CN108280952A - A kind of passenger's trailing monitoring method based on foreground object segmentation - Google Patents
A kind of passenger's trailing monitoring method based on foreground object segmentation Download PDFInfo
- Publication number
- CN108280952A CN108280952A CN201810074904.7A CN201810074904A CN108280952A CN 108280952 A CN108280952 A CN 108280952A CN 201810074904 A CN201810074904 A CN 201810074904A CN 108280952 A CN108280952 A CN 108280952A
- Authority
- CN
- China
- Prior art keywords
- monitoring
- image
- monitoring system
- trails
- main control
- 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.)
- Granted
Links
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
- G08B13/19602—Image analysis to detect motion of the intruder, e.g. by frame subtraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
Abstract
The invention discloses a kind of, and the passenger based on foreground object segmentation trails monitoring method comprising has the following steps：Step S1, monitoring system carry out Image Acquisition；Step S2, model selection；Step S3 trails monitoring：Monitoring system is directed to acquired image, and the Pixel-level that pedestrian image is carried out using deep neural network model is divided, and whether there is the case where being more than a people in the image of multiframe comprehensive descision segmentation later, and if it exists, then sends out " trailing alarm " signal to main control computer；Step S4 swarms into monitoring：Monitoring system compares present image and background image, and monitoring region is divided into multiple subregions, if there is object appearance in subregion, " swarming into alarm " signal is sent out to main control computer；Step S5 terminates monitoring.The present invention substantially increases monitoring accuracy, reduces the working strength of staff, at the same it is of low cost, be easily achieved.
Description
Technical field
The present invention relates to pedestrian monitoring methods more particularly to a kind of passenger based on foreground object segmentation to trail monitoring side
Method.
Background technology
Pedestrian monitoring system is equal AT STATION, and the crowd is dense that place is relatively conventional, and existing self-service channel is isolated equipped with two sects
Passenger, it is desirable that every time can only a passenger pass through channel, the passenger complete examination and leaving channel after, just perhaps next trip
Visitor enters channel, does not also allow adult to carry small children into channel, but it is (including straight in many ways to there is someone in actual use
Vertical walking, creeps or bends over forward etc. at row of squatting) enter channel in company with positive frequent flight passenger, there is also adults to carry small children into channel
Situation.These problems are coped with, existing pedestrian monitoring method includes following several：
The first is light curtain monitoring object, and one side of light curtain is equidistantly equipped with multiple infrared transmitting tubes, and another side is corresponding
Have an infrared receiving tube that identical quantity equally arranges, each infrared transmitting tube is corresponding, and there are one corresponding infrared receivers
Pipe, and on same straight line.When on same straight line infrared transmitting tube, there is no barrier between infrared receiving tube
When, the modulated signal (optical signal) that infrared transmitting tube is sent out can smoothly reach infrared receiving tube.Infrared receiving tube receives modulation
After signal, corresponding internal circuit exports low level, and in the case where there is barrier, the modulated signal that infrared transmitting tube is sent out
(optical signal) cannot smoothly reach infrared receiving tube, and at this moment the infrared receiving tube does not receive modulated signal, corresponding internal electricity
Road output is high level.When do not have in light curtain object by when, the modulated signal (optical signal) that all infrared transmitting tubes are sent out is all
The corresponding infrared receiving tube that the other side can smoothly be reached, to make internal circuit all output low levels.In this way, by inside
Circuit state, which carries out analysis, can monitor the information of object presence or absence.The defect of picking sensor technology is, for
Channel whether someone enter can accurate measurements come out, two people enter side by side or two modest people are immediately following entrance, Wu Faqu
It is divided to a people or two people, can causes to fail to report or report by mistake, the transmitting and receiving of infrared-ray is hindered if any dust, can cause to miss
Alarm.The heavy luggage carried for passenger may also judge by accident as people.
Second is range image analysis, and by depth image layering cutting, the analysis to each layer of progress connected domain, judgement is
It is no to have the appearance of two people's features；Lower layer, there are two independent communication domains, and movement locus has been monitored whether by frame-to-frame differences, then can be preliminary
It is judged as two people, then is analyzed by multilayer union, for example meet conditions above there are three layer, then sends out " tail to main control computer
With alarm " signal.If without monitoring signals, enters and swarm into monitoring.The defect of range image analysis technology is that forefathers carry on the back
Embrace entrances before and after mountain-climbing packet and two people, it is all similar round in depth map upper layer cutting figure, it is difficult to distinguish, can cause to report by mistake or
Person fails to report；Descendant's crouching row is creeped and heavy luggage is difficult to differentiate between forefathers in tow, can also cause to report by mistake or fail to report.
