CN109657633A - A kind of far infrared security-protecting and monitoring method based on the learning machine that transfinites - Google Patents
A kind of far infrared security-protecting and monitoring method based on the learning machine that transfinites Download PDFInfo
- Publication number
- CN109657633A CN109657633A CN201811603224.6A CN201811603224A CN109657633A CN 109657633 A CN109657633 A CN 109657633A CN 201811603224 A CN201811603224 A CN 201811603224A CN 109657633 A CN109657633 A CN 109657633A
- Authority
- CN
- China
- Prior art keywords
- far infrared
- gradient
- infrared image
- target object
- pedestrian
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/20—Movements or behaviour, e.g. gesture recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
-
- 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
- G06V10/255—Detecting or recognising potential candidate objects based on visual cues, e.g. shapes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Human Computer Interaction (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Psychiatry (AREA)
- Social Psychology (AREA)
- Image Processing (AREA)
Abstract
The invention proposes a kind of far infrared security-protecting and monitoring methods based on the learning machine that transfinites, it is acquired by data, extract characteristic value, be trained characteristic value with the learning machine that transfinites, realization pedestrian in far infrared safety monitoring is detected, the present invention can real-time perfoming pedestrian detection, especially suitable for the monitoring under the conditions of night or haze etc., and manpower is greatly saved in the present invention, to reduce costs the benefit and safety that improve operation.
Description
Technical field
The present invention relates to field of computer technology more particularly to a kind of far infrared safety monitoring sides based on the learning machine that transfinites
Method.
Background technique
With the fast development of the relevant technologies such as computer vision, security protection intelligent monitoring and security fields, need to carry out intelligence
The occasion that can be monitored is more and more.Visual light imaging technology and computer vision technique are quite mature, but due to visible
The defect of trimmed book body, so that imaging effect of the imaging system under the adverse circumstances such as night, rainy day, haze is poor.Tradition is based on
There are security risks at unattended night for the security-protecting and monitoring method of visible light, it is therefore necessary to for harsh weather such as nights
Environment researches and develops a set of far infrared security-protecting and monitoring method.
Summary of the invention
For the defects in the prior art, the present invention provides a kind of far infrared safety monitoring sides based on the learning machine that transfinites
Method is marked the pedestrian detected on infrared image, and enabling video monitoring, person clearly has found pedestrian, and the monitoring
Method is additionally provided with warning function, provides alarming machine by monitoring room or supervisor's mobile phone terminal in the case where unattended
System.
In order to solve the above-mentioned technical problems, the present invention provides a kind of far infrared safety monitoring sides based on the learning machine that transfinites
Method thes improvement is that: described method includes following steps:
(1) far infrared image data set, including pedestrian and non-pedestrian two types are established, wherein set pedestrian's data set as
Target object；
(2) characteristic value is carried out to the far infrared image data set using infrared image gradient orientation histogram algorithm to mention
It takes, the edge feature including the target object, comprising:
1) far infrared image gradient is calculated；
2) gradient orientation histogram is calculated；
3) histogram normalized；
4) feature vector of histogram of gradients is calculated；
(3) characteristic value is trained by transfiniting learning machine as classifier；
(4) target object is detected；
(5) setting terminal warning function.
Wherein: step 1) the calculating far infrared image gradient, comprising:
1. the far infrared image is divided into lesser unit area, if unit is non-integer in partition process,
The far infrared image is expanded by the way of being rounded downwards；
2. horizontally and vertically being slided with step-length t, and calculate far infrared image level gradient and vertical ladder
Degree；
3. the gradient magnitude and gradient direction of far infrared image in computing unit:
In above-mentioned formula, g is gradient magnitude；gxFor the gradient in the direction horizontal direction x；gyFor the gradient in the direction vertical direction y；θ
For gradient direction；Δ be level off to 0 dimensionless.
