Image semantic segmentation method adopting inverse attention and pixel similarity learning

A technique for semantic segmentation and attention, applied in the field of computer vision

Inactive Publication Date: 2018-02-09
盐城禅图智能科技有限公司
View PDF4 Cites 14 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide an image semantic segmentation method using inverse attention and pixel similarity learning, which solves the image segmentation problem of complex image scenes

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
  • Image semantic segmentation method adopting inverse attention and pixel similarity learning
  • Image semantic segmentation method adopting inverse attention and pixel similarity learning
  • Image semantic segmentation method adopting inverse attention and pixel similarity learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0077] Embodiment 1: On the PACALConT1exT1 data set, use a kind of image semantic segmentation method that refers to inverse attention and pixel similarity learning according to the present invention and FCN-8s, BoxSup, ConT1exT1, VeryDeep, DeepLabv2-ASPP, RefineNeT1-101 Compared with RefineNeT1-152, HolisT1ic, and ModelA2, the experimental results are shown in Table 1:

[0078]

[0079]

[0080] Table 1 Comparison results of different semantic segmentation methods on the PASCALConT1exT1 test set

[0081] As can be seen from Table 1, the comparison results on the PASCALConT1exT1 test set, a kind of image semantic segmentation method that refers to inverse attention and pixel similarity learning according to the present invention has three points in terms of pixel accuracy, average pixel accuracy and average IoU accuracy. In terms of all indicators, it is better than other methods, and the accuracy has increased by about 1%.

[0082] On the PACALPerson-ParT1 data set, an...

Embodiment 2

[0086] Embodiment 2: On the NYUDv2 data set, an image semantic segmentation method using inverse attention and pixel similarity learning according to the present invention and GupT1aeT1al, FCN-32s, ConT1exT1, HolisT1ic, RefineNeT1, DeepLabv2-ASPP, DeepLab-LFOV For comparison, the experimental results are shown in Table 3:

[0087]

[0088]

[0089] Table 3 Comparison results of various semantic segmentation methods on the NYUDv2 test set

[0090] It can be seen from Table 3 that an image semantic segmentation method using inverse attention and pixel similarity learning according to the present invention obtains a better average IoU accuracy, reaching 42.1%.

Embodiment 3

[0091] Embodiment 3: On the MIT1ADE20K data set, use a kind of image semantic segmentation method that refers to inverse attention and pixel similarity learning according to the present invention and FCN-8s, DilaT1edNeT1, DilaT1edNeT1Cascade, HolisT1ic, PSPNeT1, DeepLabv2-ASPP, DeepLab -LFOV for comparison, the experimental results are shown in Table 4:

[0092]

[0093] Table 4 Comparison results of various semantic segmentation methods on the MIT1ADE20K test set

[0094] It can be seen from Table 4 that since the MIT1ADE20K data set is relatively large and the data set contains samples in various complex scenarios, it is difficult to perform semantic segmentation on the data set. From the comparison results, it can be seen that a reference inverse attention method described in the present invention Compared with the existing best method PSPNeT1, the image semantic segmentation method based on force and pixel similarity learning has improved by about 1%.

[0095] An image...

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 present invention provides an image semantic segmentation method adopting inverse attention learning and pixel similarity learning. The method comprises the following steps that: step A, preliminary semantic segmentation is performed on input images, and the branch networks of different scales in a DeepLab v2 ResneT1 101 network are adopted to extract the features of the input images of different scales; step B, on the basis of the step A, an inverse attention layer is adopted to segment the boundaries of the input images; step C, on the basis of the step A, a pixel similarity learning layer is adopted to further segment the boundaries of the input images; step D, the inverse attention layer and the pixel similarity learning layer are optimized, and a corresponding loss function is defined; and step E, network parameters are trained. According to the image semantic segmentation method of the present invention, the inverse attention mechanism is adopted to correct boundary locatingbetween a target area and a background area; the pixel similarity learning mechanism is adopted to solve the problems of boundary locating ambiguity and boundary smoothing between the target area andthe background area; and therefore, the effective segmentation of a fused area between the target area and the background area can be realized.

Description

technical field [0001] The invention relates to the field of computer vision, in particular to an image semantic segmentation method using inverse attention and pixel similarity learning. Background technique [0002] Image semantic segmentation technology is one of the hot and difficult issues in the field of computer vision. This technology divides the image into several different semantic regions, and recognizes the category of each region to obtain the final semantic annotation. Image semantic segmentation technology is automatically Driving, positioning and navigation based on image semantics, medical image analysis and other fields have a wide range of application values. [0003] According to the different semantic feature extraction, image semantic segmentation is usually divided into two categories: traditional and non-traditional methods; traditional image semantic segmentation methods include threshold-based image semantic segmentation, region-based image semantic...

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): G06T7/12G06T7/194
CPCG06T7/12G06T7/194G06T2207/20081G06T2207/20084
Inventor 肖立智李涛赵雪专裴利沈李冬梅朱晓珺曲豪张栋梁汪伟邹香玲郭航宇
Owner 盐城禅图智能科技有限公司
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