CN109902551A - The real-time stream of people's statistical method and device of open scene - Google Patents
The real-time stream of people's statistical method and device of open scene Download PDFInfo
- Publication number
- CN109902551A CN109902551A CN201811331786.XA CN201811331786A CN109902551A CN 109902551 A CN109902551 A CN 109902551A CN 201811331786 A CN201811331786 A CN 201811331786A CN 109902551 A CN109902551 A CN 109902551A
- Authority
- CN
- China
- Prior art keywords
- pedestrian
- frame
- people
- statistical regions
- currently active
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Image Analysis (AREA)
Abstract
This specification provides a kind of real-time stream of people's statistical method of open scene, comprising: extracts the currently active frame from the live video stream for shooting the open scene;The video flowing is shot by being placed in the camera above open scene;Detect each pedestrian in the currently active frame;Using pedestrian's weight recognizer, identical pedestrian at least one valid frame in the currently active frame and before is identified;According to pedestrian's weight recognizer as a result, calculating stream of people's quantity Jing Guo the open scene.
Description
Technical field
This specification is related to technical field of data processing, more particularly to the real-time stream of people's statistical method and dress of open scene
It sets.
Background technique
Stream of people's statistics plays effect in a variety of commercial applications, for example, can count inside market by different periods
The information such as number, the crowd flow direction of distribution, hold activity for market and provide reference;It can obtain outside market on pavement
Whether appropriate crowd's amount of flow statistical information is conducive to assessment market addressing;Etc..
The statistics of real-time crowd's flow has even more important meaning, can be timely to the statistics of monitoring area real-time traffic
The number and crowd's data on flows at scene are obtained, is conducive to the more efficient organization work of management unit, provides number for science decision
According to support.Real-time stream of people statistics based on video usually requires that camera has visual angle vertically downward, and in many applied fields
Conjunction is all difficult to meet this condition.
Summary of the invention
In view of this, this specification provides a kind of real-time stream of people's statistical method of open scene, comprising:
The currently active frame is extracted from the live video stream for shooting the open scene;The video flowing is by being placed in opening
Camera shooting above scene;
Detect each pedestrian in the currently active frame;
Using pedestrian's weight recognizer, identical row at least one valid frame in the currently active frame and before is identified
People;
According to pedestrian's weight recognizer as a result, calculating stream of people's quantity Jing Guo the open scene.
This specification additionally provides a kind of real-time stream of people's statistic device of open scene, comprising:
Valid frame extraction unit, for extracting the currently active frame from the live video stream for shooting the open scene;Institute
It states video flowing and is shot by being placed in the camera above open scene;
Pedestrian detection unit, for detecting each pedestrian in the currently active frame;
Pedestrian's weight recognition unit, for identifying at least one in the currently active frame and before using pedestrian's weight recognizer
Identical pedestrian in a valid frame;
Flow rate calculation unit, for according to pedestrian's weight recognizer as a result, calculating the stream of people Jing Guo the open scene
Quantity.
A kind of computer equipment that this specification provides, comprising: memory and processor;Being stored on the memory can
The computer program run by processor;When the processor runs the computer program, above-mentioned web access realization side is executed
Step described in method.
A kind of computer readable storage medium that this specification provides, is stored thereon with computer program, the computer
When program is run by processor, step described in the implementation method that the above-mentioned web applied in CDN node is accessed is executed.
By above technical scheme as it can be seen that in the embodiment of this specification, shot above open scene based on camera
Video flowing is identified in the currently active frame by detecting the pedestrian in the currently active frame, and using pedestrian's weight recognizer
Pedestrian and identical pedestrian in valid frame before, obtain stream of people's quantity by open scene, thus solve needs vertically to
Under camera visual angle carry out the limitation of stream of people's statistics, and the timeliness of detection can be provided while reducing operation cost
Property, and the accuracy of detection can be improved.
Detailed description of the invention
Fig. 1 is a kind of open scene in this specification embodiment, the exemplary diagram of camera angle and statistical regions;
Fig. 2 is a kind of flow chart of real-time stream of people's statistical method of open scene in this specification embodiment;
Fig. 3 is the structural representation of the stream of people's quantity statistics software run on embedded board in this specification application example
Figure;
Fig. 4 is a kind of hardware structure diagram for running the equipment of this specification embodiment;
Fig. 5 is a kind of building-block of logic of real-time stream of people's statistic device of open scene in this specification embodiment.
Specific embodiment
The embodiment of this specification proposes a kind of real-time stream of people's statistical method of new opening scene, from being placed in open field
Above scape camera shooting live video stream in extract the currently active frame, detect the pedestrian in the currently active frame and identify with
Whether the pedestrian in valid frame is identical before, to count stream of people's quantity by open scene.The embodiment of this specification is not necessarily to
Camera visual angle vertically downward can be suitable for most of application, and computational load is low, when with good
Also there is very high accuracy under the premise of effect property.
