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

A human clothing segmentation method based on semantic consistency

A technology of semantic segmentation and consistency, applied in the field of computer vision, can solve problems such as failure to extract reasonable features, insufficient data volume and limited deep learning effect, etc., achieve good application value, improve final effect, improve accuracy and efficiency

Active Publication Date: 2018-12-11
ZHEJIANG UNIV
View PDF8 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Deep learning can effectively solve the problem of feature representation, but in the face of the lack of sufficient and accurately labeled data sets in clothing segmentation, the lack of data limits the effect of deep learning, and the deformable characteristics of clothing make ordinary Convolution cannot extract reasonable features

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
  • A human clothing segmentation method based on semantic consistency
  • A human clothing segmentation method based on semantic consistency
  • A human clothing segmentation method based on semantic consistency

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0077] The implementation method of this embodiment is as described above, and the specific steps will not be described in detail. The following only shows the effect of the case data. The present invention is implemented on three data sets with ground-truth labels, namely:

[0078] Fashionista v0.2 dataset: This dataset contains 685 images with 56 categories of semantic labels.

[0079] Refined Fashionista dataset: This dataset contains 685 images with 25 categories of semantic labels.

[0080] CFPD dataset: This dataset contains 2682 images with 23 categories of semantic labels.

[0081] In this example, a picture is selected for each data set to conduct an experiment. First, the closest picture is obtained by calculating the similarity, and then the features of the two pictures are extracted respectively, and the adjacent pairs of this group of pictures in the flow space are compared. Relationships are jointly modeled to obtain the final semantic segmentation graph, such ...

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 human body clothing segmentation method based on semantic consistency, which is used for analyzing the clothing area semantic situation of each frame under the condition thata single-person clothing picture of a single frame is given. The method comprises the following steps: acquiring an image data set for training human body clothing segmentation and defining an algorithm target; searching for adjacent pictures in semantic space for each single frame image in the data set and forming picture pairs; jointly modeling the adjacent relations of each pair of images in the flow pattern space; establishing a prediction model of laundry segmentation; using the prediction model to analyze the semantic information of the clothes in the picture. The method is suitable forclothing segmentation analysis in real images, and has a good effect and robustness in various complex situations.

Description

technical field [0001] The invention relates to the field of computer vision, in particular to a method for segmenting human clothing based on semantic consistency. Background technique [0002] As a low-level visual technology, clothing semantic segmentation is often used as auxiliary information for some high-level visual tasks, such as clothing retrieval and clothing attribute analysis. The goal of clothing segmentation is to give an image and predict the classification label of each pixel in the image. The key factors for clothing segmentation mainly include the large apparent differences within clothing categories, the non-rigidity of clothing, and the extreme deformability of clothing. Traditional methods generally regard the task of clothing segmentation as a semantic segmentation problem. Although some methods have made breakthroughs in classification accuracy, they do not make full use of the information of existing data. [0003] Due to the effectiveness of stati...

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
IPC IPC(8): G06K9/34G06K9/00G06N3/04
CPCG06V40/10G06V10/267G06N3/045
Inventor 李玺吉炜
Owner ZHEJIANG 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