The third is target monitoring algorithm, comes out the pedestrian monitoring in channel using conventional target monitoring algorithm, passes through
The pedestrian's quantity monitored determines whether that trailing situation occurs, and detects one, normal through detecting two or more list
People then alarms.The defect of target monitoring algorithm is, the problem of due to field angle, double imaging easily overlaps, at this time mesh
The rectangle frame result for marking monitoring algorithm output has great deviation, flase drop missing inspection inevitably occurs.
Invention content
The technical problem to be solved in the present invention is, in view of the deficiencies of the prior art, provides a kind of based on foreground target point
The passenger that cuts trails monitoring method, and accurate alarm is sent out when entering channel now with more people to reality, and reach monitoring time it is short,
Monitoring range is wide, can accurate measurements to tailer simultaneously and alarm and other effects.
In order to solve the above technical problems, the present invention adopts the following technical scheme that.
A kind of passenger's trailing monitoring method based on foreground object segmentation, this method is based on monitoring system and main control computer is real
Existing, the method includes having the following steps：Step S1, the monitoring system carry out Image Acquisition；Step S2, model selection：Institute
It states monitoring system and model selection is carried out according to the control instruction that main control computer is sent, trail monitoring pattern if entering, execute step
If rapid S3 thens follow the steps S4 into monitoring pattern is swarmed into；Step S3 trails monitoring：The monitoring system is directed to and is acquired
Image, the Pixel-level that pedestrian image is carried out using deep neural network model are divided, in the image of multiframe comprehensive descision segmentation later
The case where with the presence or absence of more than a people, and if it exists, then send out " trailing alarm " signal to the main control computer；Step S4, is swarmed into
Monitoring：The monitoring system compares present image and background image, and monitoring region is divided into multiple subregions, if sub-district
There is object appearance in domain, then sends out " swarming into alarm " signal to main control computer；Step S5 terminates monitoring：If the monitoring system
" terminating monitoring " instruction for receiving main control computer transmission, then stop monitoring, if not receiving " terminating monitoring " instruction, persistently presets
Monitoring is ended automatically after time.
Preferably, in the step S1, the image of the monitoring system acquisition is color video frequency image.
Preferably, further include waiting for monitoring step after the step S1：The monitoring system is according to the change of background image
Change to determine whether that object enters, and main control computer is waited for send out control instruction.
Preferably, in the step S2, when someone swipes the card into monitoring region, the main control computer is to the monitoring
System sends control instruction.
Preferably, in the step S3, the deep neural network model passes through offline instruction in advance by the monitoring system
It gets out.
Preferably, in the step S4, the monitoring system is divided into 5 sub-regions by region is monitored.
Preferably, in the step S5, if not receiving " terminate monitoring " instruction, the monitoring systems stay after 5 seconds oneself
It is dynamic to terminate monitoring.
Preferably, in the step S5, if the monitoring system does not generate " trailing alarm " signal and " swarming into alarm " letter
Number, then it shows " normal through ".
Preferably, further include step S6, reset：Outlet is closed after pedestrian leaves monitoring region, the monitoring system connects
" removing alarm " instruction that main control computer is sent out is received, and updates background image.
Preferably, the monitoring region is walkway.
Passenger disclosed by the invention based on foreground object segmentation trails monitoring method, having compared to existing technologies
Beneficial effect is that the monitoring system can accurately and effectively find that more people enter channel by trailing monitoring and swarming into monitoring
Situation simultaneously provides alarm, while preserving anti-trailing video, compared to existing technologies, effectively reduces wrong report and fails to report, significantly
Improve monitoring accuracy, reduce the working strength of staff, at the same it is of low cost, be easily achieved.
Description of the drawings
Fig. 1 is that the present invention is based on the passengers of foreground object segmentation to trail monitoring method flow chart.
Specific implementation mode
The present invention is described in more detail with reference to the accompanying drawings and examples.