The embodiment of this specification may operate in any equipment with calculating and storage capacity, such as mobile phone, plate
The equipment such as computer, PC (Personal Computer, PC), notebook, server;Can also by operate in two or
The logical node of more than two equipment realizes the various functions in this specification embodiment.
The embodiment of this specification is used to count in open scene by real-time crowd's flow.Held is in opening
The top of scene shoots the crowd in open scene with angle obliquely, generates live video stream.In some reality
In, need to count is stream of people's quantity by some scheduled statistical regions in open scene.Statistical regions are open
Statistical regions can be completely covered in a given area in scene, the coverage of camera.A kind of open scene, Statistical Area
The example of domain and camera angle is as shown in Figure 1, be wherein statistical regions inside solid box.
In the embodiment of this specification, the process of real-time stream of people's statistical method of open scene is as shown in Figure 2.
Step 210, the currently active frame is extracted from the live video stream for shooting open scene.
Camera above open scene is mounted on by lasting output with the video flowing of viewing angles obliquely, video flowing by
Continuous frame frame image construction.It can be continual to meet each of the condition from video flowing based on certain condition
Frame image zooming-out comes out and is used as valid frame, carries out the statistics of crowd's flow by continuously recognizing the pedestrian in valid frame.It will
The last one valid frame extracted from video flowing is as the currently active frame.
The condition for extracting valid frame can requirement according to practical application scene to statistical time precision, operation the present embodiment
The processing capacity of equipment etc. because being usually arranged, for example, can will be with the one of upper one effective frame period N (N is natural number) frame
As next valid frame, can also extract a frame from every M (M is greater than 1 natural number) a successive frame and be used as has frame image
Imitate frame.
Step 220, each pedestrian in the currently active frame is detected.
After extracting the currently active frame, judge to whether there is in the currently active frame by the algorithm of target detection of deep learning
Human body, and if so, position each pedestrian position and the pedestrian occupied by partial image region.
In this specification embodiment without limitation to the algorithm of target detection of use, Faster R-CNN can such as be used
(Faster Regions with Convolutional Neural Network features, using convolutional neural networks spy
The fast target region recognition of sign), SSD (Single Shot MultiBox Detector, the prediction of single target more frames) etc..
Not only lower calculation amount is being required, but also is requiring that YOLO (You Only can be used in the application scenarios of Detection accuracy
Live Once) algorithm of target detection, the image range and location information of each pedestrian are extracted from the currently active frame, often may be used
To reach better effect.
Step 230, using pedestrian weight recognizer, identify in the currently active frame with phase at least one valid frame before
Same pedestrian.
When some pedestrian passes through from open scene, can be photographed in multiple valid frames.When carrying out stream of people's statistics, need
Identical pedestrian is found out in each valid frame, avoided repeatedly counting the same person, can just be obtained accurate data.
Pedestrian identifies that (Person ReID, Person Re-identification) can utilize computer vision skill again
Art judges in image with the presence or absence of specific pedestrian, can be used to carry out the same camera or personage's tracking across camera.
In the embodiment of this specification, judged in all pedestrians detected in the currently active frame using pedestrian's weight recognizer,
Which is the pedestrian before being already present in valid frame, which is the emerging pedestrian in the currently active frame.
N number of valid frame before the currently active frame whether being appeared in by searching for some pedestrian in the currently active frame
In, to judge whether the pedestrian is emerging pedestrian.Due in this specification embodiment camera with angle diagonally downward
Shoot open scene, when pedestrian is more intensive, it is possible that some pedestrian some valid frame or certain it is several it is continuous effectively
The situation blocked by other people without being detected in frame, choosing biggish N value can inciting somebody to action to avoid mistake in this case
Pedestrian's repeat count, but biggish N value also brings along bigger computational load.It, can be according to opening in practical application scene
Pedestrian's concentration of scene, the interval time of adjacent valid frame, the factors such as the processing capacity of equipment that run the present embodiment, come
Select N value appropriate.
In one implementation, in available the currently active frame each pedestrian external appearance characteristic and position feature, then
According to the external appearance characteristic and position feature of pedestrian, determine some pedestrian in the currently active frame whether be before N number of valid frame
In already present pedestrian, if it is not, generating new character recognition and label marks the pedestrian;If so, with existing character recognition and label
Mark the pedestrian.
