CN108021949A - Crowded degree detection method, device, system and electronic equipment - Google Patents
Crowded degree detection method, device, system and electronic equipment Download PDFInfo
- Publication number
- CN108021949A CN108021949A CN201711451839.7A CN201711451839A CN108021949A CN 108021949 A CN108021949 A CN 108021949A CN 201711451839 A CN201711451839 A CN 201711451839A CN 108021949 A CN108021949 A CN 108021949A
- Authority
- CN
- China
- Prior art keywords
- image
- crowded
- crowding
- feature
- grader
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- 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
- G06V20/53—Recognition of crowd images, e.g. recognition of crowd congestion
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Probability & Statistics with Applications (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The present invention provides a kind of crowded degree detection method, device, system and electronic equipment, is related to image identification technical field, method includes：Obtain the first image to be detected, i.e. one kind under plurality of application scenes in acquired image；Extract the first crowded feature in the first image, i.e. the background area feature of the first image；First crowded feature is input in trained crowding detection grader in advance, obtains the classification results of the crowded degree of the first image；Crowding detects the training sample data gathered under plurality of application scenes of the grader by reaching certain threshold value to quantity and is trained to obtain.This method can be directed to different application scenarios, with reference to the background information of the image of collection, the degree of crowding under the scene is detected, reduces the dependence to high configuration hardware device, realize that quickly and accurately crowding identifies, meets under plurality of application scenes to the quick judgement demand of crowded degree.
Description
Technical field
The present invention relates to image identification technical field, more particularly, to a kind of crowded degree detection method, device, system
And electronic equipment.
Background technology
At present, the crowded degree algorithm of mainstream generally uses the mode identification method of the number number of people, first the background of the stream of people
Remove, then differentiate the number of the number of people in remaining image according to human head shapes, moved because the image in video exists
A series of consecutive variations of state, so by that can draw accurate personnel's number after processing.This method is to video equipment
Required precision it is higher, and since it is desired that cause the backstage of algorithm requirement to be counted to handle video flowing with complex algorithm
Calculation amount is bigger, so accurate number of people handling result could be obtained with fast speed by generally requiring specialized hardware.
For most application scenarios, it is not necessary to know in public place specific number how many, and only need to understand
One fuzzy concept, than such place, whether crowded and crowded degree be how high.Such as a bus stop or one
Waiting hall, administrative staff need to only know whether the crowded of waiting person reaches some degree, then easy to the different reply of startup
Pattern；Can and for example for public transport compartment, whether passenger is only concerned has seat, when not having seat, be squeezed into when in a hurry.
For the above situation, existing traditional crowded degree detection method, is analyzed by the processing to video data,
Determine number of people quantity in crowd, and then obtain the crowding under corresponding scene, there are recognition speed is slow, required hardware for this process
The problem of device configuration is high.
The content of the invention
In view of this, it is an object of the invention to provide a kind of crowded degree detection method, device, system and electronics to set
It is standby, different application scenarios can be directed to, with reference to the background information of the image of collection, detect the degree of crowding under the scene, are dropped
The low dependence to high configuration hardware device, realizes that quickly and accurately crowding identifies, meets to gather around crowd under plurality of application scenes
Squeeze the quick of degree and judge demand.
In a first aspect, an embodiment of the present invention provides a kind of crowded degree detection method, including：
Obtain the first image to be detected；First image is one kind in acquired image under plurality of application scenes；
Extract the first crowded feature in the first image；First crowded feature includes：The background of first image
Provincial characteristics；
First crowded feature is input in trained crowding detection grader in advance, obtains the first image
The classification results of crowded degree；
Wherein, crowding detection grader carries out classification based training by reaching the training sample data of certain threshold value to quantity
Obtain；Training sample data include acquired image under plurality of application scenes.
With reference to first aspect, an embodiment of the present invention provides the first possible embodiment of first aspect, wherein,
Before obtaining the first image to be detected, further include：
Obtain the second image to be trained；Second image includes quantity and reaches the plurality of application scenes of certain threshold value lower being adopted
The image of collection；
Extract the second crowded feature in the second image；Second crowded feature includes：The background of second image
Provincial characteristics；
By the second crowded feature, training sample data and verification sample data are integrated into；
Training sample data are trained by machine learning algorithm, obtain crowding detection grader.
With reference to first aspect, an embodiment of the present invention provides second of possible embodiment of first aspect, wherein,
After obtaining crowding detection grader, further include：
The model parameter of the first classification results and crowding detection grader is obtained, the first classification results are by training sample
The output result that data obtain after inputting to crowding detection grader；
Verification sample data input crowding is detected in grader, obtains the second classification results；
Judge whether the second recognition result is consistent with the first recognition result；
If it is not, then model parameter is adjusted.
With reference to first aspect, an embodiment of the present invention provides the third possible embodiment of first aspect, wherein,
Before extracting the second crowded feature in the second image, further include：
Pretreatment is normalized to the second image；
Judge that normalizing pretreated second image whether there is default defect；Default defect includes：Image masking, number
At least one of in being obscured according to mistake, image；
If it is, image is reacquired, and using the image of reacquisition as the second image.
With reference to first aspect, an embodiment of the present invention provides the 4th kind of possible embodiment of first aspect, wherein, will
First crowded feature is input in trained crowding detection grader in advance, obtains the crowded degree of the first image
Classification results, specifically include：
Determine the data format corresponding to crowding detection grader；
It is binary data format by the first crowded Feature Conversion if data format includes binary data format
Crowding detection grader is inputted afterwards, with the classification results of corresponding first image of generation.
