Eureka AIR delivers breakthrough ideas for toughest innovation challenges, trusted by R&D personnel around the world.

Rapid identification method based on lightweight DAB-Net network

A recognition method and a lightweight technology, applied in the field of target recognition, can solve problems such as difficulty in meeting real-time requirements and great impact on recognition effects, and achieve fast speed and meet real-time requirements

Pending Publication Date: 2020-07-03
SHENYANG POLYTECHNIC UNIV
View PDF2 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Benefits of technology

This patented system improves upon previous methods by optimizing parameters for better performance while reducing weight compared to existing systems that use multiple processors or threads simultaneously. It also allows for faster data analysis times due to its ability to be optimized over different dimensions (fourth dimension) at once. Overall, this new approach provides improved efficiency and accuracy in detecting road signage accurately across various resolutions.

Problems solved by technology

This patented technical problem addressed in this patents relates to improving the accuracy and speed at which identifying targets (target objects) within images captured with cameras or sensors becomes possible due to their high dimensionality compared to other types of imagery. Current techniques have limitations including slow response times caused by large amounts of time required for pixel values to transfer across different layers before being detected.

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
  • Rapid identification method based on lightweight DAB-Net network
  • Rapid identification method based on lightweight DAB-Net network
  • Rapid identification method based on lightweight DAB-Net network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0040] The ground traffic sign recognition method based on semantic segmentation usually uses large networks such as VGG16 and Res101 in the front-end feature extraction layer. Using a large network to improve detection accuracy usually reduces detection speed. In order to improve the detection speed, the feature extraction network can be replaced with a lightweight network, and at the same time, the amount of calculation is reduced by using atrous convolution and convolution decomposition to increase the receptive field.

[0041] A fast identification method based on a lightweight DAB-Net network. This method is based on a lightweight DAB-Net network structure, performs model quantification, and uses multi-thread optimization technology. The specific steps of this method are:

[0042] Such as figure 1 , figure 2 As shown, Step 1: Based on the lightweight DAB-Net network structure;

[0043] The lightweight DAB-Net network structure is as follows:

[0044]The original imag...

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 belongs to the field of target identification, and particularly relates to a rapid identification method based on a lightweight DAB-Net network. The method includes: taking a lightweightDAB-Net network structure as a basis, carrying out model quantification; parallelizing the algorithm by adopting multi-thread pipelined calculation; decomposing the algorithm into a plurality of tasks, allocating corresponding pairs of threads, and executing an independent task by each thread; gPU forward propagation operation time is taken as the minimum cycle period, and image reading, data input processing, data output processing and Opencv subsequent processing are included in GPU operation time consumption; awakening other waiting threads through a thread synchronization signal to achieve thread synchronization; data input at the current moment comes from data output of the previous moment and the previous state, and parallel algorithm optimization acceleration is achieved by delaying data output of four frames. According to the invention, the identification speed is improved based on the lightweight DAB-Net network.

Description

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

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
Owner SHENYANG POLYTECHNIC 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
Eureka Blog
Learn More
PatSnap group products