BiSeNet V2-based pavement image semantic segmentation method and device

A semantic segmentation and image technology, applied in the field of navigation, can solve the problems of inapplicable pavement element semantic segmentation and inability to accurately obtain pavement element semantic segmentation results, so as to improve the extraction ability and deepen the network depth.

Pending Publication Date: 2022-05-31
智道网联科技(北京)有限公司
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, the BiSeNetV2 of the related technology is not suitable for the semantic segmentation of road surface elements when the vehicle is driving automatically, and part of the road surface may be segmented into the background, and the semantic segmentation result of the road surface elements in the road surface image cannot be accurately obtained.

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
  • BiSeNet V2-based pavement image semantic segmentation method and device
  • BiSeNet V2-based pavement image semantic segmentation method and device
  • BiSeNet V2-based pavement image semantic segmentation method and device

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0051] The technical solutions of the embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0052] figure 1 It is a schematic flowchart of a method for semantic segmentation of road surface images based on BiSeNet V2 shown in the embodiment of the present application.

[0053] see figure 1 , a road image semantic segmentation method based on BiSeNet V2, including:

[0054] In step S101, the semantic feature map of 1 / 32 resolution of the road image is obtained through the semantic branch of the transformed BiSeNet V2, where based on BiSeNet V2, convolution is added to each downsampling layer of the semantic branch of BiSeNet V2 Layer, batch normalization layer, activation layer, delete the detail branch of BiSeNet V2, add the global pooling layer, and obtain the modified BiSeNet V2.

[0055] In one embodiment, the Semantic Branch of the existing BiSeNet V2 can obtain road image resolutions of 1 / 2, 1 / 4, 1 / 8, 1 / 16, ...

Embodiment 2

[0066] figure 2 It is another schematic flowchart of the method for semantic segmentation of road surface images based on BiSeNet V2 shown in the embodiment of the present application. figure 2 relative to figure 1 The protocol of the present application is described in more detail.

[0067] see figure 2 , a road image semantic segmentation method based on BiSeNet V2, including:

[0068] In step S201, the semantic feature map of 1 / 32 resolution of the road image is obtained through the semantic branch of the transformed BiSeNet V2, where based on BiSeNet V2, convolution is added to each downsampling layer of the semantic branch of BiSeNet V2 Layer, batch normalization layer, activation layer, delete the detail branch of BiSeNet V2, add the global pooling layer, and obtain the modified BiSeNet V2.

[0069] For this step, reference may be made to the description of step S101, which will not be repeated here.

[0070] In step S202, perform global average pooling on the se...

Embodiment 3

[0087] Corresponding to the aforementioned embodiment of the application function realization method, the present application also provides a BiSeNet V2-based road surface image semantic segmentation device, electronic equipment and corresponding embodiments.

[0088] image 3 It is a schematic structural diagram of a device for semantic segmentation of road surface images based on BiSeNet V2 shown in the embodiment of the present application.

[0089] see image 3 , a road image semantic segmentation device based on BiSeNet V2, including a semantic feature acquisition module 301, a pooling feature acquisition module 302, a fusion feature acquisition module 303, and a segmentation result acquisition module 304.

[0090] The semantic feature acquisition module 301 is used to obtain the semantic feature map of 1 / 32 resolution of the road surface image through the semantic branch of the transformed BiSeNet V2, wherein based on BiSeNet V2, each downsampling layer of the semantic ...

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 road surface image semantic segmentation method and device based on BiSeNet V2. The method comprises the following steps: obtaining a semantic feature map of 1/32 resolution of a pavement image through a semantic branch of a modified BiSeNet V2, and on the basis of the BiSeNet V2, adding a convolution layer, a batch normalization layer and an activation layer on each down-sampling layer of the semantic branch of the BiSeNet V2, deleting detail branches of the BiSeNet V2, and adding a global pooling layer to obtain the modified BiSeNet V2; performing global pooling on the semantic feature map with the 1/32 resolution ratio through the global pooling layer to obtain a global pooling feature map; fusing the global pooling feature map and the semantic feature map with the 1/32 resolution ratio through a feature fusion module of the transformed BiSeNet V2 to obtain a fused feature map; and through the transformed BiSeNet V2, according to the fused feature map, obtaining a semantic segmentation result of the pavement elements of the pavement image. According to the scheme provided by the invention, the semantic segmentation result of the pavement elements in the pavement image can be accurately obtained.

Description

technical field [0001] The present application relates to the technical field of navigation, and in particular, to a method and device for semantic segmentation of road images based on BiSeNet V2. Background technique [0002] In the related art, the road surface elements include road traffic signs, such as various lines, arrows, characters, elevation marks, raised road signs and outline signs marked on the road surface. The accurate recognition of road elements by autonomous vehicles through images is an important prerequisite for autonomous vehicles to complete safe and intelligent driving. Related technologies use neural networks to semantically segment images, and obtain identification results of pavement elements according to the semantic segmentation results, which can provide accurate road information for the navigation of autonomous vehicles, and enable autonomous vehicles to achieve safe autonomous driving functions. [0003] The related technology adopts BiSeNet V...

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): G06V10/26G06V20/58G06V10/80G06V10/764G06V10/82G06K9/62G06N3/04
CPCG06N3/045G06F18/2415G06F18/253
Inventor 孟鹏飞贾双成朱磊李成军
Owner 智道网联科技(北京)有限公司
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