Edge identification method based on deep learning

An edge recognition and deep learning technology, applied in the field of edge recognition based on deep learning, can solve the problems of difficult extraction of edge information features and low precision, and achieve low versatility, enhanced robustness, and small target positioning problems Effect

Pending Publication Date: 2021-07-20
GUANGZHOU UNIVERSITY
View PDF4 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, these methods also have certain disadvantages: it is difficult to extract edge information features in special cases through color and shape, such as rainy days, foggy days, occlusions, etc., resulting in low accuracy

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
  • Edge identification method based on deep learning
  • Edge identification method based on deep learning
  • Edge identification method based on deep learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0026] Such as figure 1 As shown, the embodiment is an edge recognition method based on depth learning, mainly including the following steps:

[0027] S1, the network structure is set, and the anchor is set on the network structure, and the target prediction is performed, and the seven-layer convolution layer extracts the network characteristics.

[0028] In this embodiment, the network structure construction is set by normalizing the image by normalizing the image, and dividing the entire image is divided into S × S mesh, each grid detects the central point position of the target image, if the target The image center point position falls in the mesh, set the target possibility PR (Object) = 1, otherwise the target possibility PR (Object) = 0.

[0029] Such as figure 2 As shown, in this embodiment, a quantitative number prediction box is generated by human set anchor anchor, and each prediction box generates coordinate information (X, Y, W, H) and confidence (confidence); , X is t...

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 relates to a learning-based edge identification method, which comprises the following steps: S1, constructing a network structure, setting an anchor point on the network structure, performing target prediction, and designing seven convolutional layers to extract network features; S2, extracting edge information through convolution layer periphery filling, and fusing a residual network and convolution down-sampling operation; S3, adding batch normalization operation to each convolutional layer, adopting a Softmax function for normalization, carrying out 0 and 1 classification on the image, and generating a target probability possible value; and S4, carrying out target network detection, and adopting a multi-scale training mode to adjust the resolution of an input image. According to the invention, a T-YOLO detection algorithm is provided from the edge information of the target, the problem of low detection speed is solved, the accuracy of target positioning is realized, and the recognition precision is improved.

Description

Technical field [0001] The present invention relates to the field of edge recognition, and more particularly to an edge recognition method based on depth learning. Background technique [0002] At present, the application scenario of edge recognition technology is increasing, for example, identifying handwritten, identifying facial contours, identifying traffic signs. In the existing edge recognition technique, by using the boundary to find the area, the identification and scene analysis of the object can be realized, due to the target edge, the image texture characteristics, etc., the edge detection, therefore, in many methods of edge detection, existence Hold various limitations and insufficient, such as: slow detection speed, low recognition accuracy, and unable to achieve a small target precision positioning. [0003] The existing color space recognition method, or identifies edge information according to the shape characteristics, or with color and shape feature fusion ident...

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): G06T7/13G06K9/46G06K9/62G06N3/04G06N3/08
CPCG06T7/13G06N3/084G06V10/462G06N3/047G06N3/045G06F18/23213G06F18/2415
Inventor 朱静凌兴涛明家辉王坤辉李林钟绮岚何泳隆赵宣博尹邦政谢斌盛
Owner GUANGZHOU UNIVERSITY
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
Try Eureka
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap