CN106096499A - A kind of video image culminant star moon pattern detection method and system - Google Patents
A kind of video image culminant star moon pattern detection method and system Download PDFInfo
- Publication number
- CN106096499A CN106096499A CN201610363846.0A CN201610363846A CN106096499A CN 106096499 A CN106096499 A CN 106096499A CN 201610363846 A CN201610363846 A CN 201610363846A CN 106096499 A CN106096499 A CN 106096499A
- Authority
- CN
- China
- Prior art keywords
- image
- foreground
- detection
- star moon
- video image
- 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
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
-
- 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
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
- G06F18/2148—Generating training patterns; Bootstrap methods, e.g. bagging or boosting characterised by the process organisation or structure, e.g. boosting cascade
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/46—Extracting features or characteristics from the video content, e.g. video fingerprints, representative shots or key frames
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to the method and system of a kind of video image culminant star moon pattern detection, the method is by a frame video image area-of-interest;Area-of-interest is carried out foreground detection by default foreground detection method and obtains foreground image;With the detection of classifier star moon pattern comprising default star moon feature in foreground image profile, this foreground image of labelling also outwards expands the statistical regions that one default enlarged area of formation detects as subsequent frame image centered by this foreground image region, then with the star moon pattern in statistical regions in this foreground image place frame of detection of classifier later continuous print predetermined number two field picture, alarm is exported when the number of times detected is eligible;Thus provide a kind of effective detection means for effectively carrying out star moon pattern detection in video image.
Description
Technical field
The invention belongs to image identification technical field, be specifically related to the method for a kind of video image star moon pattern detection and be
System.
Background technology
In some unsafe areas, for the needs of social safety, need some personages wearing special dress ornament are entered
Row prevention detection the most in advance identifies.Although the image that can use image capture device Real-time Collection monitoring region at present enters
Row monitoring, but existing monitoring technology is only monitoring afterwards, for some non-persistent personages such as the band star moon pattern mark people
Thing cannot be accomplished detect identification in advance and give warning in advance, it is impossible to accomplishes to monitor in advance, prevent in advance.
Summary of the invention
It is an object of the invention to solve above-mentioned technical problem and provide a kind of video image culminant star moon pattern detection
Method and system.
For achieving the above object, the present invention adopts the following technical scheme that
A kind of video image culminant star moon pattern detection method, comprise the following steps:
Obtain a frame video image and area-of-interest to be detected in this frame video image is set;
Described area-of-interest is carried out foreground detection, the foreground image of output binaryzation by default foreground detection method;
Obtain described foreground image profile and in this profile with the detection of classifier comprising default star moon feature whether
There is star moon pattern;
If star moon pattern detected in this profile, then this foreground image of labelling be with this foreground image region
The heart outwards expands and forms the statistical regions that a default enlarged area detects as subsequent frame image, then examines with described grader
Survey in this foreground image place frame continuous print predetermined number two field picture later in this statistical regions, whether have star moon pattern;
If detecting in described predetermined number two field picture, the frame number of star moon pattern more than presetting flase drop frame number, then exports report
Alert prompting.
Described area-of-interest is the human face region that in this frame video image, definition meets default definition values.
Described default foreground detection method is Vibe foreground detection method.
Described grader is adaboost cascade of strong classifiers.
The step tool of the profile of the described foreground image of described acquisition is:
After the foreground image of described binaryzation does repeatedly corrosion, connection becomes a region, then is obtained by rim detection
The profile of foreground image place rectangle.
The present invention also aims to provide the system of a kind of video image culminant star moon pattern detection, including:
Detection region arranges module, is used for obtaining a frame video image and arranges sense to be detected in this frame video image emerging
Interest region;
Foreground image acquisition module, for described area-of-interest is carried out foreground detection by default foreground detection method,
The foreground image of output binaryzation;
Pattern detection module, for obtaining the profile of described foreground image and with comprising default star moon feature in this profile
Detection of classifier whether have star moon pattern;
Subsequent frame detection module, after star moon pattern detected in this profile, this foreground image of labelling with before this
Outwards expand centered by scape image region and form the statistical regions that a default enlarged area detects as subsequent frame image,
Then with in continuous print predetermined number two field picture after this foreground image place frame of described detection of classifier in this statistical regions
Whether there is star moon pattern;
Alarm module, for detecting that in described predetermined number two field picture the frame number of star moon pattern is more than presetting flase drop
Frame number, exports alarm.
