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

Distributed clustering method of visual objects based on time series

A distributed clustering and time series technology, applied in the field of image recognition, can solve problems such as low efficiency and complex clustering methods, achieve high efficiency, improve clustering efficiency, and reduce time spent

Active Publication Date: 2019-01-18
UNIV OF ELECTRONICS SCI & TECH OF CHINA
View PDF11 Cites 3 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] In order to solve technical problems such as complex clustering methods and low efficiency, the present invention provides a distributed clustering method for visual objects based on time series

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
  • Distributed clustering method of visual objects based on time series

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0032] DBSCAN (density-based spatial clustering of applications with noise) is a more classic density-based clustering algorithm. The input parameters do not need to include the final number of categories, which is in line with our current scene. Its specific definition is as follows: X is a set of points that need to be clustered, and there are three types of points, core points, density-reachable points, and outlier points. If a point p has minPts points (including point p itself) within the range of distance ε, then p is called the core point, and the points within the range of ε become directly reachable by p. Second, if there exists a path p 1 ,...,p n , by p 1 = p and p n =q, and each p i+1 all by p i directly reachable, then q is said to be reachable by p. Finally, all points that are not reachable by any point are called outliers.

[0033] However, only relying on the above-mentioned DBSCAN cannot complete realistic tasks, because the time complexity of the abov...

Embodiment 2

[0045] The time-series-based distributed clustering method for visual objects proposed in the above-mentioned embodiment 1 is used for the actual algorithm coding test, and the specific process is as follows:

[0046] Step S1, collect pictures of the visual target through the camera. The video information can be obtained by connecting with the camera through the RTSP protocol, and the target detection program can be written through the recognition SDK developed by the laboratory, and the picture information and the time information of the picture can be saved to the server hard disk.

[0047] Step S2, through the feature extraction interface of the recognition SDK in the laboratory, feature extraction can be performed on the picture saved in the previous step. After actual testing, it takes about 3 hours to get all the feature data on 4 1080Ti Nvidia graphics cards for 1 million visual target images. The feature data can be stored in the file in binary form. The feature dimen...

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 distributed clustering method of visual objects based on time series, which comprises the following steps of: data acquiring step: acquiring a large number of pictures of visual objects by using a camera; feature extraction step: carrying out feature extraction and time sorting on the collected pictures; distributed clustering computing: collecting the time series characteristics of visual objects by the camera, and segmenting the data, and calculating the categories by multiple nodes. the invention can quickly complete the first iteration of the algorithm according to the time continuity of the visual object picture under the scene gathered by the camera, greatly reducing the time required for clustering, and greatly improving the clustering efficiency. The invention combines the distributed multi-computer implementation, further speeds up the clustering process, can realize the data processing of million level or million level, the algorithm is simple and the efficiency is high.

Description

technical field [0001] The invention relates to the technical field of image recognition, in particular to a distributed clustering method for visual objects based on time series. Background technique [0002] With the development of artificial intelligence deep learning technology, various new model training methods have been proposed. Especially in the field of image recognition, the accuracy of recognition has been continuously improved, and in some cases it can even exceed the accuracy of human naked eye recognition. However, the training of the model requires a large amount of labeled image data, and some data can be downloaded through the Internet. But these data are limited. In more cases, the pictures are obtained by collecting the visual objects in real life through the camera. The problem is that these pictures collected by the camera are unlabeled data. First, manual classification is required, and then the It serves as the data source for training the model. Th...

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): G06K9/62
CPCG06F18/23
Inventor 段翰聪刘长红孙月闵革勇
Owner UNIV OF ELECTRONICS SCI & TECH OF CHINA
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