Video human body tumble detection method and system based on track weighted depth convolution sequence pooling descriptor

A technology of deep convolution and detection methods, applied in neural learning methods, image enhancement, instruments, etc., can solve problems such as unfavorable video spatiotemporal feature encoding, and achieve the effect of reducing redundancy

Active Publication Date: 2019-04-16
SHANDONG UNIV
View PDF4 Cites 26 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Second, videos usually contain a lot of redundant information, which is very unfavorable for video spatio-temporal feature encoding

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 human body tumble detection method and system based on track weighted depth convolution sequence pooling descriptor
  • Video human body tumble detection method and system based on track weighted depth convolution sequence pooling descriptor
  • Video human body tumble detection method and system based on track weighted depth convolution sequence pooling descriptor

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0059] In one or more implementations, a video human fall detection method based on trajectory weighted depth convolution order pooling descriptor is disclosed, such as figure 1 As shown, it mainly includes the following steps:

[0060] (1) All frames of the collected RGB video are input into the VGG-16 convolutional network to calculate the convolutional feature maps, and then these convolutional feature maps are normalized using the method of space-time normalization;

[0061] (2) Calculate improved dense trajectories based on the collected RGB video, and these trajectories can describe the trajectories of moving figures in the video. According to these improved dense trajectories, the trajectory attention map is calculated, and the trajectory attention map can help locate the character area in the video;

[0062] (3) Weight the trajectory attention map of each frame to the corresponding convolution feature map to obtain the trajectory weighted convolution features of the c...

Embodiment 2

[0134] A video human body fall detection system based on trajectory weighted depth convolution order pooling descriptors disclosed in one or more embodiments includes a server, the server includes a memory, a processor, and is stored on the memory and can be processed A computer program running on a processor, and when the processor executes the program, the method for detecting human falls in video based on trajectory-weighted depthwise convolution order pooling descriptors described in Embodiment 1 is realized.

Embodiment 3

[0136] A computer-readable storage medium disclosed in one or more implementations, on which a computer program is stored, and when the program is executed by a processor, the sequential pooling based on trajectory weighted depth convolution as described in Embodiment 1 is performed. Descriptor-based video human fall detection method.

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 human body tumble detection method and system based on a track weighted depth convolution sequence pooling descriptor. The method comprises the steps of obtaining a convolution feature map of each frame; providing a new track attention map and can be used for positioning a character area in a video; the convolutional feature map of the video frame and the track attention map are weighted to obtain track weighted convolutional features, and the features can effectively describe visual features of a character area in the video; and a clustering pooling method is provided to eliminate redundancy in the sequence; and finally, the track weighted convolution feature sequence is encoded by using a sequential pooling method, and the obtained result is the track weighted depth convolution sequential pooling descriptor. By using the descriptor, the current highest accuracy is obtained on the SDUFall data set, and a good effect is also obtained on the UR data set and the multi-view data set.

Description

technical field [0001] The invention belongs to the technical field of human body fall detection, and in particular relates to a video human body fall detection method and system based on trajectory weighted depth convolution order pooling descriptors. Background technique [0002] Worldwide, the population over the age of 60 is growing at a much faster rate than other age groups. From 2006 to 2050, the number is expected to increase from 6.88 million to 2 billion. In China, people over the age of 65 accounted for about 8.87% of the total population in 2010, but by 2050, the number of people over the age of 65 is expected to increase to 30%. As described in the WHO report, falls are a serious problem among older adults. About 28-35% of people over the age of 65 fall each year. About 32-42% of people over the age of 70 will fall every year. Falls are the leading cause of death for people over the age of 79. The elderly generally live alone, so if a fall occurs and lacks ti...

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): G06K9/00G06K9/62G06N3/04G06N3/08G06T7/269
CPCG06N3/08G06T7/269G06T2207/30196G06T2207/10016G06V40/23G06V20/40G06N3/045G06F18/23
Inventor 马昕张智勐宋锐荣学文田新诚田国会李贻斌
Owner SHANDONG 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
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