A video anomaly detection system and method based on weak supervised learning

An abnormal event and detection system technology, applied in the field of video abnormal event detection system, can solve problems such as difficult to build a unified model for normal behavior, save manual labeling costs and labor time, save human labor and time costs, and improve accuracy rate effect

Active Publication Date: 2019-03-22
深圳龙岗智能视听研究院
View PDF5 Cites 14 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Due to the various patterns of normal behavior events in daily life, coupled with the changes in behavior expressions caused by different video shooting scenes and shooting angles, it is difficult to build a unified model for all normal behaviors

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 anomaly detection system and method based on weak supervised learning
  • A video anomaly detection system and method based on weak supervised learning
  • A video anomaly detection system and method based on weak supervised learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0026] figure 2 For the network structure diagram of the proposed model of the present invention, as figure 2 As shown, the system of this embodiment includes: a feature extraction deep network model 5, a normal / abnormal event classification model 6, a hidden layer 7, a hidden layer 2 8, and a hidden layer 3 9.

[0027] image 3 Divide the structure diagram for the video clip hierarchy, such as image 3 As shown, this embodiment includes: zero-level division 11 of video segments, first-level division 12 of video segments, second-level division 13 of video segments, and three-level division 14 of video segments.

[0028] figure 1 It is a flowchart of the present invention, wherein s1-s3 correspond to specific implementation steps 1)-3) in turn. A video anomaly event detection method based on weakly supervised learning, the overall operation process is described as follows:

[0029] 1) Divide the input video into segments, construct behavioral package S1: Given a segment ...

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 abnormal event detection system based on weak supervised learning and a method thereof. The method is based on a depth learning framework and represents the problem ofweak supervised video abnormal event detection as a multi-instance learning model. For a video sequence, it is divided into several behavior instances, and the depth network model is used to extract the multi-level shape for each behavior instance. In order to realize the task of abnormal event detection in a given video, a motion joint feature representation and a normal/abnormal behavior classifier are constructed to score the behavior instances. The method of the invention only needs weakly labeled samples to construct a model, thereby saving a large amount of human labor and time cost, andhaving higher detection accuracy for common abnormal events in daily life. In the currently published test data set, a leading level of detection has been achieved.

Description

technical field [0001] The present invention relates to the technical field of video behavior analysis, in particular to a video abnormal event detection system and method based on weakly supervised learning. The method adopts a deep learning framework and designs a weakly supervised learning strategy to train normal / abnormal video behavior classification On this basis, the detection of abnormal video behavior events is completed. Background technique [0002] Video behavior anomaly event detection has been a research hotspot in the field of computer vision for a long time. With the increasing popularity of high-definition surveillance cameras in cities in our country, the resulting massive surveillance videos have brought a heavy workload to video operators. At the same time, the existing video behavior detection technology cannot detect abnormal events (such as violent terrorist crimes) in time, and then remind the staff to prevent the further development of the situation...

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): G06K9/00G06K9/62G06N3/04
CPCG06V20/44G06V20/41G06N3/045G06F18/24Y02T10/40
Inventor 安欣赏李楠楠张世雄张子尧李革张伟民
Owner 深圳龙岗智能视听研究院
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