The present invention, by above technical scheme, can detect on the character physical in image rapidly in video image
Whether there is star moon pattern, and report to the police when the number of times detecting star moon pattern reaches the secondary numerical value set, thus for regard
Frequently image detects on the character physical in image the star moon pattern effectiveness provide one and effectively detect solution.
Accompanying drawing explanation
Fig. 1 provides for the embodiment of the present invention the video image culminant star moon pattern detection the flow chart of method;
The schematic diagram of cascade of strong classifiers shown in Fig. 2.
Detailed description of the invention
Below, in conjunction with example, substantive distinguishing features and the advantage of the present invention are further described, but the present invention not office
It is limited to listed embodiment.
Shown in Figure 1, a kind of video image culminant star moon pattern detection method, including:
S101: obtain a frame video image and area-of-interest to be detected in this frame video image is set;
S102: described area-of-interest is carried out foreground detection, the prospect of output binaryzation by default foreground detection method
Image;
S103: obtain the profile of described foreground image and with the detection of classifier comprising default star moon feature in this profile
Whether there is star moon pattern;
S104: if star moon pattern being detected in this profile, then this foreground image of labelling with this foreground image location
Outwards expand centered by territory and form the statistical regions that a default enlarged area detects as subsequent frame image, then with described point
Class device detects in this foreground image place frame continuous print predetermined number two field picture later whether have star moon figure in this statistical regions
Case;
S105: the frame number of star moon pattern is more than presetting flase drop frame number, then if detecting in described predetermined number two field picture
Output alarm, otherwise returns S101 detection.
Wherein, in the present invention, described area-of-interest is that in this frame video image, definition meets default definition values
Human face region, i.e. refers to human face region region the most clearly in whole video image.
Concrete, in step S101, obtain a frame video image and arrange in this frame video image to be detected interested
Region particularly as follows:
Obtain the width of sequence of video images and height, according to the width of sequence of video images and high calculate area-of-interest
Position and width in video image are high, thus realize the setting to area-of-interest.
After arranging area-of-interest, follow-up detection will be carried out in area-of-interest, by arranging region of interest
Territory, fewer than detection time in entire image a lot, thus improve detection speed.
In described step S102, described area-of-interest is carried out foreground detection by default foreground detection method, mainly
In order to judge whether described area-of-interest exists foreground image, as there is the image then exporting binaryzation, otherwise return
S101 detects, and when foreground image having been detected, obtains the profile of foreground image and with comprising star moon feature in prospect profile
Grader detect, to determine whether star moon pattern.
The described grader comprising star moon feature is adaboost cascade classifier.
In the present invention, described default foreground detection method is Vibe (Visual Background extractor) prospect
Detection method.
Wherein, in the present invention, the step tool of the profile of the described foreground image of described acquisition is:
After the foreground image of described binaryzation does repeatedly corrosion, connection becomes a region, then is obtained by rim detection
The profile of foreground image place rectangle.
Described Vibe foreground detection method i.e. utilizes ViBe model to realize, below to ViBe foreground detection model
Illustrate.
1, ViBe model operation principle
Background object refers to the static or object slowly moved, and foreground object is the object of corresponding movement.
Object detection can be regarded as a classification problem, namely determine whether a pixel belongs to background dot.At ViBe mould
In type, background model is that each background dot stores a sample set, then each new pixel value and sample set is carried out
Relatively judge whether to belong to background dot, thus may know that if a new observed value belongs to background dot, then it should
With the sampled value in sample set relatively.
Concretely, note v (x): the pixel value at x point;M (x)=V1, V2 ... VN} is the background sample set (sample at x
This collection size is N);SR (v (x)): R centered by the x region as radius, if M (x) [SR (v (x)) ∩ v1, v2 ...,
VN}}] more than given threshold value min, it is considered that x point belongs to background dot.
