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

Examination illegal behavior detection method based on 3D convolution

A detection method and convolution technology, applied in the direction of image data processing, image enhancement, instruments, etc., can solve the problems of supervision and analysis of all examinees' behavior, and achieve the effect of reducing human resource consumption, improving capture rate, and reducing time cost.

Pending Publication Date: 2021-12-28
四川天翼网络股份有限公司
View PDF2 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Although the technical solution also obtains the supervision data of the examination video, it does not directly supervise and analyze the behavior of all candidates. Therefore, the technical scheme needs to be improved in the recognition accuracy of examination violations

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
  • Examination illegal behavior detection method based on 3D convolution

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0025] In this example, if figure 1 As shown, a 3D convolution-based examination violation detection method includes the following steps:

[0026] Step 1: Obtain the real-time monitoring video of the exam, and use the adaptive video lens segmentation strategy to divide the video into multiple video clips to form a video sequence, and capture the short-term video sequence in the video sequence according to the motion change information between adjacent frames in the video clip dependencies.

[0027] Step 2: Construct a space-time pyramid pooling network, and extract the inter-temporal dependencies in video clips through the constructed space-time pyramid pooling network.

[0028] Step 3: Input the video sequence into the spatial stream network and the local multi-region network for feature extraction, and extract the global spatial features and local region features in the video sequence respectively.

[0029] Step 4: Fuse the global spatial features and local area features a...

Embodiment 2

[0036] In the present embodiment, a kind of examination violation behavior detection method based on 3D convolution, comprises the following steps:

[0037]Step 1: Obtain the real-time monitoring video of the exam, and use the adaptive video lens segmentation strategy to divide the video into multiple video clips to form a video sequence, and capture the short-term video sequence in the video sequence according to the motion change information between adjacent frames in the video clip dependencies.

[0038] Step 2: Construct a space-time pyramid pooling network, and extract the inter-temporal dependencies in video clips through the constructed space-time pyramid pooling network.

[0039] Step 3: Input the video sequence into the spatial stream network and the local multi-region network for feature extraction, and extract the global spatial features and local region features in the video sequence respectively.

[0040] Step 4: Fuse the global spatial features and local area fe...

Embodiment 3

[0046] In the present embodiment, a kind of examination violation behavior detection method based on 3D convolution, comprises the following steps:

[0047] Step 1: Obtain the real-time monitoring video of the exam, and use the adaptive video lens segmentation strategy to divide the video into multiple video clips to form a video sequence, and capture the short-term video sequence in the video sequence according to the motion change information between adjacent frames in the video clip dependencies.

[0048] Step 2: Construct a space-time pyramid pooling network, and extract the inter-temporal dependencies in video clips through the constructed space-time pyramid pooling network.

[0049] Step 3: Input the video sequence into the spatial stream network and the local multi-region network for feature extraction, and extract the global spatial features and local region features in the video sequence respectively.

[0050] Step 4: Fuse the global spatial features and local area f...

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 an examination illegal behavior detection method based on 3D convolution, and the method comprises the following steps: segmenting an examination real-time monitoring video into a plurality of video segments to form a video sequence, and capturing a short-time dependency relationship in the video sequence; extracting a medium-time dependency relationship in the video clip through the constructed space-time pyramid pooling network; respectively inputting the video sequence into a spatial stream network and a local multi-area network for feature extraction, and respectively extracting global spatial features and local area features; fusing the global spatial features and the local region features and testing a feature fusion result to obtain a global feature vector, inputting the global feature vector into softmax for classification, and obtaining a classification result of student actions. According to the invention, the manpower resource consumption and the time cost required by the playback of the examination monitoring video can be greatly reduced, the continuous illegal behaviors of the examinee in the examination room can be analyzed, and the capturing efficiency of the illegal behaviors is improved.

Description

technical field [0001] The invention relates to the technical field of behavior detection, in particular to a method for detecting violations in examinations based on 3D convolution. Background technique [0002] Exams are an important way to test learning outcomes, but in the process of examinations, there have been various forms of violations since ancient times. These behaviors disturb the order of the examination room and make the examination impossible to be conducted fairly and justly. Therefore, it is necessary to monitor the examination, find out the violations in the examination process, and punish them. At present, the means of detecting violations is mainly to install camera monitoring, record the examination process as a video, and then manually observe the video to find out the violations. This will inevitably take up a lot of human resources, and the time consumption is also huge. Therefore, it is very necessary to study intelligent recognition technology, an...

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): G06K9/00G06K9/62G06T7/269G06N3/04
CPCG06T7/269G06T2207/10016G06T2207/30196G06T2207/30232G06N3/045G06F18/2415G06F18/253
Inventor 刘栓
Owner 四川天翼网络股份有限公司
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