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

A traffic sign detection method in automatic driving based on a YOLOv3 network

A traffic sign and automatic driving technology, applied in character and pattern recognition, instruments, computer components, etc., can solve problems such as low detection accuracy and detection speed that cannot meet real-time requirements, and achieve enhanced robustness and improved detection Accuracy, reduction of variation and effectiveness of information

Active Publication Date: 2019-03-08
BEIJING INFORMATION SCI & TECH UNIV
View PDF4 Cites 29 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to solve the problem that the existing YOLOv3 network target detection algorithm has low detection accuracy and the detection speed cannot meet the real-time requirements

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
  • A traffic sign detection method in automatic driving based on a YOLOv3 network
  • A traffic sign detection method in automatic driving based on a YOLOv3 network
  • A traffic sign detection method in automatic driving based on a YOLOv3 network

Examples

Experimental program
Comparison scheme
Effect test

specific Embodiment approach 1

[0022] Specific embodiment one: the traffic sign detection method in the automatic driving based on YOLOv3 network described in the present embodiment, this method specifically comprises the following steps:

[0023] Step 1, based on the GTSDB data set, make training set data and test set data with traffic sign target labels;

[0024] Step 2. Cluster the real target frames marked in the training set data, and use the area intersection over union ratio (IOU) as the rating index to obtain the initial candidate target frame of the predicted traffic sign target in the training set data, and use the initial candidate target frame as The initial network parameters of the YOLOv3 network; (the advantage of this is that the convergence speed of the training process can be accelerated); call the initial network parameters of the YOLOv3 network, and input the training set data into the YOLOv3 network for training until the loss function of the training set data output Values ​​less than ...

specific Embodiment approach 2

[0028] Specific implementation mode two: the difference between this implementation mode and specific implementation mode one is: the specific process of the step one is:

[0029] The GTSDB data set contains a total of M images, and after marking the traffic signs in the M images, the marked M images are randomly divided into two parts: a training set and a test set.

specific Embodiment approach 3

[0030] Embodiment 3: This embodiment is different from Embodiment 2 in that: the data volume ratio of the training set and the test set is 8:1.

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 traffic sign detection method in automatic driving based on a YOLOv3 network, and belongs to the field of traffic sign detection. The method solves the problems that an existing YOLOv3 network target detection algorithm is not high in detection precision and the detection speed cannot meet the real-time requirement. According to the invention, an improved loss function isprovided, so that the influence of a large target error on a small target detection effect is reduced, and the detection accuracy of a small-size target is improved. An improved activation function is provided, a negative value is reserved, meanwhile, changes and information propagated to the next layer are reduced, and the robustness of the algorithm to noise is enhanced. The real frames in thetraffic sign data set are clustered by using a K-means algorithm to realize the pre-fetching of a target frame position and accelerate convergence of the network. The detection precision mAP of the traffic sign detection model on a test set reaches 92.88%, the detection speed reaches 35FPS, and the requirement for real-time performance is completely met. The method can be applied to the field of traffic sign detection.

Description

technical field [0001] The invention belongs to the field of traffic sign detection, and in particular relates to a traffic sign detection method in automatic driving. Background technique [0002] Object detection is an important research direction in the field of autonomous driving. Its main detection targets are divided into two categories: stationary targets and moving targets. Stationary targets such as traffic lights, traffic signs, lanes, obstacles, etc.; moving targets such as vehicles, pedestrians, non-motor vehicles, etc. Among them, traffic sign detection provides rich and necessary navigation information for unmanned vehicles during driving, which is a fundamental work of great significance. [0003] Traditional target detection methods are mainly divided into the following steps: preprocessing, selecting candidate regions, extracting target features and feature classification. Commonly used features such as SIFT (scale-invariant feature transform), HOG (histo...

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): G06K9/00G06K9/62
CPCG06V20/582G06V2201/07G06F18/23213G06F18/214
Inventor 王超
Owner BEIJING INFORMATION SCI & TECH 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

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com