The invention discloses a kind of, and the passenger based on foreground object segmentation trails monitoring method, please refers to Fig. 1, this method base
Realize that the method includes having the following steps in monitoring system and main control computer：
Step S1, the monitoring system carry out Image Acquisition；
Step S2, model selection：The monitoring system carries out model selection according to the control instruction that main control computer is sent, if
Into monitoring pattern is trailed, S3 is thened follow the steps, if into monitoring pattern is swarmed into, thens follow the steps S4；
Step S3 trails monitoring：The monitoring system is directed to acquired image, is carried out using deep neural network model
The Pixel-level of pedestrian image is divided, and whether there is the case where being more than a people in the image of multiframe comprehensive descision segmentation later, and if it exists,
Then " trailing alarm " signal is sent out to the main control computer；
Step S4 swarms into monitoring：The monitoring system compares present image and background image, and will monitoring region point
At multiple subregions, if there is object appearance in subregion, " swarming into alarm " signal is sent out to main control computer；
Step S5 terminates monitoring：If the monitoring system receives " terminating monitoring " instruction of main control computer transmission, stop
Monitoring persistently ends automatically monitoring if not receiving " terminating monitoring " instruction after preset time.
In the above method, the monitoring system by trail monitor and swarm into monitoring, can accurately and effectively find more people into
The case where entering channel simultaneously provides alarm, while preserving anti-trailing video and effectively reducing wrong report and leakage compared to existing technologies
Report, substantially increases monitoring accuracy, reduces the working strength of staff, at the same it is of low cost, be easily achieved.
About acquired image, in the step S1, the image of the monitoring system acquisition is color video frequency image.
Further include waiting for monitoring step in the present embodiment, after the step S1：The monitoring system is according to background image
Variation determine whether that object enters, and main control computer is waited for send out control instruction.
Further, in the step S2, when someone swipes the card into monitoring region, the main control computer is to the monitoring
System sends control instruction.
As a preferred method, in the step S3, the deep neural network model is passed through by the monitoring system
Off-line training obtains in advance.
In the step S4 of the method for the present invention, the monitoring system is divided into 5 sub-regions by region is monitored.
As a preferred method, in the step S5, if not receiving " terminating monitoring " instruction, the monitoring system is held
Monitoring is ended automatically after 5 seconds continuous.
In the case of normal pass, in the step S5, if the monitoring system do not generate " trail alarm " signal and
" swarming into alarm " signal, then show " normal through ".
The present embodiment further includes step S6, is resetted：Outlet is closed after pedestrian leaves monitoring region, the monitoring system connects
" removing alarm " instruction that main control computer is sent out is received, and updates background image.
In the present embodiment, the monitoring region is walkway.
Passenger disclosed by the invention based on foreground object segmentation trails monitoring method, as shown in Figure 1, its practical application mistake
Implementation procedure in journey can refer to following examples：
Step 1, color video frequency image acquires.
Step 2, it waits to be monitored：It determines whether that object enters by background subtraction, and channel main control computer is waited for send
Various command signals.
Step 3, start monitoring pattern：When someone swipes the card into channel, main control computer sends " starting to monitor " instruction, into
Enter to trail mode monitoring, if not receiving " starting to monitor " instruction, enters and swarm into mode monitoring.
Step 4, monitoring is trailed：To the deep neural network model that the image of input is completed with off-line training is shifted to an earlier date, carry out
The Pixel-level of pedestrian is divided, and then whether there is the case where being more than a people in multiframe comprehensive descision segmentation image, if it find that having
More people exist, then send out " trailing alarm " signal to main control computer.
Step 5, monitoring is swarmed into：Present image and background are compared, passage area is divided into 5 zonules, if finding
There is object in exit, then sends out " swarming into alarm " signal to main control computer.
Step 6, terminate monitoring：If receiving " terminating monitoring " instruction that main control computer is sent, stop monitoring, if not having,
Then monitoring is ended automatically after 5 seconds.If not finding to alarm, show " normal through ".
Step 7, it resets：After people goes out, outlet portal is closed, receives " removing alarm " instruction that main control computer is sent, update
Background.
Passenger disclosed by the invention based on foreground object segmentation trails monitoring method, the example segmentation based on deep learning
The accurate Pixel-level class prediction to foreground target may be implemented in algorithm Mask RCNN, and Mask RCNN are divided into three parts, the
One is that core network is used for carrying out feature extraction, and second is that header structure is used for doing bounding box identification (classification and recurrence), the
Three are exactly that mask prediction is used for carrying out Pixel-level segmentation to each foreground target.For the anti-trailing scene in channel, it is contemplated that
Process object only can be people, and knapsack luggage is background classes, and the complicated field of anti-trailing channel can be directed to using example partitioning algorithm
Scape carries out Pixel-level examination, is hugged together Ru double, forefathers normal stand descendant squats down, and bends over can normally detect；Especially
It is to be directed to baby and child, the detail differences area of child and baby and luggage can be accurately utilized very much using Pixel-level segmentation
It separates, reduces flase drop.It is disposed for embedded device, Mask RCNN core networks is substituted for and are aiming at mobile terminal design
Mobilenet network structures, realize and anti-real-time anti-trailings for trailing scene monitored.Compared to existing technologies, this hair
It is bright to monitor the case where someone (people or more people) enters channel in company with passenger in many ways, including adult carries small children entrance
The case where channel, and can identify the luggage and articles for entering channel in company with passenger, luggage and articles will not be mistaken for personnel.This hair
Bright that accurate alarm is prompted in the case where there is more people to enter channel, monitoring time is short, and monitoring range covers entire channel, tailer
One, which enters, to find, alarm is timely.
