CN110992629A - Method for detecting static human body based on video monitoring - Google Patents
Method for detecting static human body based on video monitoring Download PDFInfo
- Publication number
- CN110992629A CN110992629A CN201911245118.XA CN201911245118A CN110992629A CN 110992629 A CN110992629 A CN 110992629A CN 201911245118 A CN201911245118 A CN 201911245118A CN 110992629 A CN110992629 A CN 110992629A
- Authority
- CN
- China
- Prior art keywords
- target
- static
- human body
- static human
- time
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B13/00—Burglar, theft or intruder alarms
- G08B13/18—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength
- G08B13/189—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems
- G08B13/194—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems using image scanning and comparing systems
- G08B13/196—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems using image scanning and comparing systems using television cameras
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
- G08B21/18—Status alarms
- G08B21/24—Reminder alarms, e.g. anti-loss alarms
Abstract
A method for detecting a static human body based on video monitoring comprises the steps of extracting a target from a real-time video image by adopting a Faster-RCNN deep learning network model and a background elimination method to obtain two coordinate parameters of the same target frame at the same time, analyzing the two coordinate parameters to judge whether the target is in a static state, and judging that the target is static and giving an alarm if the target is in the static state because the static target is blended into a background in background subtraction method modeling and cannot be detected, so when the target detected by the Faster-RCNN deep learning network model is not extracted in the background subtraction method, the target is in the static state; according to the invention, the static human body is identified through video monitoring, and an alarm is given for abnormal conditions such as long-time static, no-standing, faint and the like of an operator, so that the monitor can check or rescue in time, the work load of the monitor for watching the monitoring video for a long time is effectively reduced, the work efficiency is improved, and the labor cost is saved.
Description
Technical Field
The invention belongs to the technical field of pattern recognition, and particularly relates to a method for detecting a static human body based on video monitoring.
Background
In recent years, with the rapid development of economy, information technology is gradually popularized in various industries, and especially video monitoring is widely applied to social production and life, and plays a great role in ensuring life and property safety, maintaining social order and the like. The enterprise adopts technologies such as network, computer, video monitoring and the like to efficiently manage personnel and production flow, and the detection of the safety state of operating personnel becomes a problem to be solved urgently in the production safety of the enterprise.
At present, various methods are proposed for pedestrian detection, and the methods are roughly divided into three categories: the method is based on a global feature method, a human body part method and a stereoscopic vision method, and particularly utilizes the combination of a motion-based feature method and deep learning in the global feature method to obtain higher recognition rate, but the methods are suitable for detecting moving human bodies and lack the detection of static human bodies. At present, the cameras installed in workshops of most enterprises are used for monitoring the safety states of operating personnel, the monitoring personnel are in a manual judgment stage, the workload of the monitoring personnel is increased due to long-time monitoring, visual fatigue easily occurs, and the operating personnel in abnormal states such as static state, no-standing state, faint and the like cannot be rescued in time.
Disclosure of Invention
The invention provides a method for detecting a static human body based on video monitoring to solve the problems mentioned in the background technology.
The invention is realized by adopting the following technical scheme:
a method for detecting a static human body based on video monitoring comprises the steps of extracting a target from a real-time video image by adopting a Faster-RCNN deep learning network model and a background elimination method to obtain two coordinate parameters of the same target frame at the same time, analyzing the two coordinate parameters to judge whether the target is in a static state, judging that the target is static and giving an alarm if the target is in the static state when the target detected by the Faster-RCNN deep learning network model is not extracted in the background subtraction method because the static target is not detected due to the fact that the static target is blended into a background in background subtraction method modeling, and specifically comprising the following steps:
(1) image acquisition module acquiring and processing video stream
The real-time video stream is collected through a camera, the real-time video stream collected by the camera is uploaded to a static human body detection server through a switch for video decoding, and finally decoded data are stored at a designated position;
(2) target extraction module performs target extraction
The static human body detection server performs target extraction on the video stream data stored at the designated position in the step (1) by adopting a fast-RCNN deep learning network model and a background elimination method to obtain two coordinate parameters of the same target frame at the same time, extracts effective sample data, and updates coordinate parameter samples in a data set; the two methods respectively carry out modeling analysis on video stream data, and when the two methods are both analyzed, the two results are simultaneously stored in an SQL database for comparative analysis;
(3) alarm module alarm
The static human body detection server analyzes the two coordinate parameter samples of the same target at the same time stored in the SQL database in the step (2) and judges whether the target is in a static state; when the target detected by the fast-RCNN deep learning network model is not extracted in the background subtraction method, the target is judged to be in a static state; and setting static stay time, when the static target exceeds the set time, judging that the target is static by the system and giving an alarm, changing the target frame into red, and prompting monitoring personnel to take measures in time.
