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

Single-stage semi-supervised image human body target detection method

A human target and detection method technology, applied in the field of computer vision, can solve cumbersome problems

Pending Publication Date: 2020-07-03
SOUTH CHINA UNIV OF TECH
View PDF2 Cites 8 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0007] The purpose of the present invention is to overcome the cumbersome problem of the existing semi-supervised training process, and proposes a single-stage semi-supervised image human target detection method, which can obtain good detection results with only one round of training, effectively saving plenty of time

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
  • Single-stage semi-supervised image human body target detection method
  • Single-stage semi-supervised image human body target detection method
  • Single-stage semi-supervised image human body target detection method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0063] The present invention will be further described below in conjunction with specific examples.

[0064] Such as Figure 1 to Figure 3 As shown, the single-stage semi-supervised image human target detection method provided by this embodiment includes the following steps:

[0065] S1. Divide the video frame data into a collection of real label images No ground truth label image collection and the test data set details as follows:

[0066] The image of the video frame needs to be zoomed in order to achieve the ideal training effect and reduce the amount of data calculation; classify the video frame data according to the needs, and first divide the video frame data into training data and test data sets Two categories; then the training data is divided into two categories: the real label image set and the set of untrue labeled images The ratio of 1:19, that is, the training data is equal to A ground-truth label image is denoted as which is A label-free image ...

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 single-stage semi-supervised image human body target detection method, and the method comprises the steps: selecting a small number of images with real labels from video frames, and enabling the remaining other video frames to serve as images without real labels; the two types of images are simultaneously sent into a deep network to train the network; wherein the trainingprocess is different, the image with the real label can be normally trained, but the image without the real label cannot be normally trained, so that high-confidence position information can be obtained through the network to serve as a temporary label of the image without the real label, and then normal training is carried out; in order to prevent the network from being deviated by the image ofthe temporary label, limitation is carried out through subsequent screening and weight setting; and until the network model is trained to a preset number of times. According to the method, two types of images are trained at the same time, only one stage is needed, and a large amount of time cost is saved.

Description

technical field [0001] The invention relates to the technical field of computer vision, in particular to a single-stage semi-supervised image human target detection method. Background technique [0002] Pedestrian detection is the use of computer vision technology to identify whether there are pedestrians in images or video frames and give precise positioning. This technology has a wide range of applications and can be combined with pedestrian tracking, pedestrian re-identification and other technologies, and can be well applied to artificial intelligence systems, vehicle assisted driving systems, intelligent video surveillance, human behavior analysis, intelligent transportation and other real-world scenarios. [0003] Due to some unique characteristics of pedestrians, the appearance is easily affected by clothing color, scale, occlusion, posture and viewing angle, etc., making pedestrian detection a hot research topic that is not only valuable but also challenging in the f...

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): G06K9/00G06N3/04G06N3/08
CPCG06N3/08G06V40/20G06V20/41G06V2201/07G06N3/045
Inventor 陈学贤吴斯
Owner SOUTH CHINA UNIV OF TECH
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