The above is preferred embodiments of the present invention, is not intended to restrict the invention, all technology models in the present invention
Interior done modification, equivalent replacement or improvement etc. are enclosed, should be included in the range of of the invention protect.
Claims (10)
1. a kind of passenger based on foreground object segmentation trails monitoring method, which is characterized in that this method be based on monitoring system and
Main control computer realizes that the method includes having the following steps：
Step S1, the monitoring system carry out Image Acquisition；
Step S2, model selection：The monitoring system carries out model selection according to the control instruction that main control computer is sent, if into
Monitoring pattern is trailed, S3 is thened follow the steps, if into monitoring pattern is swarmed into, thens follow the steps S4；
Step S3 trails monitoring：The monitoring system is directed to acquired image, and pedestrian is carried out using deep neural network model
The Pixel-level of image is divided, the case where in the image of multiframe comprehensive descision segmentation later with the presence or absence of a people is more than, and if it exists, then to
The main control computer sends out " trailing alarm " signal；
Step S4 swarms into monitoring：The monitoring system compares present image and background image, and will monitoring region be divided into it is more
Sub-regions send out " swarming into alarm " signal if there is object appearance in subregion to main control computer；
Step S5 terminates monitoring：If the monitoring system receives " terminating monitoring " instruction of main control computer transmission, stop supervising
It surveys, if not receiving " terminating monitoring " instruction, monitoring is persistently ended automatically after preset time.
2. the passenger based on foreground object segmentation trails monitoring method as described in claim 1, which is characterized in that the step
In S1, the image of the monitoring system acquisition is color video frequency image.
3. the passenger based on foreground object segmentation trails monitoring method as described in claim 1, which is characterized in that the step
Further include waiting for monitoring step after S1：The monitoring system determines whether that object enters according to the variation of background image,
And main control computer is waited for send out control instruction.
4. the passenger based on foreground object segmentation trails monitoring method as described in claim 1, which is characterized in that the step
In S2, when someone swipes the card into monitoring region, the main control computer sends control instruction to the monitoring system.
5. the passenger based on foreground object segmentation trails monitoring method as described in claim 1, which is characterized in that the step
In S3, the deep neural network model is obtained by the monitoring system by shifting to an earlier date off-line training.
6. the passenger based on foreground object segmentation trails monitoring method as described in claim 1, which is characterized in that the step
In S4, the monitoring system is divided into 5 sub-regions by region is monitored.
7. the passenger based on foreground object segmentation trails monitoring method as described in claim 1, which is characterized in that the step
In S5, if not receiving " terminating monitoring " instruction, the monitoring systems stay ends automatically monitoring after 5 seconds.
8. the passenger based on foreground object segmentation trails monitoring method as described in claim 1, which is characterized in that the step
In S5, if the monitoring system does not generate " trailing alarm " signal and " swarming into alarm " signal, show " normal through ".
9. the passenger based on foreground object segmentation trails monitoring method as described in claim 1, which is characterized in that further include step
Rapid S6 resets：Outlet is closed after pedestrian leaves monitoring region, the monitoring system receives " the removing report that main control computer is sent out
It is alert " instruction, and update background image.