With reference to first aspect, an embodiment of the present invention provides the 5th kind of possible embodiment of first aspect, wherein,
After obtaining the crowded degree classification results of the first image, further include：
By the first image and its corresponding classification results, training sample database is stored in as new training sample data
In.
Second aspect, the embodiment of the present invention provide a kind of crowded degree detection device, and device includes：
Image collection module, for obtaining the first image to be detected；First image under plurality of application scenes by gathering
Image in one kind；
Characteristic extracting module, for extracting the first crowded feature in the first image；First crowded feature bag
Include：The background area feature of first image；
Image classification module, grader is detected for the first crowded feature to be input to trained crowding in advance
In, obtain the classification results of the crowded degree of the first image；Wherein, crowding detects grader by reaching certain to quantity
The training sample data of threshold value carry out classification based training and obtain；Training sample data include the figure gathered under plurality of application scenes
Picture.
The third aspect, the embodiment of the present invention provide a kind of crowded degree detecting system, and system includes：Background server,
Terminal device and the crowded degree detection device as described in second aspect；
Background server, terminal device and crowded degree detection device communicate to connect two-by-two；
Background server, receives the crowded degree classification results to image transmitted by crowd's crowding detection device,
And crowded degree classification results are sent to terminal device.
Fourth aspect, the embodiment of the present invention provide a kind of electronic equipment, including memory, processor, are stored on memory
There is the computer program that can be run on a processor, processor realizes the side described in above-mentioned first aspect when performing computer program
The step of method.
5th aspect, the embodiment of the present invention also provide a kind of meter for the non-volatile program code that can perform with processor
Calculation machine computer-readable recording medium, program code make processor perform the method described in first aspect.
The embodiment of the present invention brings following beneficial effect：
In crowded degree detection method provided in an embodiment of the present invention, the first image to be detected is obtained first；The
One image is one kind in acquired image under plurality of application scenes；Wherein, plurality of application scenes includes but not limited to：Public transport
Compartment, rail train carriage, steamer cabin, various waiting rooms and have distinguishing mark thing (such as ground at bus station, ticket window
Ceramic tile line) room or ground.Then the first crowded feature in the first image is extracted；First crowded feature bag
Include：The background area feature of first image；The first crowded feature finally is input to trained crowding in advance to detect
In grader, the classification results of the crowded degree of the first image are obtained；Wherein, crowding detects grader by being reached to quantity
Training sample data progress classification based training to certain threshold value obtains；Training sample data include that plurality of application scenes is lower is gathered
Image.
The crowded degree detection method that the present embodiment is provided, beforehand through under large number of plurality of application scenes
Training sample data, carry out multi-angle feature extraction, and then training obtain crowding detection grader, then pass through the classification
Device, and the background area feature extracted from new image is combined, crowding identification is carried out to new image, this method is not required to
Determine the number of people quantity in image, you can obtain the crowding of current scene image, crowding detection speed can be greatly improved
Degree, and the dependence to high configuration hardware device is reduced, realize that quickly and accurately crowding identifies, meets a variety of applied fields
Judgement demand under scape to crowded degree.
Other features and advantages of the present invention will illustrate in the following description, also, partly become from specification
Obtain it is clear that or being understood by implementing the present invention.The purpose of the present invention and other advantages are in specification, claims
And specifically noted structure is realized and obtained in attached drawing.
To enable the above objects, features and advantages of the present invention to become apparent, preferred embodiment cited below particularly, and coordinate
Appended attached drawing, is described in detail below.
Brief description of the drawings
, below will be to specific in order to illustrate more clearly of the specific embodiment of the invention or technical solution of the prior art
Embodiment or attached drawing needed to be used in the description of the prior art are briefly described, it should be apparent that, in describing below
Attached drawing is some embodiments of the present invention, for those of ordinary skill in the art, before not making the creative labor
Put, other attached drawings can also be obtained according to these attached drawings.
Fig. 1 is a kind of flow chart for crowded degree detection method that the embodiment of the present invention one provides；
Fig. 2 is the flow chart for another crowded degree detection method that the embodiment of the present invention one provides；
Fig. 3 is the flow chart for another crowded degree detection method that the embodiment of the present invention one provides；
Fig. 4 is the flow chart for another crowded degree detection method that the embodiment of the present invention one provides；
Fig. 5 is the flow chart for another crowded degree detection method that the embodiment of the present invention one provides；
Fig. 6 is a kind of schematic diagram of crowded degree detection device provided by Embodiment 2 of the present invention；
Fig. 7 is a kind of schematic diagram for crowded degree detecting system that the embodiment of the present invention three provides；
Fig. 8 is the schematic diagram for a kind of electronic equipment that the embodiment of the present invention four provides.
Embodiment
To make the purpose, technical scheme and advantage of the embodiment of the present invention clearer, below in conjunction with attached drawing to the present invention
Technical solution be clearly and completely described, it is clear that described embodiment is part of the embodiment of the present invention, rather than
Whole embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art are not making creative work premise
Lower all other embodiments obtained, belong to the scope of protection of the invention.
Existing crowded degree detection method, is analyzed by the processing to video data, determines number of people number in crowd
Amount, and then obtains the crowding under corresponding scene, and there are the problem of recognition speed is slow, required hardware device configuration is high for this process.