Wherein: when step 2) the calculating gradient orientation histogram, the gradient direction being split, is divided into 8
Bin carries out statistics with histogram according to 8 bin, gradient orientation histogram abscissa be [0,7), the corresponding gradient side each bin
To for [0 °, 45 °), [45 °, 90 °) ..., [315 °, 360 °) }, ordinate is then accumulative gradient magnitude.
Wherein: when step 3) histogram normalized, accumulative gradient magnitude that the far infrared image set is obtained into
Row normalized.
Wherein: when the feature vector of step 4) the calculating histogram of gradients, judging that the far infrared image slices vegetarian refreshments is
It is no to be less than setting regions: if being not more than the setting regions, the feature vector dimension of entire far infrared image to be selected to be characterized
Description carries out classification based training；If more than the setting regions, then the far infrared image pixel point is cut, made it equal to
The setting regions.
Wherein: step (3) is trained the characteristic value, comprising:
I the learning machine parameter that transfinites) is determined；The learning machine input layer number that transfinites is to calculate the feature of histogram of gradients
The dimension of vector；
II the output matrix of hidden layer) is calculated；
III) according to least squares estimate find out hidden layer to output layer weight；
IV) pedestrian and the two kinds of data of non-pedestrian are trained by the learning machine that transfinites.
Wherein: when step (4) detects the target object, by the sliding window of the unit area to be detected
Far infrared image on slided, slide into a position every time, calculate the characteristic value in corresponding far infrared image gradient direction,
Then it will be compared by the training result that learning machine obtains that transfinites: if containing target object in matrix frame, being sorted out
For positive sample, that is, detection is completed, and goes out pedestrian with rectangle frame in the picture；If there is no target object in matrix frame, continue by
The sliding window of the unit area is slided on far infrared image to be detected, until containing target in the matrix frame
Object.
Wherein: step (5) setting terminal warning function, comprising:
The accounting for successively traversing target object described in the visual field of infrared safety monitoring, when accounting is more than 1/10, triggering
Level-one alarm mechanism；Secondary alarm mechanism is triggered when accounting is more than 1/5, issues secondary alarm signal, the prison to supervisor
Control person is tracked according to target object in real time, and judges whether to alarm.
Wherein: the lesser unit area includes 8*8 unit and its with lower unit.
Implement of the invention, has the advantages that
(1) the classifier selection in pedestrian of the present invention identification transfinites learning machine instead of support vector machines, overcome support to
The shortcomings that high extensive sample of amount machine runing time complexity is difficult to carry out.
(2) present invention is marked the pedestrian detected on infrared image, and enabling video monitoring, person clearly sends out
Existing people.
(3) present invention carries out the effect of security protection patrol worker at night, is also applicable in unattended security protection occasion.When infrared
After detecting pedestrian in image, alarm mechanism is provided, prompt video supervisor's monitoring area has pedestrian, in order to prevent video monitoring
Person can be networked by security threat and be linked into public security alarm system, i.e., can be with when video monitoring person discovery is by security threat
One it is bonded enter public security alarm system.
Detailed description of the invention
Fig. 1 is flow chart provided by the invention.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiments of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, ordinary skill people
Member's every other embodiment obtained without making creative work, shall fall within the protection scope of the present invention.
A kind of far infrared security-protecting and monitoring method based on the learning machine that transfinites provided in this embodiment, flow chart as shown in Figure 1,
Specifically comprise the following steps:
(1) far infrared image data set, including pedestrian and non-pedestrian two types (also referred to as positive and negative example) are established, wherein
Pedestrian's data set is set as target object.
(2) characteristic value is carried out to the far infrared image data set using infrared image gradient orientation histogram algorithm to mention
It takes, the edge feature including the target object, comprising:
1) far infrared image gradient is calculated, comprising:
1. setting far infrared image width as M, the far infrared image is divided into lesser unit area, i.e. 8*8 by a height of N
Unit and its with lower unit, the pixel that the present embodiment chooses 16*16 forms a unit, and infrared image width is divided into int (M/
8) unit, height are divided into int (N/8) unit, and int indicates to be rounded downwards here；
2. each unit 2*2 forms a block, the present embodiment carries out horizontally and vertically by taking step-length t=8 as an example
Sliding, and calculate far infrared image level gradient and vertical gradient；
3. the gradient magnitude and gradient direction of far infrared image in computing unit:
In above-mentioned formula, g is gradient magnitude；gxFor the gradient in the direction horizontal direction x；gyFor the gradient in the direction vertical direction y；θ
For gradient direction；Δ be level off to 0 dimensionless, the present embodiment takes Δ=10-8。
2) gradient orientation histogram is calculated.
3, according to the method described in claim 2, it is characterized by: when step 2) the calculating gradient orientation histogram,
The gradient direction is split, 8 bin (module) are divided into, carries out statistics with histogram according to 8 bin, gradient direction is straight
Square figure abscissa be [0,7), the corresponding gradient direction of each bin be [0 °, 45 °), [45 °, 90 °) ..., [315 °,
360 °) }, ordinate is then accumulative gradient magnitude.
3) histogram normalized.The present embodiment is normalized according to accumulative gradient magnitude obtained above,
Here normalization is mainly normalized to the accumulative gradient magnitude of general image rather than each bin is individually normalized, and reason exists
In the gradient of image be sensitive to overall brightness.
4) feature vector of histogram of gradients is calculated, comprising:
According to above-mentioned data, far infrared image width is divided into int (M/8) unit, and high score is int (N/8) unit, 2*2 composition
One block, using 8 pixels as step-length, then entire Infrared Image Features vector dimension P is [int (M/8) -1] × [int (N/8) -
1] × 4 × 8, judge whether the far infrared image slices vegetarian refreshments is less than setting regions: if selecting no more than the setting regions
The feature vector dimension of entire far infrared image is characterized description and carries out classification based training；If more than the setting regions, then by institute
It states far infrared image pixel point to be cut, makes it equal to the setting regions.The setting regions of the present embodiment by supervisor or
Other users setting.The present embodiment setting regions takes 128*64, will be red if infrared image pixel M × N is greater than 128*64
Outer image is cut, and 128*64 pixel is cut to, and the purpose of cutting is that the Infrared Image Features vector dimension calculated will not
It is too big.The wide unit number of infrared image is 128 ÷ 8=16 at this time, and high unit number is 64 ÷ 8=8, and unit number is in total
128, a total of 15*7=105 block of infrared image, the feature vector dimension P of infrared image is 105*4*8=3360.
(3) characteristic value is trained by transfiniting learning machine as classifier, is specifically included:
I the learning machine parameter that transfinites) is determined；The learning machine input layer number that transfinites is to calculate the feature of histogram of gradients
The dimension P of vector；
II the output matrix of hidden layer) is calculated:
Given Q different infrared images (training sample) It is input variable,It is pair
The target output variable answered, hidden layer output matrix are
WhereinIt is the weight variable and threshold value for connecting m-th of hidden node and input node, g respectively
() is activation primitive.
Single hidden layer feedforward neural networks meet
Above formula is reduced to H β=T, whereinIt is the weight vector for connecting m-th hidden node and output node.
III) according to least squares estimate find out hidden layer to output layer weight:
In formula,For the generalized inverse of hidden layer output matrix H.
IV) pedestrian and the two kinds of data of non-pedestrian are trained by the learning machine that transfinites.
(4) target object is detected；The present embodiment is first by the sliding window of the unit area to be checked
It is slided on the far infrared image of survey, slides into a position every time, calculate the feature in corresponding far infrared image gradient direction
Then value will be compared by the training result that learning machine obtains that transfinites: if containing target object in matrix frame, be returned
Class is positive sample, that is, completes detection, and go out pedestrian with rectangle frame in the picture；If there is no target object in matrix frame, continue
The sliding window of the unit area is slided on far infrared image to be detected, until containing mesh in the matrix frame
Mark object.