2, ViBe model initialization method
ViBe initializes the process of the sample set being exactly filler pixels, owing to can not comprise pixel in a two field picture
Spatial and temporal distributions information, may utilize close pixel and had close spatial and temporal distributions characteristic, specifically: for a picture
Vegetarian refreshments, the pixel value of the random neighbours' point selecting it is as its model sample value.M0 (x)=v0 (y | y ∈ NG (x)) }, t
=0 initial time, NG (x) is neighbours' point.This initial method advantage is that the reaction for noise is sensitiveer, amount of calculation
Little speed is fast, can the detection carrying out moving object quickly.
3, ViBe model modification strategy
More New Policy can be conservative more New Policy+foreground point method of counting.Foreground point counts: unite pixel
Meter, if certain pixel N continuous time is detected as prospect, is then updated to background dot.Randomly sub-sampled: new at each
Frame of video in all go to update the sample value of each pixel in background model it is not necessary that, when a pixel quilt
When being categorized as background dot, it has the probability of 1/ φ to go to update background model.
Concrete update method: each background dot has the probability of 1/ φ to go to update the model sample value of oneself, also has simultaneously
The probability of 1/ φ goes to update the model sample value of its neighbours' point.The sample value of more new neighbor make use of the spatial transmission of pixel value
Characteristic, background model is gradually to external diffusion, and this is also beneficial to the faster identification in Ghost region.Reach when foreground point counting simultaneously
It is changed into background during to marginal value, and has the probability of 1/ φ to go to update the model sample value of oneself.Selecting sample to be replaced
When the sample value of this concentration, randomly select a sample value and be updated, so can ensure that the smooth life of sample value
Cycle, the probability that such a sample value is not updated at moment t was (N-1)/N owing to being random renewal, it is assumed that the time is
Continuous print, then after past time of dt, the probability that sample value still retains is:
Also can write,
Showing the most whether a sample value is replaced unrelated with time t, randomized policy is suitable.
Drawn by foreground detection is a bianry image, and foreground image pixel value is 1, and background image pixel values is 0,
It is not a connected region due to foreground image, this bianry image is done 5 corrosion, connected and become a region, then
The size and location of foreground image areas place rectangle is obtained by rim detection, so that it is determined that the profile of foreground image.
Wherein, in the present invention, described grader is adaboost cascade classifier.Adaboost cascade classifier is to pass through
Acquire the positive sample of 5000 features comprising star moon pattern and an adaboost cascade point that 300,000 negative samples are trained to
Class device.Detecting at the contour area of foreground image with this grader, if detecting star moon pattern, exporting it in the picture
Position and size.
The principle of Adaboost cascade classifier is as follows:
Adaboost cascade classifier is to utilize Weak Classifier that classification capacity is general by certain method superposition
(boost) get up, the strong classifier that composition and classification is very capable.The arthmetic statement of Adaboost training strong classifier is as follows:
Given a series of training sample (x1,y1),(x2,y2),...(xn,yn), wherein xiRepresent i-th sample, yiWhen=1
For positive sample (face), yi=0 represents negative sample (non-face).To each feature featurej, train a Weak Classifier hj
(x), the Weak Classifier afterwards each feature generated calculating weighted error:
The grader with minimal weight error e j is added in strong classifier, and the probability updating training sample divides
Cloth:
Wherein βj=εj/1-εj, ei=0 represents that sample xi is correctly classified, and ei=1 represents and correctly do not classified, final structure
One-tenth strong classifier:
Wherein, b is the threshold value arranged, and is defaulted as 0.
Adaboost cascade classifier as shown in Figure 2, by multiple Haar-like features F1F2…FnJointly form certainly
Plan tree, by inputting data and the plurality of Haar-like feature F1F2…FnCompare, it is judged that meet and be output as face
(face), the most incongruent it is output as non-face (not face).