Based on this, the embodiment of the present invention provides a kind of crowded degree detection method, device, system and electronic equipment, energy
Different application scenarios are enough directed to, with reference to the background information of the image of collection, the degree of crowding under the scene is detected, reduces to height
The dependence of hardware device is configured, realizes that quickly and accurately crowding identifies, meets under plurality of application scenes to crowded degree
Quickly judge demand.
For ease of understanding the present embodiment, a kind of crowded degree disclosed in the embodiment of the present invention is detected first
Method describes in detail.
Embodiment one：
An embodiment of the present invention provides a kind of crowded degree detection method, this method can be applied to a variety of specific crowds
In scene.Shown in Figure 1, this method comprises the following steps：
S101：Obtain the first image to be detected；First image is one in acquired image under plurality of application scenes
Kind.
In the embodiment of the present invention, application scenarios include but not limited to：It is public transport compartment, rail train carriage, steamer cabin, each
Kind waiting room, public transport station, ticket window and the room or ground that have distinguishing mark thing (such as floor ceramic tile line).Gathered around
Squeeze before degree identification, the image under certain application scenarios to be detected is obtained first, specifically, can be by being installed on the application
Camera in scene is shot, and obtains image to be detected, or can also be shot by other filming apparatus such as camera
Above-mentioned image to be detected is obtained, in order to which the image with hereinafter to be trained is distinguished, here with the first graphical representation.
S102：Extract the first crowded feature in the first image；First crowded feature includes：First image
Background area feature.
After image to be detected is got, feature extraction is carried out to the background area in the image, specifically, can tie
Fractional order theory and method, and the technology such as image enhancement, reconstruct, segmentation, target following are closed, background area is carried out to above-mentioned image
The feature extraction in domain, such as：For bus application scenarios, the feature of its background area includes but not limited to：Whether there is on seat
Contrast characteristic that people sits, passageway OR gate mouth have the contrast characteristic at unmanned station, the contrast characteristic for having nobody to help on handrail.For difference
Application scenarios, the background area feature extracted in its corresponding image is also different, with reference to its background area feature, utilizes
Advance trained grader is identified, and can quickly obtain the crowded degree under respective application scene.Specific identification
Process is as follows：
S103：First crowded feature is input in trained crowding detection grader in advance, obtains first
The classification results of the crowded degree of image.
Wherein, crowding detection grader carries out classification based training by reaching the training sample data of certain threshold value to quantity
Obtain；Training sample data include acquired image under plurality of application scenes.
In the present embodiment, crowding detection grader is by being carried out to the image gathered under substantial amounts of plurality of application scenes
What feature extraction and training obtained.It is thus possible to improve the accuracy of image recognition, that is, image to be detected can be obtained
Accurate crowding.
By the crowded feature of said extracted, that is, background area feature, above-mentioned trained crowding is input to
Detect in grader, it is possible to obtain the crowded degree classification results of first image, that is, crowded angle value, crowded angle value
It can be indicated, can also be represented in a manner of grade using percentage.Specifically, determine that crowding detection grader institute is right
The data format answered；It is binary number by the first crowded Feature Conversion if data format includes binary data format
According to crowding detection grader is inputted after form, with the classification results of corresponding first image of generation.
It should be noted that the grader of the training in the embodiment of the present invention can use various ways to carry out, such as, often
The a variety of neutral nets having based on deep learning seen, or can also be other moulds that can realize feature extraction and identification
Type, as a kind of preferred embodiment, the embodiment of the present invention is multi-faceted to image progress multi-angle using convolutional neural networks
Background characteristics extracts, and to realize, quickly crowded degree identifies.
Shown in Figure 2, the training process of above-mentioned grader is as follows, that is, obtains to be detected first in step S101
It is further comprising the steps of before image：
S201：Obtain the second image to be trained；Second image includes the plurality of application scenes that quantity reaches certain threshold value
Lower acquired image.
In the embodiment of the present invention, application scenarios include but not limited to：It is public transport compartment, rail train carriage, steamer cabin, each
Kind waiting room, public transport station, ticket window and the room or ground that have distinguishing mark thing (such as floor ceramic tile line).Gathered around
Squeeze before degree detection classifier training, the image under above-mentioned plurality of application scenes to be trained is obtained first, specifically, can lead to
Cross the camera being installed in each application scenarios to be shot, obtain image to be trained, in order to be detected with above
Image distinguished, here with the second graphical representation.
S202：Extract the second crowded feature in the second image；Second crowded feature includes：Second image
Background area feature.
Getting after the image trained, feature extraction is being carried out to the background area in the image, specifically, can tie
Fractional order theory and method, and the technology such as image enhancement, reconstruct, segmentation, target following are closed, background area is carried out to above-mentioned image
The feature extraction in domain, such as：For bus application scenarios, the feature of its background area includes but not limited to：Whether there is on seat
Contrast characteristic that people sits, passageway OR gate mouth have the contrast characteristic at unmanned station, the contrast characteristic for having nobody to help on handrail.For difference
Application scenarios, the background area feature extracted in its corresponding image is also different, with reference to different under plurality of application scenes
Background area feature, a variety of crowdings for different application scene can be trained and detect graders, for various applications
Crowding under scene quickly identifies.
S203：By the second crowded feature, training sample data and verification sample data are integrated into.