(5) setting terminal warning function specifically includes:
After infrared safety monitoring detects pedestrian, pedestrian, the square of pedestrian can be selected with rectangle frame in monitoring image
Shape frame is having a size of a × b；Define pedestrian accounting be
The accounting for successively traversing target object described in the visual field of infrared safety monitoring, when accounting is more than 1/10, triggering
Level-one alarm mechanism；Secondary alarm mechanism is triggered when accounting is more than 1/5, is issued secondary alarm signal to supervisor, is reminded prison
Control person's distance detection local is closer, and supervisor is tracked it according to the pedestrian that frame real-time in monitoring selects, and disobeys if pedestrian exists
Judicial act has threat video monitoring person behavior, can carry out a key report by the warning device being linked into public security system
It is alert.
In conclusion FPGA can be used the invention proposes a kind of far infrared security-protecting and monitoring method based on the learning machine that transfinites
Chip realize, pedestrian is detected, this method can real-time perfoming pedestrian detection, especially suitable for items such as night or hazes
Monitoring under part.Manpower is greatly saved in this method, to reduce costs the benefit and safety that improve operation.
Finally, it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；Although
Present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that: it still may be used
To modify the technical solutions described in the foregoing embodiments or equivalent replacement of some of the technical features；
And these are modified or replaceed, technical solution of various embodiments of the present invention that it does not separate the essence of the corresponding technical solution spirit and
Range.
Claims (9)
1. a kind of far infrared security-protecting and monitoring method based on the learning machine that transfinites, it is characterised in that: described method includes following steps:
(1) far infrared image data set, including pedestrian and non-pedestrian two types are established, wherein setting pedestrian's data set as target
Object；
(2) characteristics extraction, packet are carried out to the far infrared image data set using infrared image gradient orientation histogram algorithm
Include the edge feature of the target object, comprising:
1) far infrared image gradient is calculated；
2) gradient orientation histogram is calculated；
3) histogram normalized；
4) feature vector of histogram of gradients is calculated；
(3) characteristic value is trained by transfiniting learning machine as classifier；
(4) target object is detected；
(5) setting terminal warning function.
2. according to the method described in claim 1, it is characterized by: step 1) the calculating far infrared image gradient, comprising:
1. the far infrared image is divided into lesser unit area, if unit is non-integer in partition process, use
The mode being rounded downwards expands the far infrared image；
2. horizontally and vertically being slided with step-length t, and calculate far infrared image level gradient and vertical gradient；
3. the gradient magnitude and gradient direction of far infrared image in computing unit:
In above-mentioned formula, g is gradient magnitude；gxFor the gradient in the direction horizontal direction x；gyFor the gradient in the direction vertical direction y；θ is ladder
Spend direction；Δ be level off to 0 dimensionless.
3. according to the method described in claim 2, it is characterized by: when step 2) the calculating gradient orientation histogram, to institute
It states gradient direction to be split, is divided into 8 bin, carry out statistics with histogram, gradient orientation histogram abscissa according to 8 bin
For [0,7), the corresponding gradient direction of each bin be [0 °, 45 °), [45 °, 90 °) ..., [315 °, 360 °) }, ordinate is then
To add up gradient magnitude.
4., will be described remote red according to the method described in claim 1, it is characterized by: when step 3) histogram normalized
The accumulative gradient magnitude that outer image set obtains is normalized.
5. according to the method described in claim 1, it is characterized by: the step 4) feature vector for calculating histogram of gradients
When, judge whether the far infrared image slices vegetarian refreshments is less than setting regions: if selection is entire remote no more than the setting regions
The feature vector dimension of infrared image is characterized description and carries out classification based training；It, then will be described remote red if more than the setting regions
Outer image pixel point is cut, and the setting regions is made it equal to.