Viola-Jones detector utilizes waterfall (Cascade) algorithm classification device to be organized as the cascade classifier of screening type,
Each node of cascade is that AdaBoost trains the strong classifier obtained.Each node in cascade arranges threshold value b so that almost
All face samples can be transferred through, and most non-face sample can not pass through.Node arranges from simple to complex, and position is more
Node rearward is the most complicated, i.e. comprises the most Weak Classifiers.So energy minimization refusal image but amount of calculation during region, logical
Know and ensure the high detection rate of grader and low reject rate.Such as it is 99.9% at discrimination, when reject rate is 50%, (99.9%
Face and 50% non-face can pass through), total discrimination of 20 nodes is: 98%, and false acceptance rate is only:
0.0001%.
In the present invention, in order to ensure the precision of detection, the method for successive frame statistics is used to get rid of flase drop, concrete, permissible
It is to add up in units of 10 frames, when certain frame detects star moon pattern, according to previously obtained foreground image areas at image
In position and size, centered by the rectangle of foreground image areas place, outwards expand the 1/3 of its length of side, using this region as
Statistical regions, if the star moon pattern detected in this statistical regions in continuous print 9 frame after this frame is less than 5 frames, it is determined that for by mistake
Inspection, returns step S101 and continues detection, be otherwise judged to just detecting, and output is reported to the police.
The present invention also aims to provide the system of a kind of video image culminant star moon pattern detection, including:
Detection region arranges module, is used for obtaining a frame video image and arranges sense to be detected in this frame video image emerging
Interest region;
Foreground image acquisition module, for described area-of-interest is carried out foreground detection by default foreground detection method,
The foreground image of output binaryzation;
Pattern detection module, for obtaining the profile of described foreground image and with comprising default star moon feature in this profile
Detection of classifier whether have star moon pattern;
Subsequent frame detection module, after star moon pattern detected in this profile, this foreground image of labelling with before this
Outwards expand centered by scape image region and form the statistical regions that a default enlarged area detects as subsequent frame image,
Then with in continuous print predetermined number two field picture after this foreground image place frame of described detection of classifier in this statistical regions
Whether there is star moon pattern;
Alarm module, for detecting that in described predetermined number two field picture the frame number of star moon pattern is more than presetting flase drop
Frame number, exports alarm.
The present invention, by above technical scheme, can detect on the character physical in image rapidly in video image
Whether there is star moon pattern, and report to the police when the number of times detecting star moon pattern reaches the secondary numerical value set, thus for regard
Frequently image detects on the character physical in image the star moon pattern effectiveness provide one and effectively detect solution.
This video image culminant star moon pattern detection the detection process of system and method, refer to foregoing video
The image culminant star moon pattern detection the detection process of method and method, be the most no longer described in detail.
The present invention, by above technical scheme, can detect on the character physical in image rapidly in video image
Whether there is star moon pattern, and report to the police when the number of times detecting star moon pattern reaches the secondary numerical value set, thus for regard
Frequently image detects on the character physical in image the star moon pattern effectiveness provide one and effectively detect solution.
The above is only the preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For Yuan, under the premise without departing from the principles of the invention, it is also possible to make some improvements and modifications, these improvements and modifications also should
It is considered as protection scope of the present invention.
Claims (10)
1. one kind the video image culminant star moon pattern detection method, it is characterised in that comprise the following steps:
Obtain a frame video image and area-of-interest to be detected in this frame video image is set;
Described area-of-interest is carried out foreground detection, the foreground image of output binaryzation by default foreground detection method;
Obtain the profile of described foreground image and whether have star with the detection of classifier comprising default star moon feature in this profile
Month pattern;
If star moon pattern being detected in this profile, then this foreground image of labelling and centered by this foreground image region to
Outer expansion forms the statistical regions that a default enlarged area detects as subsequent frame image, then should with described detection of classifier
After the frame of foreground image place, whether continuous print predetermined number two field picture there is star moon pattern in this statistical regions;
If detecting in described predetermined number two field picture, the frame number of star moon pattern more than presetting flase drop frame number, then exports warning and carries
Show.
The most according to claim 1 the video image culminant star moon pattern detection method, it is characterised in that described area-of-interest
The human face region of default definition values is met for definition in this frame video image.