Extracting above-mentioned crowded feature, that is, after the background area feature under various application scenarios, its is whole
Two kinds of data are combined into, one kind is training sample data, and another kind is verification sample data, it is generally the case that training sample data
Ratio with verification sample data is 7：3.
Above-mentioned sample data is according to the needs of accuracy of identification, it may be necessary to the mark of image is manually participated in, with supervised learning
Mode reaches identification progress demand.
S204：Training sample data are trained by machine learning algorithm, obtain crowding detection grader.
After above-mentioned training sample data are obtained, based on deep learning algorithm (such as convolutional neural networks CNN technologies), from big
The above-mentioned training sample data learning of amount establishes model, obtains above-mentioned crowding detection grader.
It is shown in Figure 3, it is further comprising the steps of after crowding detection grader is obtained：
S301：The model parameter of the first classification results and crowding detection grader is obtained, the first classification results are will instruction
Practice the output result obtained after sample data is inputted to crowding detection grader.
S302：Verification sample data input crowding is detected in grader, obtains the second classification results.
S303：Judge whether the second recognition result is consistent with the first recognition result；If it is not, then model parameter is carried out
Adjustment.
After above-mentioned crowding detection grader is trained, the model parameter of the grader is carried out real scene test with
Verification, specifically, above-mentioned verification sample data is input in above-mentioned crowding detection grader, judge to export result whether with
The situation of input training sample data is consistent, and above-mentioned verification sample data includes：From different automobile types, the different degree of crowding, difference
The background area feature extracted in image under illumination condition, different camera angle scenes, for above-mentioned a variety of verification samples
Notebook data, judges whether its output result is consistent with the output result of training sample data, if it is different, then further according to survey
Test result carries out the adjustment of model parameter, the model of grader is modified with it is perfect.
It is shown in Figure 4, it is further comprising the steps of before the second crowded feature in extracting the second image：
S401：Pretreatment is normalized to the second image.
S402：Judge that normalizing pretreated second image whether there is default defect；Default defect includes：Image hides
Cover, at least one of during error in data, image obscure.
S403：If it is, image is reacquired, and using the image of reacquisition as the second image.
Shot, obtained after the image trained by the camera being installed in each application scenarios, treated to above-mentioned
Pretreatment is normalized in trained image, it is ensured that picture format standard, preliminary classification is carried out to multisource video.If to figure
After pretreatment, image still suffer from masking, error in data, it is fuzzy the defects of, then other view data are reacquired, if repeatedly
Still go wrong after pretreatment, control warning device is alarmed.
As a kind of preferred embodiment, after the crowded degree classification results of the first image are obtained, further include with
Lower step, it is shown in Figure 5：
S501：By the first image and its corresponding classification results, training sample is stored in as new training sample data
In database.
As crowding detects grader middle application in practical situations, constantly newly identified image and its recognition result are increased
It is added in training sample database, so that constantly grader or algorithm are updated and be optimized, adapts it to different
New opplication scene, improves the crowding recognition accuracy under different application scene.
Crowding described in the present embodiment is the quantizating index of application-oriented scene, various ways can be used to define.Such as
Crowding can be defined according to personnel's ride comfort degree.It is extremely crowded to be defined as 5 (in-cars by taking bus crowding as an example
People can not again be gone up), it is crowded to be defined as 4 (ridership that stand in passageway exceedes half, but does not arrive the extreme degree of crowding), moderately gather around
Squeeze and be defined as 3 (stand a small amount of passenger on passageway), be comfortably defined as 2 (having empty seat, but seating capacity exceedes half), it is empty
Spare time was defined as 1 (empty seat exceedes half).Also available units area carries number to define to crowding, such as unit area
(square metre) interior number is crowded more than X people；Or according to the handling capacity of passengers that place and vehicle are appraised and decided as standard definition.
A kind of specific application scenarios are set forth below, by taking public transport compartment as an example, are taken first from multiple compartment video cameras
The view data under different light conditions is obtained, then according to several degree of crowding indexes defined above, first manually mode mark
Note the degree of crowding under different automobile types, illumination condition；Then selection sort device (such as deep learning neutral net), is trained,
After the completion of training, it is ensured that resolution reaches requirement, such as not up to requires, then increases sample data or replace grader or adjustment point
Class device parameter etc.；The crowding that grader up to standard is used for real time video data is analyzed, in practice constantly by expanding sample
The mode such as this is trained this grader and is updated and is deployed in application scenarios in backstage evolution, reaches preferably recognition effect.
In the present embodiment, application scenarios include but not limited to：This crowding algorithm except for it is crowded degree identification in addition to,
It can be additionally used in the degree of crowding analysis of article, such as the container degree of crowding, the warehouse item degree of crowding.These application scenarios
It is characterized as：Identify target object shape size can be approximately fixed, place background possess distinguishing mark thing (as seat, handrail,
Furring tile line, markings etc.).
In crowded degree detection method provided in an embodiment of the present invention, the first image to be detected is obtained first；The
One image is one kind in acquired image under plurality of application scenes；Wherein, plurality of application scenes includes but not limited to：Public transport
Compartment, rail train carriage, steamer cabin, various waiting rooms and have distinguishing mark thing (such as ground at bus station, ticket window
Ceramic tile line) room or ground.Then the first crowded feature in the first image is extracted；First crowded feature bag
Include：The background area feature of first image；The first crowded feature finally is input to trained crowding in advance to detect
In grader, the classification results of the crowded degree of the first image are obtained；Wherein, crowding detects grader by being reached to quantity
Training sample data progress classification based training to certain threshold value obtains；Training sample data include that plurality of application scenes is lower is gathered
Image.