6. according to the method described in claim 3, it is characterized by: step (3) is trained the characteristic value, comprising:
I the learning machine parameter that transfinites) is determined；The learning machine input layer number that transfinites is to calculate the feature vector of histogram of gradients
Dimension；
II the output matrix of hidden layer) is calculated；
III) according to least squares estimate find out hidden layer to output layer weight；
IV) pedestrian and the two kinds of data of non-pedestrian are trained by the learning machine that transfinites.
7. according to the method described in claim 6, it is characterized by: when step (4) detects the target object, by institute
The sliding window for stating unit area is slided on far infrared image to be detected, slides into a position, calculating pair every time
The characteristic value in far infrared image gradient direction is answered, then will be compared by the training result that learning machine obtains that transfinites:
If containing target object in matrix frame, it is classified as positive sample, that is, completes detection, and go out pedestrian with rectangle frame in the picture；If
There is no target object in matrix frame, then continues to carry out the sliding window of the unit area on far infrared image to be detected
Sliding, until containing target object in the matrix frame.
8. according to the method described in claim 1, it is characterized by: step (5) setting terminal warning function, comprising:
The accounting for successively traversing target object described in the visual field of infrared safety monitoring triggers level-one when accounting is more than 1/10
Alarm mechanism；Secondary alarm mechanism is triggered when accounting is more than 1/5, issues secondary alarm signal, the supervisor to supervisor
It is tracked in real time according to target object, and judges whether to alarm.
9. according to the method described in claim 1, it is characterized by: the lesser unit area include 8*8 unit and its with
Lower unit.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811603224.6A CN109657633A (en) | 2018-12-26 | 2018-12-26 | A kind of far infrared security-protecting and monitoring method based on the learning machine that transfinites |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811603224.6A CN109657633A (en) | 2018-12-26 | 2018-12-26 | A kind of far infrared security-protecting and monitoring method based on the learning machine that transfinites |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109657633A true CN109657633A (en) | 2019-04-19 |
Family
ID=66116434
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811603224.6A Pending CN109657633A (en) | 2018-12-26 | 2018-12-26 | A kind of far infrared security-protecting and monitoring method based on the learning machine that transfinites |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109657633A (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103902976A (en) * | 2014-03-31 | 2014-07-02 | 浙江大学 | Pedestrian detection method based on infrared image |
| CN104239852A (en) * | 2014-08-25 | 2014-12-24 | 中国人民解放军第二炮兵工程大学 | Infrared pedestrian detecting method based on motion platform |
| CN106446792A (en) * | 2016-08-31 | 2017-02-22 | 大连楼兰科技股份有限公司 | Pedestrian detection feature extraction method in road traffic auxiliary driving environment |
| CN106600631A (en) * | 2016-11-30 | 2017-04-26 | 郑州金惠计算机***工程有限公司 | Multiple target tracking-based passenger flow statistics method |
| KR101866381B1 (en) * | 2016-06-02 | 2018-06-12 | 중앙대학교 산학협력단 | Apparatus and Method for Pedestrian Detection using Deformable Part Model |
| CN108399708A (en) * | 2018-05-17 | 2018-08-14 | 中恒智能工业设备（深圳）有限公司 | A kind of infrared thermal imagery alarm and alarm implementation method |
-
2018
- 2018-12-26 CN CN201811603224.6A patent/CN109657633A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103902976A (en) * | 2014-03-31 | 2014-07-02 | 浙江大学 | Pedestrian detection method based on infrared image |