The most according to claim 1 the video image culminant star moon pattern detection method, it is characterised in that described default prospect examine
Survey method is Vibe foreground detection method.
The most according to claim 1 the video image culminant star moon pattern detection method, it is characterised in that described grader is
Adaboost cascade of strong classifiers.
5. according to described in any one of claim 1-4 the video image culminant star moon pattern detection method, it is characterised in that described in obtain
The step tool of the profile taking described foreground image is:
After the foreground image of described binaryzation does repeatedly corrosion, connection becomes a region, then obtains prospect by rim detection
The profile of image place rectangle.
6. one kind the video image culminant star moon pattern detection system, it is characterised in that including:
Detection region arranges module, for obtaining a frame video image and arranging region of interest to be detected in this frame video image
Territory;
Foreground image acquisition module, for described area-of-interest is carried out foreground detection by default foreground detection method, output
The foreground image of binaryzation;
Pattern detection module, for obtaining the profile of described foreground image and with comprising dividing of default star moon feature in this profile
Class device has detected whether star moon pattern;
Subsequent frame detection module, after star moon pattern detected in this profile, this foreground image of labelling with this foreground picture
The statistical regions that a default enlarged area detects is formed as subsequent frame image, then as outwards expanding centered by region
With in continuous print predetermined number two field picture after this foreground image place frame of described detection of classifier in this statistical regions whether
There is star moon pattern;
Alarm module, for detecting that in described predetermined number two field picture the frame number of star moon pattern is more than presetting flase drop frame
Number, exports alarm.
The most according to claim 6 the video image culminant star moon pattern detection system, it is characterised in that described area-of-interest
The human face region of default definition values is met for definition in this frame video image.
The most according to claim 6 the video image culminant star moon pattern detection system, it is characterised in that described default prospect examine
Survey method is Vibe foreground detection method.
The most according to claim 6 the video image culminant star moon pattern detection system, it is characterised in that described grader is
Adaboost cascade of strong classifiers.
10. according to described in any one of claim 6-9 the video image culminant star moon pattern detection system, it is characterised in that described
Pattern detection module includes:
Image outline acquiring unit, after the foreground image of described binaryzation is done repeatedly corrosion, connection becomes a region,
The profile of foreground image is obtained again by rim detection.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610363846.0A CN106096499A (en) | 2016-05-26 | 2016-05-26 | A kind of video image culminant star moon pattern detection method and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610363846.0A CN106096499A (en) | 2016-05-26 | 2016-05-26 | A kind of video image culminant star moon pattern detection method and system |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106096499A true CN106096499A (en) | 2016-11-09 |
Family
ID=57229772
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610363846.0A Pending CN106096499A (en) | 2016-05-26 | 2016-05-26 | A kind of video image culminant star moon pattern detection method and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106096499A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108322699A (en) * | 2018-01-23 | 2018-07-24 | 东南大学 | A method of based on video images detection arc-flash |
CN110363197A (en) * | 2019-06-22 | 2019-10-22 | 东北电力大学 | Based on the video area-of-interest exacting method for improving visual background extraction model |
CN112733666A (en) * | 2020-12-31 | 2021-04-30 | 湖北亿咖通科技有限公司 | Method, equipment and storage medium for collecting difficult images and training models |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102147869A (en) * | 2011-03-31 | 2011-08-10 | 上海交通大学 | Pedestrian detection method based on foreground analysis and pattern recognition |
CN103198332A (en) * | 2012-12-14 | 2013-07-10 | 华南理工大学 | Real-time robust far infrared vehicle-mounted pedestrian detection method |