The crowded degree detection method that the present embodiment is provided, beforehand through under large number of plurality of application scenes
Training sample data, carry out multi-angle feature extraction, and then training obtain crowding detection grader, then pass through the classification
Device, and the background area feature extracted from new image is combined, crowding identification is carried out to new image, this method is not required to
Determine the number of people quantity in image, you can obtain the crowding of current scene image, crowding detection speed can be greatly improved
Degree, and the dependence to high configuration hardware device is reduced, realize that quickly and accurately crowding identifies, meets a variety of applied fields
Judgement demand under scape to crowded degree.
It should be noted that list some preprocess methods, algorithm name in the deep learning algorithm that the present embodiment uses
Title and control parameter, are not limited only to above method, algorithm and parameter in practical applications, all to reach with reference to scene characteristic factor
To the crowded method, algorithm and the parameter that are quickly judged spent in the protection domain of this patent.
Embodiment two：
The embodiment of the present invention provides a kind of crowded degree detection device, and shown in Figure 6, which includes：Image obtains
Modulus block 61, characteristic extracting module 62, image classification module 63.
Wherein, image collection module 61, for obtaining the first image to be detected；First image is under plurality of application scenes
One kind in acquired image；Characteristic extracting module 62, for extracting the first crowded feature in the first image；First
Crowded feature includes：The background area feature of first image；Image classification module 63, for by the first crowded feature
It is input in trained crowding detection grader in advance, obtains the classification results of the crowded degree of the first image；Wherein,
Crowding detects grader and is obtained by reaching the training sample data progress classification based training of certain threshold value to quantity；Training sample
Data include acquired image under plurality of application scenes.
In the crowded degree detection device that the embodiment of the present invention is provided, modules are detected with foregoing crowded degree
Method has identical technical characteristic, therefore, can equally realize above-mentioned function.The specific works mistake of modules in the present apparatus
Journey is referring to above method embodiment, and details are not described herein.
Embodiment three：
The embodiment of the present invention provides a kind of crowded degree detecting system, and shown in Figure 7, which includes：Backstage takes
Business device 71, terminal device 72 and the crowded degree detection device 73 as described in embodiment two.
Wherein, background server 71, terminal device 72 and crowded degree detection device 73 communicate to connect two-by-two；From the background
Server 71, receives the crowded degree classification results to image transmitted by crowd's crowding detection device 73, and by crowd
Crowding classification results are sent to terminal device 72.
Above-mentioned background server 71 can be the monitoring center in different application industry, and above-mentioned terminal device 72 can be each
The information promulgating platform of a industry, display, the handheld terminal etc. of relevant staff, specifically, by above-mentioned crowded degree
Detection device 73 is applied to different related application industries, can further realize multiple functions, such as, bus station staff
The current bus degree of crowding received according to handheld terminal carries out vehicle scheduling operation, for another example, railway station waiting hall
In by information promulgating platform, the crowded degree of current each Waiting Lounge is announced to the public, so that the public is according to above- mentioned information
Select the place waited.
Example IV：
The embodiment of the present invention provides a kind of electronic equipment, and shown in Figure 8, which includes：Processor 80, storage
Device 81, bus 82 and communication interface 83, the processor 80, communication interface 83 and memory 81 are connected by bus 82；Processing
Device 80 is used to perform the executable module stored in memory 81, such as computer program.When processor performs computer program
The step of realizing the method as described in embodiment of the method.
Wherein, memory 81 may include high-speed random access memory (RAM, RandomAccessMemory), also may be used
Nonvolatile memory (non-volatile memory), for example, at least a magnetic disk storage can be further included.Pass through at least one
A communication interface 83 (can be wired or wireless) realizes the communication link between the system network element and at least one other network element
Connect, internet, wide area network, local network, Metropolitan Area Network (MAN) etc. can be used.
Bus 82 can be isa bus, pci bus or eisa bus etc..The bus can be divided into address bus, data
Bus, controlling bus etc..Only represented for ease of representing, in Fig. 8 with a four-headed arrow, it is not intended that an only bus or
A type of bus.
Wherein, memory 81 is used for storage program, and the processor 80 performs the journey after execute instruction is received
Sequence, the method performed by the device for the flow definition that foregoing any embodiment of the embodiment of the present invention discloses can be applied to processor
In 80, or realized by processor 80.
Processor 80 is probably a kind of IC chip, has the disposal ability of signal.During realization, above-mentioned side
Each step of method can be completed by the integrated logic circuit of the hardware in processor 80 or the instruction of software form.Above-mentioned
Processor 80 can be general processor, including central processing unit (Central Processing Unit, abbreviation CPU), network
Processor (Network Processor, abbreviation NP) etc.；It can also be digital signal processor (Digital Signal
Processing, abbreviation DSP), application-specific integrated circuit (Application Specific Integrated Circuit, referred to as
ASIC), ready-made programmable gate array (Field-Programmable Gate Array, abbreviation FPGA) or other are programmable
Logical device, discrete gate or transistor logic, discrete hardware components.It can realize or perform in the embodiment of the present invention
Disclosed each method, step and logic diagram.General processor can be microprocessor or the processor can also be appointed
What conventional processor etc..The step of method with reference to disclosed in the embodiment of the present invention, can be embodied directly in hardware decoding processing
Device performs completion, or performs completion with the hardware in decoding processor and software module combination.Software module can be located at
Machine memory, flash memory, read-only storage, programmable read only memory or electrically erasable programmable memory, register etc. are originally
In the storage medium of field maturation.The storage medium is located at memory 81, and processor 80 reads the information in memory 81, with reference to
Its hardware completes the step of above method.