In the invention, in the step (1), the video stream is obtained through FFmpeg, and then CUDA decoding processing is adopted.
In the invention, in the step (2), the fast-RCNN deep learning network model consists of a VGG16 convolutional neural network model, an RPN region generation network and an ROI Pooling.
In the invention, in the step (2), the background subtraction method adopts a mean value modeling method to carry out background modeling.
The utility model provides a detect static human device based on video monitoring, including the camera, the switch, alarm and static human detection server, wherein, the camera is connected with the switch, the switch is connected with static human detection server, the alarm is connected with static human detection server, the camera gathers real-time image, upload to static human detection server by the switch with the real-time video stream that the camera gathered and carry out video decoding and carry out the target and draw and analysis simultaneously, judge whether there is the static human condition, if there is then send the warning through the alarm and remind.
Has the advantages that: the invention identifies the static human body through video monitoring, can be applied to monitoring workshop operators, can give an alarm aiming at abnormal conditions of long-time static, no standing, faint and the like of the operators, can be checked or rescued in time by the monitoring personnel, effectively relieves the work burden of the monitoring personnel for watching the monitoring video for a long time, improves the work efficiency and saves the labor cost.
Drawings
FIG. 1 is a flow chart illustrating a preferred embodiment of the present invention.
Detailed Description
In order to make the technical means, the creation characteristics, the achievement purposes and the effects of the invention easy to understand, the following description is further provided in combination with the specific drawings.
Referring to fig. 1, a method for detecting a still human body based on video monitoring specifically includes the following steps:
(1) image acquisition module acquiring and processing video stream
The real-time video stream is collected through a camera, and the real-time video stream collected by the camera is uploaded to a static human body detection server through a switch to be subjected to video decoding; the method comprises the steps of acquiring a video stream by using FFmpeg, processing the video stream by using CUDA decoding, and storing decoded data at a specified position;
(2) target extraction module performs target extraction
The static human body detection server performs target extraction on the video stream data stored at the designated position in the step (1) by adopting a fast-RCNN deep learning network model and a background elimination method to obtain two coordinate parameters of the same target frame at the same time, extracts effective sample data, and updates coordinate parameter samples in a data set; the two methods respectively carry out modeling analysis on video stream data, and when the two methods are both analyzed, the two results are simultaneously stored in an SQL database for comparative analysis;
(3) alarm module alarm
And (2) analyzing two coordinate parameter samples of the same target stored in an SQL database at the same time in the step (2) by the static human body detection server, judging whether the target is in a static state, defining target frame information extracted by the fast-RCNN deep learning network model as A, and similarly, B is target frame information extracted by a background subtraction method, wherein the static target cannot be detected due to the fact that the static target is blended into the background in the background subtraction method modeling, so when the target detected by the fast-RCNN deep learning network model is not extracted in the background subtraction method, the target in the A- (A ∩ B) is in the static state, namely the static target is identified, static staying time is set in the static human body detection server, when the static target exceeds the set time, the static target is judged to be static, an alarm is sent by the alarm module, and the target frame is red at the moment and the monitoring personnel is prompted to take measures in time.