10. the passenger based on foreground object segmentation trails monitoring method as described in claim 1, which is characterized in that the prison
Survey region is walkway.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810074904.7A CN108280952B (en) | 2018-01-25 | 2018-01-25 | Passenger trailing monitoring method based on foreground object segmentation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810074904.7A CN108280952B (en) | 2018-01-25 | 2018-01-25 | Passenger trailing monitoring method based on foreground object segmentation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108280952A true CN108280952A (en) | 2018-07-13 |
| CN108280952B CN108280952B (en) | 2020-03-27 |
Family
ID=62805158
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810074904.7A Active CN108280952B (en) | 2018-01-25 | 2018-01-25 | Passenger trailing monitoring method based on foreground object segmentation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108280952B (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109118519A (en) * | 2018-07-26 | 2019-01-01 | 北京纵目安驰智能科技有限公司 | Target Re-ID method, system, terminal and the storage medium of Case-based Reasoning segmentation |
| CN109635740A (en) * | 2018-12-13 | 2019-04-16 | 深圳美图创新科技有限公司 | Video object detection method, device and image processing equipment |
| CN109657608A (en) * | 2018-12-17 | 2019-04-19 | 中通服公众信息产业股份有限公司 | Trailing analysis method based on face recognition technology |
| CN109977796A (en) * | 2019-03-06 | 2019-07-05 | 新华三技术有限公司 | Trail current detection method and device |
| CN110930568A (en) * | 2019-12-05 | 2020-03-27 | 江苏中云智慧数据科技有限公司 | Video anti-trailing system and method |
| CN110930427A (en) * | 2018-09-20 | 2020-03-27 | 银河水滴科技（北京）有限公司 | Image segmentation method, device and storage medium based on semantic contour information |
| CN111144231A (en) * | 2019-12-09 | 2020-05-12 | 深圳市鸿逸达科技有限公司 | Self-service channel anti-trailing detection method and system based on depth image |
| CN112037403A (en) * | 2020-07-22 | 2020-12-04 | 四川科达乐气象科技有限公司 | Personnel safety monitoring system and method based on embedded gateway |
| CN112597928A (en) * | 2020-12-28 | 2021-04-02 | 深圳市捷顺科技实业股份有限公司 | Event detection method and related device |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2012012555A1 (en) * | 2010-07-20 | 2012-01-26 | SET Corporation | Methods and systems for audience digital monitoring |
| US20140049647A1 (en) * | 2011-02-18 | 2014-02-20 | Hella Kgaa Hueck & Co. | Method for detecting target objects in a surveillance region |
| CN105139425A (en) * | 2015-08-28 | 2015-12-09 | 浙江宇视科技有限公司 | People counting method and device |
| CN105141885A (en) * | 2014-05-26 | 2015-12-09 | 杭州海康威视数字技术股份有限公司 | Method for video monitoring and device |
| CN106961576A (en) * | 2017-03-10 | 2017-07-18 | 盛视科技股份有限公司 | The anti-trailing method of video, apparatus and system |
| CN107563349A (en) * | 2017-09-21 | 2018-01-09 | 电子科技大学 | A kind of Population size estimation method based on VGGNet |
| CN107610148A (en) * | 2017-09-19 | 2018-01-19 | 电子科技大学 | A kind of foreground segmentation method based on Binocular Stereo Vision System |
-
2018
- 2018-01-25 CN CN201810074904.7A patent/CN108280952B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2012012555A1 (en) * | 2010-07-20 | 2012-01-26 | SET Corporation | Methods and systems for audience digital monitoring |
| US20140049647A1 (en) * | 2011-02-18 | 2014-02-20 | Hella Kgaa Hueck & Co. | Method for detecting target objects in a surveillance region |
| CN105141885A (en) * | 2014-05-26 | 2015-12-09 | 杭州海康威视数字技术股份有限公司 | Method for video monitoring and device |
| CN105139425A (en) * | 2015-08-28 | 2015-12-09 | 浙江宇视科技有限公司 | People counting method and device |
| CN106961576A (en) * | 2017-03-10 | 2017-07-18 | 盛视科技股份有限公司 | The anti-trailing method of video, apparatus and system |
| CN107610148A (en) * | 2017-09-19 | 2018-01-19 | 电子科技大学 | A kind of foreground segmentation method based on Binocular Stereo Vision System |
| CN107563349A (en) * | 2017-09-21 | 2018-01-09 | 电子科技大学 | A kind of Population size estimation method based on VGGNet |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109118519A (en) * | 2018-07-26 | 2019-01-01 | 北京纵目安驰智能科技有限公司 | Target Re-ID method, system, terminal and the storage medium of Case-based Reasoning segmentation |
| CN110930427B (en) * | 2018-09-20 | 2022-05-24 | 银河水滴科技（北京）有限公司 | Image segmentation method, device and storage medium based on semantic contour information |
| CN110930427A (en) * | 2018-09-20 | 2020-03-27 | 银河水滴科技（北京）有限公司 | Image segmentation method, device and storage medium based on semantic contour information |