The computer program product for the crowded degree detection method that the embodiment of the present invention is provided, including store processing
The computer-readable recording medium for the non-volatile program code that device can perform, the instruction that said program code includes can be used for holding
Method described in row previous methods embodiment, specific implementation can be found in embodiment of the method, and details are not described herein.
It is apparent to those skilled in the art that for convenience and simplicity of description, the device of foregoing description
And the specific work process of electronic equipment, the corresponding process in preceding method embodiment is may be referred to, details are not described herein.
Flow chart and block diagram in attached drawing show multiple embodiment method and computer program products according to the present invention
Architectural framework in the cards, function and operation.At this point, each square frame in flow chart or block diagram can represent one
A part for module, program segment or code, a part for the module, program segment or code are used for realization comprising one or more
The executable instruction of defined logic function.It should also be noted that at some as the work(in the realization replaced, marked in square frame
Energy can also be with different from the order marked in attached drawing generation.For example, two continuous square frames can essentially be substantially parallel
Ground performs, they can also be performed in the opposite order sometimes, this is depending on involved function.It is also noted that block diagram
And/or the combination of each square frame and block diagram in flow chart and/or the square frame in flow chart, work(as defined in performing can be used
Can or the dedicated hardware based system of action realize, or reality can be carried out with the combination of specialized hardware and computer instruction
It is existing.
In the description of the present invention, it is necessary to explanation, term " " center ", " on ", " under ", "left", "right", " vertical ",
The orientation or position relationship of the instruction such as " level ", " interior ", " outer " be based on orientation shown in the drawings or position relationship, merely to
Easy to describe the present invention and simplify description, rather than instruction or imply signified device or element must have specific orientation,
With specific azimuth configuration and operation, therefore it is not considered as limiting the invention.In addition, term " first ", " second ",
" the 3rd " is only used for description purpose, and it is not intended that instruction or hint relative importance.
In several embodiments provided herein, it should be understood that disclosed systems, devices and methods, can be with
Realize by another way.Device embodiment described above is only schematical, for example, the division of the unit,
Only a kind of division of logic function, can there is other dividing mode when actually realizing, in another example, multiple units or component can
To combine or be desirably integrated into another system, or some features can be ignored, or not perform.It is another, it is shown or beg for
The mutual coupling, direct-coupling or communication connection of opinion can be by some communication interfaces, device or unit it is indirect
Coupling or communication connection, can be electrical, machinery or other forms.
The unit illustrated as separating component may or may not be physically separate, be shown as unit
The component shown may or may not be physical location, you can with positioned at a place, or can also be distributed to multiple
In network unit.Some or all of unit therein can be selected to realize the mesh of this embodiment scheme according to the actual needs
's.
In addition, each functional unit in each embodiment of the present invention can be integrated in a processing unit, can also
That unit is individually physically present, can also two or more units integrate in a unit.
If the function is realized in the form of SFU software functional unit and is used as independent production marketing or in use, can be with
It is stored in the non-volatile computer read/write memory medium that a processor can perform.Based on such understanding, the present invention
The part that substantially contributes in other words to the prior art of technical solution or the part of the technical solution can be with software
The form of product embodies, which is stored in a storage medium, including some instructions use so that
One computer equipment (can be personal computer, server, or network equipment etc.) performs each embodiment institute of the present invention
State all or part of step of method.And foregoing storage medium includes：USB flash disk, mobile hard disk, read-only storage (ROM, Read-
Only Memory), random access memory (RAM, Random Access Memory), magnetic disc or CD etc. are various can be with
The medium of store program codes.
Finally it should be noted that：Embodiment described above, is only the embodiment of the present invention, to illustrate the present invention
Technical solution, rather than its limitations, protection scope of the present invention is not limited thereto, although with reference to the foregoing embodiments to this hair
It is bright to be described in detail, it will be understood by those of ordinary skill in the art that：Any one skilled in the art
The invention discloses technical scope in, it can still modify the technical solution described in previous embodiment or can be light
It is readily conceivable that change, or equivalent substitution is carried out to which part technical characteristic；And these modifications, change or replacement, do not make
The essence of appropriate technical solution departs from the spirit and scope of technical solution of the embodiment of the present invention, should all cover the protection in the present invention
Within the scope of.Therefore, protection scope of the present invention answers the scope of the claims of being subject to.
Claims (10)
- A kind of 1. crowded degree detection method, it is characterised in that including：Obtain the first image to be detected；Described first image is one kind in acquired image under plurality of application scenes；Extract the first crowded feature in described first image；The first crowded feature includes：First figure The background area feature of picture；Described first crowded feature is input in trained crowding detection grader in advance, obtains first figure The classification results of the crowded degree of picture；Wherein, the crowding detection grader carries out classification based training by reaching the training sample data of certain threshold value to quantity Obtain；The training sample data include acquired image under plurality of application scenes.