The device comprises a camera, an alarm, a switch and a static human body detection server, wherein the camera acquires real-time images, the camera is connected with the switch, the switch transmits the received real-time images to the static human body detection server connected with the switch, then target extraction and analysis are carried out on the static human body detection server, whether static human body conditions exist or not is judged, and if yes, alarm reminding is sent out through the alarm.
In this embodiment, the fast-RCNN deep learning network model is composed of a VGG16 convolutional neural network model, an RPN region generation network, and ROI Pooling, and the background subtraction method is performed by using a mean value modeling method to perform background modeling.
Claims (8)
1. A method for detecting a static human body based on video monitoring is characterized in that a fast-RCNN deep learning network model and a background subtraction method are adopted to extract a target from a real-time video image to obtain two coordinate parameters of the same target frame at the same moment, then the two coordinate parameters are analyzed to judge whether the target is in a static state, the static target cannot be detected due to the fact that the static target is blended into a background in background subtraction method modeling, and therefore when the target detected by the fast-RCNN deep learning network model is not extracted in the background subtraction method, the target is judged to be in the static state, and an alarm is given out.
2. The method for detecting the static human body based on the video monitoring as claimed in claim 1 is characterized by comprising the following specific steps:
(1) image acquisition module acquiring and processing video stream
The real-time video stream is collected through a camera, the real-time video stream collected by the camera is uploaded to a static human body detection server through a switch for video decoding, and finally decoded data are stored at a designated position;
(2) target extraction module performs target extraction
The static human body detection server performs target extraction on the video stream data stored at the designated position in the step (1) by adopting a fast-RCNN deep learning network model and a background elimination method to obtain two coordinate parameters of the same target frame at the same time, extracts effective sample data, and updates coordinate parameter samples in a data set; the two methods respectively carry out modeling analysis on video stream data, and when the two methods are both analyzed, the two results are simultaneously stored in an SQL database for comparative analysis;
(3) alarm module alarm
The static human body detection server analyzes the two coordinate parameter samples of the same target at the same time stored in the SQL database in the step (2) and judges whether the target is in a static state; and when the target detected by the fast-RCNN deep learning network model is not extracted in the background subtraction method, judging that the target is in a static state, and alarming by an alarm module.
3. The method for detecting the static human body based on the video surveillance as claimed in claim 2, wherein in the step (1), the real-time video stream is obtained through FFmpeg.
4. The method for detecting the static human body based on the video surveillance as claimed in claim 2, wherein in the step (1), the real-time video stream is processed by CUDA decoding.
5. The method for detecting the static human body based on the video monitoring as claimed in claim 2, wherein in the step (2), the fast-RCNN deep learning network model is composed of a VGG16 convolutional neural network model, an RPN region generation network and ROIPooling.
6. The method for detecting the static human body based on the video monitoring as claimed in claim 2, wherein in the step (2), the background subtraction method is a mean value modeling method for background modeling.
7. The method for detecting the static human body based on the video monitoring as claimed in claim 2, wherein in the step (3), a static stay time is set in the static human body detection server.
