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

Video target detection method based on motion history image

A motion history image and target detection technology, applied in the field of video target detection based on motion history images, achieves the effects of fast speed, simple extraction, guaranteed detection speed and detection accuracy

Inactive Publication Date: 2020-01-17
BEIJING TECHNOLOGY AND BUSINESS UNIVERSITY
View PDF8 Cites 9 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0010] Aiming at the defect that the existing image target detection technology cannot make full use of the timing information in the video, this invention proposes a video target detection method based on motion history images. In the algorithm, the motion history images are used to represent the timing information in the video, and the timing information is fused To the image target detection technology, so as to better detect the video

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
  • Video target detection method based on motion history image
  • Video target detection method based on motion history image
  • Video target detection method based on motion history image

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0034] (1) The present invention uses a large-scale video target detection benchmark data set ImageNet VID proposed by the Large-Scale Visual Recognition Challenge in 2015 as an experimental data set, which contains 30 categories of targets. A subset of categories in the DET image dataset that takes into account different factors such as type of motion, video clutter, average number of object instances, and a few others can be extensively studied. Meanwhile, the dataset contains 3862 videos as training set, 555 videos as validation set, and 937 videos as test set. The training set and verification set have been fully labeled and all video clips have been cut into frames, that is, the data set is a sequence of video frames. The method in the present invention is not only applicable to the detection of vehicles and animals included in the data set, but also can be extended to other types of video target detection, such as pedestrian detection.

[0035] (2) The video frame can b...

Embodiment 2

[0053] The difference with embodiment 1 is:

[0054] The motion history image obtained in step 4 can first be processed with pseudo-color, that is, different colors are given according to the gray level of the pixels in the obtained gray-scale image, so that the motion history image can provide more information for model training. In the present invention, the grayscale color conversion method is adopted to convert the grayscale image into an RGB image, and the conversion method is as follows:

[0055] (1) Obtain the value f(x, y) of a certain pixel point (x, y) in the image;

[0056] (2) Obtain the value R(x, y) of the red channel, the value G(x, y) of the green channel, and the value B(x, y) of the blue channel of the pixel according to the following conversion formula.

[0057]

[0058]

[0059]

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 target detection method based on a motion history image, and aims to improve the speed and accuracy of video target detection. The video target detection method comprises three aspects: (1) for an input video frame sequence, calculating motion history images of the video frame sequence, and performing feature extraction on video frames and the motion history imagesthereof through a residual network; (2) fusing the extracted two parts of features, and inputting the fused features into a convolutional neural network to extract candidate boxes; and (3) obtaining avideo target detection result according to a bounding box regression algorithm and the constructed classifier. According to the video target detection method, the motion history images are added intothe model training process, so that the feature information of the video frames is provided for the model, and the association information between the video frame sequences is increased, and the accuracy of video target detection can be improved.

Description

technical field [0001] The invention relates to the field of computer vision, in particular to the field of video target detection in target detection, and specifically designs a video target detection method based on motion history images. Background technique [0002] At present, society is in an era of big data and cloud computing. With the emergence of video social software such as Douyin, Kuaishou, and Volcano Video, information on the Internet is no longer limited to text and images, and video information is also emerging in an endless stream. It is an unavoidable problem in contemporary society to mine the connection in the video information and to monitor the video information effectively. [0003] The purpose of object detection is to detect and classify multiple objects of interest in a picture or video. According to the detection object, it can be divided into image target detection and video target detection. Currently, image target detection has matured in the...

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
CPCG06N3/08G06V20/41G06V20/46G06N3/045
Inventor 李韩玉蔡强余乐李海生颜津
Owner BEIJING TECHNOLOGY AND BUSINESS UNIVERSITY
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