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

Traffic sign recognition method based on dense connection and attention mechanism

A traffic sign recognition and dense connection technology, applied in the field of traffic sign recognition based on dense connection and attention mechanism, can solve problems such as imbalance and slow down training speed, and achieve the effect of high accuracy, bright colors and regular shapes.

Active Publication Date: 2020-08-25
TIANJIN UNIV
View PDF5 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Neural networks usually need to generate a very large set of anchor boxes, only a small part of which overlaps with the ground truth, which creates a huge imbalance between positive and negative samples, slowing down the training speed

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
  • Traffic sign recognition method based on dense connection and attention mechanism
  • Traffic sign recognition method based on dense connection and attention mechanism
  • Traffic sign recognition method based on dense connection and attention mechanism

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0044] The embodiment of the present invention proposes a traffic sign recognition method based on dense connection and channel attention mechanism, see figure 1 , the method includes the following steps:

[0045] 101: Construct a dataset and perform data preprocessing;

[0046]The step 101 is specifically: downloading the data set, the data source is the Chinese traffic sign data set TT100K (Tsinghua-Tencent 100K) released by Tsinghua University, and the data set is intercepted from the street view panorama of Tencent. The training set of the dataset contains 6107 pictures, the test set contains 3073 pictures, and the image size is 2048*2048 pixels. The present invention selects categories whose occurrence frequency is greater than 100 in the data set for training, and there are 45 categories in total.

[0047] 102: Build a traffic sign recognition neural network based on dense connection and attention mechanism through the deep learning framework PyTorch;

[0048] Among t...

Embodiment 2

[0057] The scheme in embodiment 1 is further introduced below in conjunction with specific examples, see the following description for details:

[0058] 201: Construct a dataset and perform data preprocessing:

[0059] (1) The present invention uses the public TT100K (Tsinghua-Tencent 100K) data set, which is divided into two parts: training set and test set. The training set contains 6107 pictures, and the test set contains 3073 pictures, and the size of the pictures is 2048*2048 pixels. TT100K was intercepted with Tencent's street view panorama, covering a total of more than 180 types of traffic signs in China, but many of them are relatively rare and appear less frequently in the data set. The present invention adopts 45 types of traffic signs whose occurrence frequency is greater than 100 in the data set for training.

[0060] (2) Due to GPU memory limitations, the entire image cannot be directly trained for training, so the image in (1) is cropped, and the training set ...

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 recognition method based on dense connection and a channel attention mechanism. The method comprises the following steps: constructing a data set and carrying out data preprocessing; building a traffic sign recognition neural network based on dense connection and attention mechanism through a deep learning framework; inputting the pictures in the training setinto a neural network, obtaining traffic sign types and position information through forward propagation, calculating errors with information in true values, carrying out reverse propagation, and continuously updating network parameters until the errors are not reduced any more; and inputting a picture with a traffic sign, loading the trained model, and outputting a traffic sign recognition result picture. According to the method, the deep features of the network are fully utilized, so that the network has stronger representation capability, and global and local information can be better mixed.

Description

technical field [0001] The invention relates to the field of driving assistance system and target detection, in particular to a traffic sign recognition method based on dense connection and attention mechanism. Background technique [0002] In the existing automatic driving system based on vision methods, target detection is the core task, including lane line detection, vehicle detection, non-motor vehicle detection, pedestrian detection, traffic sign detection, etc. When driving on actual roads, self-driving vehicles must abide by traffic rules and make judgments based on traffic signs and actual road conditions. However, in the face of complex and changeable road scenes, vehicles need to obtain regulated driving information from the surrounding environment. “Tips”, therefore, traffic sign detection algorithms are an integral part of autonomous driving systems. Due to the regular shape and bright colors of traffic signs, in the early stage of research, domestic and foreign...

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): G06K9/00G06K9/62G06N3/08G06N3/04
CPCG06N3/084G06N3/08G06V20/582G06N3/045G06F18/214G06F18/253
Inventor 褚晶辉黄浩吕卫
Owner TIANJIN 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