F18/253
Inventor 李国强付乐孙英家方奇王天雷
Owner YANSHAN UNIV