Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Integration target attribute identification and precise retrieval method based on depth measurement learning

A technology of metric learning and target attributes, which is applied in the field of integrated target attribute recognition and precise retrieval, can solve the problems that the precise retrieval of pedestrians and vehicles has not been fully studied, and achieve the effect of improving accuracy

Active Publication Date: 2016-07-27
PEKING UNIV
View PDF3 Cites 60 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

In real-world scenarios, generally only accurate face retrieval (or face recognition) can achieve sufficient practical accuracy, and the same or even more important problem of precise pedestrian and vehicle retrieval has not been fully studied, so it is urgent to propose new methods. An algorithm model that can efficiently and accurately solve various individual target retrieval problems

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
  • Integration target attribute identification and precise retrieval method based on depth measurement learning
  • Integration target attribute identification and precise retrieval method based on depth measurement learning
  • Integration target attribute identification and precise retrieval method based on depth measurement learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0106] Embodiment 1 (vehicle precise retrieval)

[0107] This embodiment is based on an outdoor monitoring scenario. In the outdoor environment, apart from the differences in the appearance of different vehicles, illumination changes, shadows, and trees blown by the wind will bring different degrees of interference to the precise retrieval of vehicles. This embodiment focuses on how to apply the present invention to vehicle precision search on the issue.

[0108] It should be noted here that this problem cannot be solved by license plate recognition technology, because vehicle retrieval technology is designed to search for vehicles by appearance in special cases where license plate recognition cannot work; in addition, vehicle retrieval problems are different from pedestrian or face retrieval problems , in theory, pedestrians / faces with different identities must have different appearances, but if vehicles with different IDs have the same color and model, their appearances may...

Embodiment 2

[0112] Embodiment 2 (precise retrieval of pedestrians)

[0113] Unlike vehicles, pedestrians have more attributes, but not all attributes have the same effect. Here, three attribute characteristics are used: gender, coat color, pants (or skirt) color, and then add the identity ID of the pedestrian itself to To train the network model, one attribute or more attributes can be used in actual use.

[0114] Deep neural networks are still used for feature extraction, using hybrid distance metric networks for both category and individual feature extraction. Taking the network shown before as an example, after the 1024-dimensional feature Fc7 obtained by extracting the above branch, a gender classifier, a coat color classifier and a pants (or skirt) color classifier are connected to extract pedestrians that can be identified. The characteristics of gender and upper and lower body color, the following branch is connected to the clustering loss function after extracting the 1024-Witt F...

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 embodiment of the invention provides an integration target attribute identification and precise retrieval method based on depth measurement learning. The method comprises the following steps: obtaining a query picture and a picture database; according to a pre-trained deep neural network model, independently obtaining the feature vector of the query picture and the feature vectors of all pictures in the picture database, wherein the feature vectors comprise the category features of an individual target and the identity features of the individual target in the pictures; according to the feature vectors, independently calculating an Euclidean distance between the query picture and each picture in the picture database in an Euclidean space; and according to the Euclidean distances between the query picture and all pictures in the picture database, selecting a picture which exhibits highest similarity with the query picture in the picture database. The method can improve the accuracy of image retrieval.

Description

technical field [0001] The invention relates to the field of image retrieval, in particular to an integrated target attribute recognition and precise retrieval method based on deep metric learning. Background technique [0002] Accurate target retrieval and individual target attribute recognition have always been two crucial issues in the field of computer vision, and are the basis for upper-level application analysis such as individual target tracking and behavior analysis. Currently, in academia, the two belong to two independent issues. . [0003] The main goal of the former is to identify a specific individual target from a series of individual target pictures taken from different cameras. Most of the existing research focuses on the retrieval of specific types of individual targets such as faces or pedestrians. In real-world scenarios, generally only accurate face retrieval (or face recognition) can achieve sufficient practical accuracy, and the same or even more impor...

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): G06F17/30
CPCG06F16/583
Inventor 田永鸿刘弘也王耀威黄铁军
Owner PEKING UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com