Zeroshot hash image retrieval method based on supervision transfer

A picture and hashing technology, applied in the field of image hashing, can solve the problems that the category relevance cannot be reflected, and the category cannot be encoded reliably, so as to improve the retrieval accuracy, reduce the quantization error, and have a wide range of applications

Inactive Publication Date: 2017-11-14
UNIV OF ELECTRONICS SCI & TECH OF CHINA
View PDF3 Cites 14 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Doing so will cause the correlation between categories to not be reflected, so that the trained hash model can only effectively encode the categories in the training set, but cannot reliably encode a category that has never been seen. i.e. cannot reliably retrieve unseen categories

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
  • Zeroshot hash image retrieval method based on supervision transfer
  • Zeroshot hash image retrieval method based on supervision transfer
  • Zeroshot hash image retrieval method based on supervision transfer

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0051] In order to make the purpose, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the embodiments.

[0052] Picture retrieval of the present invention comprises the following steps:

[0053] Step 1: Carry out the preprocessing of picture and picture label to training sample, obtain the picture characteristic vector X of training sample, label vector Y:

[0054] Using the material model given in the document "R.Socher, M.Ganjoo, C.D.Manning, and A.Ng. Zero-shot learning through cross-modal transfer.In NIPS, 2013.", using the free corpus on Wekipedia (including nearly 500 million Vocabulary) for training, dig out a reasonable vocabulary representation method, and express the label as a label vector Y through a pre-trained model.

[0055] When extracting the original features of the picture, the excitation of the convolutional neural network can be used as the feature of th...

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 zeroshot hash image retrieval method based on supervision transfer, belonging to the technical field of image hash. Modeling is performed on a label with training data by using an existing natural language processing model, and a label space is formed; by means of a relationship between potential storage labels in the new label space, mapping from an image feature space to the label space is trained; on the basis, the mapping relationship is reflected on a hash code. Strict requirements on training samples are avoided, the application scope is wide, and especially in a large database, when categories are more and training samples cannot be found for each category, the image retrieval accuracy rate of the categories can be greatly improved by the method.

Description

technical field [0001] The invention belongs to the field of image hashing, and in particular relates to a method for accurately hashing pictures, especially a method for reasonably coding a certain type of picture when there is no training sample. Background technique [0002] With the generation of more and more multimedia data, hashing has become a powerful tool for large-scale retrieval, which can greatly shorten the time it takes to find billions of data. Since computers are very good at XOR operations, using hash codes for retrieval can meet the search needs brought about by the era of big data. [0003] Database hashing is an extremely important job, which is of great significance to many fields. Therefore, hashing has received important attention for a long time in the past, and many important algorithms have been proposed. Such as Local Sensitive Hashing (LSH) for data-independent hashing, Iterative Quantization (ITQ) for data-dependent hashing, Isotropic Hashing,...

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): G06F17/30G06K9/62
CPCG06F16/5866G06F18/2411
Inventor 杨阳罗雅丹陈纬伦沈复民邵杰申恒涛
Owner UNIV OF ELECTRONICS SCI & TECH OF CHINA
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