| CN109635740A (en) * | 2018-12-13 | 2019-04-16 | 深圳美图创新科技有限公司 | Video object detection method, device and image processing equipment |
| CN109635740B (en) * | 2018-12-13 | 2020-07-03 | 深圳美图创新科技有限公司 | Video target detection method and device and image processing equipment |
| CN109657608A (en) * | 2018-12-17 | 2019-04-19 | 中通服公众信息产业股份有限公司 | Trailing analysis method based on face recognition technology |
| CN109977796A (en) * | 2019-03-06 | 2019-07-05 | 新华三技术有限公司 | Trail current detection method and device |
| CN110930568A (en) * | 2019-12-05 | 2020-03-27 | 江苏中云智慧数据科技有限公司 | Video anti-trailing system and method |
| CN111144231A (en) * | 2019-12-09 | 2020-05-12 | 深圳市鸿逸达科技有限公司 | Self-service channel anti-trailing detection method and system based on depth image |
| WO2021114765A1 (en) * | 2019-12-09 | 2021-06-17 | 深圳市鸿逸达科技有限公司 | Depth image-based method and system for anti-trailing detection of self-service channel |
| CN111144231B (en) * | 2019-12-09 | 2022-04-15 | 深圳市鸿逸达科技有限公司 | Self-service channel anti-trailing detection method and system based on depth image |
| CN112037403A (en) * | 2020-07-22 | 2020-12-04 | 四川科达乐气象科技有限公司 | Personnel safety monitoring system and method based on embedded gateway |
| CN112597928A (en) * | 2020-12-28 | 2021-04-02 | 深圳市捷顺科技实业股份有限公司 | Event detection method and related device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108280952B (en) | 2020-03-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108280952A (en) | A kind of passenger's trailing monitoring method based on foreground object segmentation | |
| CN108037770B (en) | Unmanned aerial vehicle power transmission line inspection system and method based on artificial intelligence | |
| CN105799740B (en) | A kind of track foreign body intrusion automatic detection and method for early warning based on technology of Internet of things | |
| CN101334924B (en) | Fire hazard probe system and its fire hazard detection method | |
| CN105812733B (en) | A kind of scene monitoring guiding system of air traffic control | |
| CN103108159B (en) | Electric power intelligent video analyzing and monitoring system and method | |
| EP1281166B1 (en) | Imaging fire detector | |
| CN100582654C (en) | Height measuring method and its measuring device | |
| CN106203265A (en) | A kind of Construction Fugitive Dust Pollution based on unmanned plane collection image is derived from dynamic monitoring and coverage prognoses system and method | |
| CN101751744B (en) | Detection and early warning method of smoke | |
| CN108334103A (en) | Unmanned plane multiple spurs is from barrier-avoiding method and obstacle avoidance system | |
| Niu et al. | A UAV-based traffic monitoring system-invited paper | |
| CN109993789A (en) | A kind of the separated of shared bicycle stops determination method, device and camera | |
| CN109118479A (en) | Defects of insulator identification positioning device and method based on capsule network | |
| CN107992067A (en) | Unmanned plane inspection fault diagnosis system based on integrated gondola and AI technologies | |
| CN108898085A (en) | A kind of road disease intelligent detecting method based on mobile video | |
| CN108229407A (en) | A kind of behavioral value method and system in video analysis | |
| CN112633057B (en) | Intelligent monitoring method for abnormal behavior in bus | |
| CN106550326A (en) | A kind of scenic spot passenger flow detecting system and method based on wireless apss and WiFi network | |
| CN109766746A (en) | A kind of track foreign matter detecting method of unmanned plane video | |
| CN109644360A (en) | A method of differentiating that pedestrian flows to using WI-FI probe | |
| CN201091014Y (en) | Fire detecting device | |
| CN110723621A (en) | Device and method for detecting smoking in elevator car based on deep neural network | |
| CN107688793A (en) | A kind of outside transformer substation fire automatic monitoring method for early warning | |
| CN111898444A (en) | Aircraft landing gear state determination method based on image recognition |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CP02 | Change in the address of a patent holder | ||
| CP02 | Change in the address of a patent holder |
Address after: 1601 and 1605, East Block 2, Tian'an innovation Plaza, chegongmiao, Futian District, Shenzhen City, Guangdong Province Patentee after: MAXVISION TECHNOLOGY Co.,Ltd. Address before: 518000 Guangdong city of Shenzhen province Futian District Shatou street Che Kung Temple Tairan six road Tairan Cangsong building 511, South Tower 512, 513, 515, 516, 517, 518 (only office) Patentee before: MAXVISION TECHNOLOGY Co.,Ltd. |