- 2. according to the method described in claim 1, it is characterized in that, before the acquisition the first image to be detected, also wrap Include：Obtain the second image to be trained；Second image includes quantity and reaches the plurality of application scenes of certain threshold value lower being adopted The image of collection；Extract the second crowded feature in second image；The second crowded feature includes：Second figure The background area feature of picture；By the described second crowded feature, training sample data and verification sample data are integrated into；The training sample data are trained by machine learning algorithm, obtain the crowding detection grader.
- 3. according to the method described in claim 2, it is characterized in that, after the crowding detection grader is obtained, also wrap Include：The model parameter of the first classification results and crowding detection grader is obtained, first classification results are by described in The output result that training sample data obtain after inputting to crowding detection grader；The verification sample data is inputted in the crowding detection grader, obtains the second classification results；Judge whether second recognition result is consistent with first recognition result；If it is not, then the model parameter is adjusted.
- 4. according to the method described in claim 2, it is characterized in that, the second crowd in extraction second image gathers around Squeeze before feature, further include：Pretreatment is normalized to second image；Judge that pretreated second image of normalization whether there is default defect；The default defect includes：Image hides Cover, at least one of during error in data, image obscure；If it is, image is reacquired, and using the image of the reacquisition as second image.
- 5. according to the method described in claim 1, it is characterized in that, described be input to the described first crowded feature in advance In trained crowding detection grader, the crowded degree classification results of described first image are obtained, are specifically included：Determine the data format corresponding to the crowding detection grader；It is binary data by the described first crowded Feature Conversion if the data format includes binary data format The crowding detection grader is inputted after form, with the classification results of the corresponding described first image of generation.
- 6. according to claim 1-5 any one of them methods, it is characterised in that obtain the crowd of described first image described After crowding classification results, further include：By described first image and its corresponding classification results, training sample database is stored in as new training sample data In.
- 7. a kind of crowded degree detection device, it is characterised in that described device includes：Image collection module, for obtaining the first image to be detected；Described first image under plurality of application scenes by gathering Image in one kind；Characteristic extracting module, for extracting the first crowded feature in described first image；The first crowded spy Sign includes：The background area feature of described first image；Image classification module, grader is detected for the described first crowded feature to be input to trained crowding in advance In, obtain the classification results of the crowded degree of described first image；Wherein, the crowding detection grader passes through to quantity The training sample data progress classification based training for reaching certain threshold value obtains；The training sample data are included under plurality of application scenes Acquired image.
- A kind of 8. crowded degree detecting system, it is characterised in that the system comprises：Background server, terminal device and such as Crowded degree detection device described in the claims 7；The background server, the terminal device and the crowded degree detection device communicate to connect two-by-two；The background server, receives the crowded degree classification knot to image transmitted by the crowded degree detection device Fruit, and the crowded degree classification results are sent to the terminal device.
- 9. a kind of electronic equipment, including memory, processor, it is stored with what can be run on the processor on the memory Computer program, it is characterised in that the processor realizes that the claims 1 to 6 are any when performing the computer program Described in method the step of.
- 10. a kind of computer-readable medium for the non-volatile program code that can perform with processor, it is characterised in that described Program code makes the processor perform claim 1 to 6 any one of them method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711451839.7A CN108021949B (en) | 2017-12-27 | 2017-12-27 | Crowd crowding degree detection method, device and system and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711451839.7A CN108021949B (en) | 2017-12-27 | 2017-12-27 | Crowd crowding degree detection method, device and system and electronic equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108021949A true CN108021949A (en) | 2018-05-11 |
| CN108021949B CN108021949B (en) | 2021-03-05 |
Family
ID=62071924
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711451839.7A Active CN108021949B (en) | 2017-12-27 | 2017-12-27 | Crowd crowding degree detection method, device and system and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108021949B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110705391A (en) * | 2019-09-17 | 2020-01-17 | 四川大学锦城学院 | Seat distribution system based on human body image recognition method |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101739569A (en) * | 2009-12-17 | 2010-06-16 | 北京中星微电子有限公司 | Crowd density estimation method, device and monitoring system |
| CN102044073A (en) * | 2009-10-09 | 2011-05-04 | 汉王科技股份有限公司 | Method and system for judging crowd density in image |
| US8195598B2 (en) * | 2007-11-16 | 2012-06-05 | Agilence, Inc. | Method of and system for hierarchical human/crowd behavior detection |
| US8812344B1 (en) * | 2009-06-29 | 2014-08-19 | Videomining Corporation | Method and system for determining the impact of crowding on retail performance |