8. The utility model provides a detect static human device based on video monitoring which characterized in that, includes camera, switch, alarm and static human detection server, and wherein, the camera is connected with the switch, and the switch is connected with static human detection server, and the alarm is connected with static human detection server.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911245118.XA CN110992629A (en) | 2019-12-06 | 2019-12-06 | Method for detecting static human body based on video monitoring |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911245118.XA CN110992629A (en) | 2019-12-06 | 2019-12-06 | Method for detecting static human body based on video monitoring |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110992629A true CN110992629A (en) | 2020-04-10 |
Family
ID=70091005
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911245118.XA Pending CN110992629A (en) | 2019-12-06 | 2019-12-06 | Method for detecting static human body based on video monitoring |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110992629A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113744488A (en) * | 2021-08-24 | 2021-12-03 | 上海柏格仕厨卫有限公司 | Intelligent cabinet with smoke alarm system |
CN116092130A (en) * | 2023-04-11 | 2023-05-09 | 东莞先知大数据有限公司 | Method, device and storage medium for supervising safety of operators in oil tank |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109325474A (en) * | 2018-11-14 | 2019-02-12 | 郭道宁 | A kind of abnormal state detection method of couple of special caregiver of need |
CN109636795A (en) * | 2018-12-19 | 2019-04-16 | 安徽大学 | Monitor video remnant object detection method without tracking in real time |
CN110147723A (en) * | 2019-04-11 | 2019-08-20 | 苏宁云计算有限公司 | The processing method and system of customer's abnormal behaviour in a kind of unmanned shop |
CN110287923A (en) * | 2019-06-29 | 2019-09-27 | 腾讯科技(深圳)有限公司 | Human body attitude acquisition methods, device, computer equipment and storage medium |
-
2019
- 2019-12-06 CN CN201911245118.XA patent/CN110992629A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109325474A (en) * | 2018-11-14 | 2019-02-12 | 郭道宁 | A kind of abnormal state detection method of couple of special caregiver of need |
CN109636795A (en) * | 2018-12-19 | 2019-04-16 | 安徽大学 | Monitor video remnant object detection method without tracking in real time |
CN110147723A (en) * | 2019-04-11 | 2019-08-20 | 苏宁云计算有限公司 | The processing method and system of customer's abnormal behaviour in a kind of unmanned shop |
CN110287923A (en) * | 2019-06-29 | 2019-09-27 | 腾讯科技(深圳)有限公司 | Human body attitude acquisition methods, device, computer equipment and storage medium |
Non-Patent Citations (1)
Title |
---|
阎相伊: "基于深度卷积神经网络的企业生产安全智能监控系统", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113744488A (en) * | 2021-08-24 | 2021-12-03 | 上海柏格仕厨卫有限公司 | Intelligent cabinet with smoke alarm system |
CN116092130A (en) * | 2023-04-11 | 2023-05-09 | 东莞先知大数据有限公司 | Method, device and storage medium for supervising safety of operators in oil tank |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2017125094A1 (en) | Body fall smart control system and method therefor | |
CN102665071B (en) | Intelligent processing and search method for social security video monitoring images | |
CN107133564B (en) | Tooling cap detection method | |
CN107679471B (en) | Indoor personnel air post detection method based on video monitoring platform | |
CN110751051B (en) | Abnormal driving behavior detection method based on machine vision | |
CN103942850B (en) | Based on medical personnel's monitoring method on duty of video analysis and RFID technique | |
CN106128022B (en) | A kind of wisdom gold eyeball identification violent action alarm method | |
CN110633612B (en) | Monitoring method and system for inspection robot | |
CN112396658A (en) | Indoor personnel positioning method and positioning system based on video | |
CN104751603A (en) | Rockfall monitoring and early warning system and rockfall monitoring and early warning method | |
KR101309366B1 (en) | System and Method for Monitoring Emergency Motion based Image | |
CN114155601A (en) | Vision-based method and system for detecting dangerous behaviors of operating personnel | |
CN106228709B (en) | A kind of wisdom gold eyeball identifies that one adds paper money alarm method and device | |
CN111259718B (en) | Staircase retention detection method and system based on Gaussian mixture model | |
EP3418991A1 (en) | Body fall smart control system and method therefor | |
CN110992629A (en) | Method for detecting static human body based on video monitoring | |
CN110781853A (en) | Crowd abnormality detection method and related device | |
CN111178424A (en) | Petrochemical production site safety compliance real-time detection system and method | |
CN101930540A (en) | Video-based multi-feature fusion flame detecting device and method | |
CN112530144A (en) | Method and system for warning violation behaviors of thermal power plant based on neural network | |
CN115223249A (en) | Quick analysis and identification method for unsafe behaviors of underground personnel based on machine vision | |
CN103093197A (en) | Monitoring method and system for recognizing hanging behavior | |
CN104284149A (en) | Electric line dangerous source point oversize vehicle shape change detection video recognition method | |
CN104392201A (en) | Human fall identification method based on omnidirectional visual sense | |
CN105741503A (en) | Parking lot real time early warning method under present monitoring device |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200410 |
|
RJ01 | Rejection of invention patent application after publication |