| CN104239852A (en) * | 2014-08-25 | 2014-12-24 | 中国人民解放军第二炮兵工程大学 | Infrared pedestrian detecting method based on motion platform |
| KR101866381B1 (en) * | 2016-06-02 | 2018-06-12 | 중앙대학교 산학협력단 | Apparatus and Method for Pedestrian Detection using Deformable Part Model |
| CN106446792A (en) * | 2016-08-31 | 2017-02-22 | 大连楼兰科技股份有限公司 | Pedestrian detection feature extraction method in road traffic auxiliary driving environment |
| CN106600631A (en) * | 2016-11-30 | 2017-04-26 | 郑州金惠计算机***工程有限公司 | Multiple target tracking-based passenger flow statistics method |
| CN108399708A (en) * | 2018-05-17 | 2018-08-14 | 中恒智能工业设备（深圳）有限公司 | A kind of infrared thermal imagery alarm and alarm implementation method |
Non-Patent Citations (9)
| Title |
|---|
| ZHIJIE HUANG: "Pedestrian Detection Algorithm in Video Analysis Based on Centrist", 《2016 INTERNATIONAL CONFERENCE ON INTELLIGENT TRANSPORTATION, BIG DATA & SMART CITY》 * |
| 候启真: "《机场安全防范技术》", 30 April 2013 * |
| 吕慧: "增广四元数超限学习机的学习算法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 哈达: "基于运动目标检测的哨位夜间自动报警***", 《指挥与控制学报》 * |
| 朱聪聪: "基于梯度方向和强度直方图的红外行人检测", 《计算机工程》 * |
| 柳杨: "《数字图像物体识别理论详解与实战》", 31 March 2018 * |
| 洪晶: "海事场景的视频监控***", 《中国优秀硕士学位论文全文数据库 (信息科技辑)》 * |
| 苏松志: "《行人检测:理论与实践》", 31 March 2016 * |
| 郁磊: "《MATLAB智能算法30个案例分析》", 31 August 2015 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Gong et al. | A Real‐Time Fire Detection Method from Video with Multifeature Fusion | |
| CN111967393B (en) | Safety helmet wearing detection method based on improved YOLOv4 | |
| CN110135269B (en) | Fire image detection method based on mixed color model and neural network | |
| CN103824070B (en) | A kind of rapid pedestrian detection method based on computer vision | |
| CN104123544B (en) | Anomaly detection method and system based on video analysis | |
| CN103400111B (en) | Method for detecting fire accident on expressway or in tunnel based on video detection technology | |
| CN110428522A (en) | A kind of intelligent safety and defence system of wisdom new city | |
| CN109145689A (en) | A kind of robot fire detection method | |
| CN108875561A (en) | A kind of hidden danger repetition method of discrimination of transmission line of electricity monitoring hidden danger early warning image | |
| CN109101859A (en) | The method for punishing pedestrian in detection image using Gauss | |
| CN103514694A (en) | Intrusion detection monitoring system | |
| CN103366506A (en) | Device and method for automatically monitoring telephone call behavior of driver when driving | |
| CN110619277A (en) | Multi-community intelligent deployment and control method and system | |
| CN104881643B (en) | A kind of quick remnant object detection method and system | |
| CN105844245A (en) | Fake face detecting method and system for realizing same | |
| CN101311947A (en) | Real time intelligent control method based on natural video frequency | |
| CN105894701A (en) | Large construction vehicle identification and alarm method for preventing external damage to transmission lines | |
| CN109190475A (en) | A kind of recognition of face network and pedestrian identify network cooperating training method again | |
| CN110636281B (en) | Real-time monitoring camera shielding detection method based on background model | |
| CN109410497A (en) | A kind of monitoring of bridge opening space safety and alarm system based on deep learning | |
| CN110674753A (en) | Theft early warning method, terminal device and storage medium | |
| CN112183472A (en) | Method for detecting whether test field personnel wear work clothes or not based on improved RetinaNet | |
| CN108052865A (en) | A kind of flame detecting method based on convolutional neural networks and support vector machines | |
| CN106339657A (en) | Straw incineration monitoring method and device based on monitoring video | |
| CN109684946A (en) | A kind of kitchen mouse detection method based on the modeling of single Gaussian Background |
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: 20190419 |
|
| RJ01 | Rejection of invention patent application after publication |