| CN104123569A (en) * | 2014-07-21 | 2014-10-29 | 中国人民解放军63908部队 | Video person number information statistics method based on supervised learning |
| CN104268898A (en) * | 2014-09-15 | 2015-01-07 | 郑州天迈科技股份有限公司 | Method for detecting density of passengers in bus on basis of image analysis |
| CN104951804A (en) * | 2015-06-25 | 2015-09-30 | 西北工业大学 | Crowded scene classification method based on unsupervised feature learning and density estimation |
| CN106203331A (en) * | 2016-07-08 | 2016-12-07 | 苏州平江历史街区保护整治有限责任公司 | A kind of crowd density evaluation method based on convolutional neural networks |
| CN107145821A (en) * | 2017-03-23 | 2017-09-08 | 华南农业大学 | A kind of crowd density detection method and system based on deep learning |
| CN107229894A (en) * | 2016-03-24 | 2017-10-03 | 上海宝信软件股份有限公司 | Intelligent video monitoring method and system based on computer vision analysis technology |
| CN107341480A (en) * | 2017-07-12 | 2017-11-10 | 中国电子科技集团公司第二十八研究所 | A kind of crowd massing detection method of modified PCCNN neural network models |
| CN107483887A (en) * | 2017-08-11 | 2017-12-15 | 中国地质大学（武汉） | The early-warning detection method of emergency case in a kind of smart city video monitoring |
-
2017
- 2017-12-27 CN CN201711451839.7A patent/CN108021949B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8195598B2 (en) * | 2007-11-16 | 2012-06-05 | Agilence, Inc. | Method of and system for hierarchical human/crowd behavior detection |
| US8812344B1 (en) * | 2009-06-29 | 2014-08-19 | Videomining Corporation | Method and system for determining the impact of crowding on retail performance |
| CN102044073A (en) * | 2009-10-09 | 2011-05-04 | 汉王科技股份有限公司 | Method and system for judging crowd density in image |
| CN101739569A (en) * | 2009-12-17 | 2010-06-16 | 北京中星微电子有限公司 | Crowd density estimation method, device and monitoring system |
| CN104123569A (en) * | 2014-07-21 | 2014-10-29 | 中国人民解放军63908部队 | Video person number information statistics method based on supervised learning |
| CN104268898A (en) * | 2014-09-15 | 2015-01-07 | 郑州天迈科技股份有限公司 | Method for detecting density of passengers in bus on basis of image analysis |
| CN104951804A (en) * | 2015-06-25 | 2015-09-30 | 西北工业大学 | Crowded scene classification method based on unsupervised feature learning and density estimation |
| CN107229894A (en) * | 2016-03-24 | 2017-10-03 | 上海宝信软件股份有限公司 | Intelligent video monitoring method and system based on computer vision analysis technology |
| CN106203331A (en) * | 2016-07-08 | 2016-12-07 | 苏州平江历史街区保护整治有限责任公司 | A kind of crowd density evaluation method based on convolutional neural networks |
| CN107145821A (en) * | 2017-03-23 | 2017-09-08 | 华南农业大学 | A kind of crowd density detection method and system based on deep learning |
| CN107341480A (en) * | 2017-07-12 | 2017-11-10 | 中国电子科技集团公司第二十八研究所 | A kind of crowd massing detection method of modified PCCNN neural network models |
| CN107483887A (en) * | 2017-08-11 | 2017-12-15 | 中国地质大学（武汉） | The early-warning detection method of emergency case in a kind of smart city video monitoring |
Non-Patent Citations (3)
| Title |
|---|
| YA-LI HOU等: "People Counting and Human Detection in a Challenging Situation", 《IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS - PART A: SYSTEMS AND HUMANS》 * |
| 覃勋辉等: "多种人群密度场景下的人群计数", 《中国图象图形学报》 * |
| 陈鸿: "重点视频监控区域的人群密度估计改进算法的研究与实现", 《中国硕士学位论文全文数据库信息科技辑》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110705391A (en) * | 2019-09-17 | 2020-01-17 | 四川大学锦城学院 | Seat distribution system based on human body image recognition method |
| CN110705391B (en) * | 2019-09-17 | 2023-09-19 | 四川大学锦城学院 | Seat distribution system based on human body image recognition method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108021949B (en) | 2021-03-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109978893B (en) | Training method, device, equipment and storage medium of image semantic segmentation network | |
| CN105574550A (en) | Vehicle identification method and device | |
| CN105303193B (en) | A kind of passenger number statistical system based on single-frame images processing | |
| CN110188807A (en) | Tunnel pedestrian target detection method based on cascade super-resolution network and improvement Faster R-CNN | |
| CN108171162A (en) | Crowded degree detection method, apparatus and system | |
| CN108537117A (en) | A kind of occupant detection method and system based on deep learning | |
| CN109241871A (en) | A kind of public domain stream of people's tracking based on video data | |
| CN108073898B (en) | Method, device and equipment for identifying human head area | |
| CN104239386A (en) | Method and system for prioritizion of facial recognition matches | |
| CN112613569B (en) | Image recognition method, training method and device for image classification model | |
| CN107203775A (en) | A kind of method of image classification, device and equipment | |
| CN111274926B (en) | Image data screening method, device, computer equipment and storage medium | |
| CN111813997A (en) | Intrusion analysis method, device, equipment and storage medium | |
| CN107871314A (en) | A kind of sensitive image discrimination method and device | |
| CN108229256A (en) | A kind of road construction detection method and device | |
| CN110163109A (en) | A kind of lane line mask method and device | |
| CN113642474A (en) | Hazardous area personnel monitoring method based on YOLOV5 | |
| CN113450573A (en) | Traffic monitoring method and traffic monitoring system based on unmanned aerial vehicle image recognition | |
| CN106529470A (en) | Gesture recognition method based on multistage depth convolution neural network | |
| Zhang et al. | Image-based approach for parking-spot detection with occlusion handling | |
| CN108154199B (en) | High-precision rapid single-class target detection method based on deep learning | |
| CN106548195A (en) | A kind of object detection method based on modified model HOG ULBP feature operators | |
| Al Nasim et al. | An automated approach for the recognition of bengali license plates | |
| CN113610770A (en) | License plate recognition method, device and equipment | |
| CN109583396A (en) | A kind of region prevention method, system and terminal based on CNN two stages human testing |
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 |