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

A video copy detection system and method based on convolutional and recurrent neural network

A cyclic neural network and video copy detection technology, applied in neural learning methods, biological neural network models, neural architectures, etc., can solve the problems of time-consuming, memory-consuming, and low accuracy of detection results.

Inactive Publication Date: 2018-12-11
SOUTHEAST UNIV
View PDF3 Cites 9 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0010] Purpose of the invention: In order to overcome the deficiencies of the prior art, the present invention provides a video copy detection system and method based on convolutional and cyclic neural networks, which can solve the problem of low accuracy of detection results in detection, time-consuming and memory-consuming video matching The problem

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 video copy detection system and method based on convolutional and recurrent neural network
  • A video copy detection system and method based on convolutional and recurrent neural network
  • A video copy detection system and method based on convolutional and recurrent neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0074] Such as figure 1 As shown, the present invention provides a kind of video duplication detection system, and system comprises 5 modules, is respectively data set establishment module 1, frame feature extraction module 2, spatio-temporal feature training module 3, cycle network test module 4 and copy video matching module 5 , wherein the spatio-temporal feature training module 3 also includes a video editing module 31 and a recurrent network training module 32, the data set building module 1 mainly collects relevant data of video copy detection, and uses the public video copy detection data set CC_WEB as the training data of the recurrent neural network Set, using the public data set VCDB to verify the performance of the method proposed in the present invention, as a test data set.

[0075] The frame feature extraction module 2 is used to extract the image frame features in the CC_WEB video using the 50-layer residual convolutional neural network ResNet50. The residual co...

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 video copy detection system based on a convolutional and recurrent neural network. The system includes five modules, which are a data set establishment module, a frame feature extraction module, a spatio-temporal feature training module, a loop network test module and a copy video matching module. The spatio-temporal feature training module also includes video editing module and loop network training module. The invention adopts a residual convolutional neural network to extract deeper frame-level feature representationto improve the detection accuracy and reduce thedetection recall effectively. A twin-loop neural network is used to fuse multiple frame-level features, and the spatio-temporal feature representation is generated by using the dynamic information between frames, so the spatio-temporal fusion between sequences is realized, and video matching takes less time and occupies less memory.

Description

technical field [0001] The invention relates to a video copy detection system and method, in particular to a video copy detection system and method based on convolutional and cyclic neural networks. Background technique [0002] With the development of network multimedia technology, network video data is increasing massively, and a large amount of video data is made public on the Internet. Internet users can search for different types of videos such as politics, entertainment, sports, etc. on Youtube or MetaCafe. Although online video allows Internet users to obtain the latest information from all over the world, there are also some potential risks. Pirates can easily steal or tamper with online original videos to earn illegal income. Therefore, copy detection technology based on video analysis is of great significance to network security and copyright protection. [0003] In early video copyright protection tasks, watermarks or digital tags were inserted into video strea...

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): G06K9/00G06N3/04G06N3/08
CPCG06N3/084G06V20/46G06V20/41G06N3/045
Inventor 路小波胡耀聪
Owner SOUTHEAST 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

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