Unlock instant, AI-driven research and patent intelligence for your innovation.

Method and device for retrieving footwear

A footwear and shoe body technology, applied in the field of footwear retrieval, can solve problems such as low accuracy of footwear retrieval and complex image background, and achieve the effect of improving accuracy, solving complex background and multi-angle

Active Publication Date: 2020-11-06
SENSETIME GRP LTD
View PDF10 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] The invention provides a footwear retrieval method, which is used to solve the problem of low retrieval accuracy of footwear caused by complex image backgrounds and various shoe body angles during the footwear retrieval process

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
  • Method and device for retrieving footwear
  • Method and device for retrieving footwear
  • Method and device for retrieving footwear

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0038] see figure 1 , an embodiment of the footwear retrieval method in the embodiment of the present invention includes:

[0039] 101. Crop the input image according to the edge coordinate detection model;

[0040] The shoe retrieval device cuts the input image according to the edge coordinate detection model to obtain the shoe body image, and the input image is the shoe image to be retrieved.

[0041] Specifically, the footwear retrieval device may use the edge coordinate detection model to locate the shoe body coordinates of the input image to obtain the shoe body coordinates of the input image; then cut the input image according to the shoe body coordinates to obtain the shoe body image.

[0042] The edge coordinate detection model in the embodiment of the present invention is obtained through model training; the model training specifically includes: coordinate sample images (that is, marked footwear images), and a large number of original footwear images (unmarked shoe ...

Embodiment 2

[0060] The following embodiments describe in detail the footwear retrieval method in the embodiments of the present invention, including:

[0061] The training of the edge coordinate detection model:

[0062] Taking coordinate sample images (ie, marked shoe images) and a large number of original shoe images (unmarked shoe images) as input, use Faster R-CNN to train the convolutional neural network, and constantly update iterations Coordinate model of shoe body. Wherein, the coordinate sample image is a footwear image marked with edge coordinate positions, and the marked footwear image refers to a footwear image marked with shoe body coordinates, and the shoe body coordinates may be manually marked The coordinates of the bounding box, such as image 3 The coordinates of two points in the upper left corner and the lower right corner of the shown rectangular frame; the convolutional neural network may be a neural network with five convolutional layers and two fully connected la...

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 provides a footwear retrieval method, which is used to solve the problem of low accuracy of footwear retrieval caused by complex image background and various shoe body angles in the footwear retrieval process; the method includes: according to the edge coordinate detection model, the input image is Perform cropping to obtain the shoe body image; use a key point detection model to perform key point detection on the shoe body image to obtain key points of the shoe body image; perform image adjustment according to the key points of the shoe body image, so that the The shoe body orientation and image size of the shoe body image are consistent with the comparison image in the database; the depth feature extraction is carried out on the shoe body image after the image adjustment by using the feature extraction model, and the depth features of the shoe body image are obtained; The depth feature of the shoe body image is compared with the depth feature of the comparison image in the database, and the retrieval result of the input image is output according to the result of the similarity comparison.

Description

technical field [0001] The invention relates to the field of data processing, in particular to a footwear retrieval method. Background technique [0002] With the rapid popularization of the Internet and the rise and development of e-commerce, image retrieval technology based on computer vision has achieved unprecedented development. General e-commerce websites distinguish and classify pictures by manually adding labels to product pictures, but cannot retrieve and give feedback on unlabeled product pictures provided by users. In response to the above problems, a large number of retrieval methods based on the visual information of commodity pictures have been continuously proposed. At present, most of the popular methods are based on convolutional neural networks to extract deep features of pictures, calculate the similarity measure with the deep features of pictures in the database, and Perform ascending sorting, and return the first few sorted results to the user, which ar...

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 Patents(China)
IPC IPC(8): G06K9/62G06K9/34G06K9/46G06K9/42G06F16/58G06Q30/06
CPCG06F16/583G06Q30/0601G06V10/32G06V10/267G06V10/40G06V10/44G06F18/22
Inventor 陈海峰邱石张伟
Owner SENSETIME GRP LTD