CN103473788A (en) * | 2013-07-31 | 2013-12-25 | 中国电子科技集团公司第三十八研究所 | Indoor fire and flame detection method based on high-definition video images |
CN104866843A (en) * | 2015-06-05 | 2015-08-26 | 中国人民解放军国防科学技术大学 | Monitoring-video-oriented masked face detection method |
CN105095869A (en) * | 2015-07-24 | 2015-11-25 | 深圳市佳信捷技术股份有限公司 | Pedestrian detection method and apparatus |
CN105389586A (en) * | 2015-10-20 | 2016-03-09 | 浙江大学 | Method for automatically detecting integrity of shrimp body based on computer vision |
-
2016
- 2016-05-26 CN CN201610363846.0A patent/CN106096499A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102147869A (en) * | 2011-03-31 | 2011-08-10 | 上海交通大学 | Pedestrian detection method based on foreground analysis and pattern recognition |
CN103198332A (en) * | 2012-12-14 | 2013-07-10 | 华南理工大学 | Real-time robust far infrared vehicle-mounted pedestrian detection method |
CN103473788A (en) * | 2013-07-31 | 2013-12-25 | 中国电子科技集团公司第三十八研究所 | Indoor fire and flame detection method based on high-definition video images |
CN104866843A (en) * | 2015-06-05 | 2015-08-26 | 中国人民解放军国防科学技术大学 | Monitoring-video-oriented masked face detection method |
CN105095869A (en) * | 2015-07-24 | 2015-11-25 | 深圳市佳信捷技术股份有限公司 | Pedestrian detection method and apparatus |
CN105389586A (en) * | 2015-10-20 | 2016-03-09 | 浙江大学 | Method for automatically detecting integrity of shrimp body based on computer vision |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108322699A (en) * | 2018-01-23 | 2018-07-24 | 东南大学 | A method of based on video images detection arc-flash |
CN108322699B (en) * | 2018-01-23 | 2020-02-18 | 东南大学 | Method for detecting arc flash based on video image |
CN110363197A (en) * | 2019-06-22 | 2019-10-22 | 东北电力大学 | Based on the video area-of-interest exacting method for improving visual background extraction model |
CN112733666A (en) * | 2020-12-31 | 2021-04-30 | 湖北亿咖通科技有限公司 | Method, equipment and storage medium for collecting difficult images and training models |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10445567B2 (en) | Pedestrian head identification method and system | |
CN105787472B (en) | A kind of anomaly detection method based on the study of space-time laplacian eigenmaps | |
Chackravarthy et al. | Intelligent crime anomaly detection in smart cities using deep learning | |
CN110502965A (en) | A kind of construction safety helmet wearing monitoring method based on the estimation of computer vision human body attitude | |
CN102509075B (en) | Remnant object detection method and device | |
CN110188807A (en) | Tunnel pedestrian target detection method based on cascade super-resolution network and improvement Faster R-CNN | |
CN111881730A (en) | Wearing detection method for on-site safety helmet of thermal power plant | |
Benyang et al. | Safety helmet detection method based on YOLO v4 | |
CN105303191A (en) | Method and apparatus for counting pedestrians in foresight monitoring scene | |
CN106355154B (en) | Method for detecting frequent passing of people in surveillance video | |
CN105718857A (en) | Human body abnormal behavior detection method and system | |
CN103605983B (en) | Remnant detection and tracking method | |
CN105160317A (en) | Pedestrian gender identification method based on regional blocks | |
CN110378233B (en) | Double-branch anomaly detection method based on crowd behavior prior knowledge | |
CN103390151B (en) | Method for detecting human face and device | |
CN102930248A (en) | Crowd abnormal behavior detection method based on machine learning | |
CN106778650A (en) | Scene adaptive pedestrian detection method and system based on polymorphic type information fusion | |
CN112364778A (en) | Power plant safety behavior information automatic detection method based on deep learning | |
CN112489368A (en) | Intelligent falling identification and detection alarm method and system | |
CN112464701A (en) | Method for detecting whether people wear masks or not based on light weight characteristic fusion SSD | |
CN106096499A (en) | A kind of video image culminant star moon pattern detection method and system | |
CN109344886A (en) | Occlusion number plate distinguishing method based on convolutional neural network | |
CN105975952A (en) | Beard detection method and system in video image | |
CN104200218A (en) | Cross-view-angle action identification method and system based on time sequence information | |
CN109447064B (en) | A kind of duplicate rows License Plate Segmentation method and system based on CNN |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20161109 |
|
RJ01 | Rejection of invention patent application after publication |