Specifically, to algorithm of target detection output each pedestrian position and the pedestrian occupied by part figure
As region, generating the position feature of the pedestrian by the position of each pedestrian, (partial region that such as pedestrian occupies is in Picture Coordinate
Coordinate in system), the image of the partial region as occupied by the pedestrian generates the external appearance characteristic of the pedestrian (such as clothes color, clothing
Take texture, handbag, knapsack, cap etc.).Using the position feature and external appearance characteristic of some pedestrian in the currently active frame, at it
It searches whether to have existed the pedestrian in preceding N number of valid frame, be used in combination if it does not, generating new character recognition and label for the pedestrian
The character recognition and label of generation marks the pedestrian, and character recognition and label is used to uniquely represent a pedestrian, can be call number, character string
Deng without limitation.If having existed the pedestrian, which has had the character recognition and label of oneself, existing before continuing to use
Character recognition and label marks the pedestrian in the currently active frame.All pedestrians detected in the currently active frame are executed one by one
Search procedure is stated, until each pedestrian detected is marked with character recognition and label.
Can be according to the needs of practical application scene, whether the pedestrian to select to identify in different valid frames is the same person
The algorithm of Shi Caiyong, without limitation.For example, Hungary Algorithm can be used, according to the external appearance characteristic and position feature of pedestrian
The currently active frame is carried out to match with the pedestrian in valid frame before.
Step 240, according to pedestrian's weight recognizer as a result, calculating stream of people's quantity Jing Guo the open scene.
The concrete mode for counting open scene stream of people quantity can be determined according to the needs of practical application, this specification
Embodiment is without limitation.(in example below, statistical time section is that the accumulative period is carried out to the stream of people) illustrated below:
First example: emerging pedestrian in each valid frame in statistical time section can be added up, will
Accumulation result is as stream of people's quantity in statistical time section.If some pedestrian in the currently active frame top n valid frame therewith
In pedestrian it is different, then the pedestrian be the currently active frame in emerging pedestrian;It is new in all valid frames in statistical time section
Pedestrian's quantity summation of appearance, can be considered as pedestrian's total quantity of statistical time section.
Second example: in the implementation that each pedestrian is marked with character recognition and label, when available statistics
Between pedestrian emerging in each valid frame and the pedestrian left in section.Wherein, the pedestrian left can occur from some
The pedestrian not occurred in N number of valid frame in adjacent valid frame before valid frame and in the valid frame and later, with
The pedestrian is avoided to be blocked and statistic bias caused by temporary extinction by other pedestrians.Stream of people's quantity in statistical time section, can be with
It is emerging pedestrian's quantity summation in all valid frames in statistical time section, is also possible to all valid frames in statistical time section
In pedestrian's quantity summation for leaving, can also be result (such as emerging pedestrian's number to perform mathematical calculations to above-mentioned two value
The mean value of amount summation and the pedestrian's quantity summation left).
In second example, the time span that each pedestrian is in open scene can also be counted, such as by some row
Elapsed time between the valid frame that people leaves and the emerging valid frame of the pedestrian is in open scene as the pedestrian
Duration.
Need to count be by the application of stream of people's quantity of some scheduled statistical regions in open scene, it is right
Some valid frame, if some pedestrian N number of having of being appeared in the valid frame in statistical regions and before the valid frame
The pedestrian does not appear in statistical regions in effect frame, then (enters the pedestrian as pedestrian emerging in the valid frame
The pedestrian of statistical regions);If some pedestrian appear in some valid frame before adjacent valid frame statistical regions and
It is to appear in statistical regions in the valid frame and subsequent N number of valid frame, then using the pedestrian as leaving in the valid frame
Pedestrian (leaves the pedestrian of statistical regions).
It, can be using stringenter judgment criteria, to obtain to statistical regions are entered and left in above-mentioned application
More accurate stream of people's quantity.For example, can some pedestrian appeared in N number of valid frame before outside statistical regions without
When once appearing in statistical regions and being appeared in statistical regions in the currently active frame, it is believed that the pedestrian is the currently active
Frame enters statistical regions;System is never appeared in when some pedestrian once appears in statistical regions in N number of valid frame before
When outside meter region and being appeared in outside statistical regions in the currently active frame, it is believed that the pedestrian leaves statistics in the currently active frame
Region.Stream of people's quantity by statistical regions can be calculated according to the number for entering statistical regions, can also be united according to leaving
The number in region is counted to calculate, can also be calculated according to the number for entering statistical regions with the number of statistical regions is left.
As it can be seen that in the embodiment of this specification, from the live video stream for the camera shooting being placed in above open scene
It is middle to extract the currently active frame, it is identified currently by detecting the pedestrian in the currently active frame, and using pedestrian's weight recognizer
Pedestrian in valid frame and identical pedestrian in valid frame before, obtain stream of people's quantity by open scene, not only without hanging down
Straight downward camera visual angle can be suitable for most of application, and computational load is low, can provide detection
Timeliness, and the accuracy of detection can be improved.
Since computational load is lower, the method for this specification embodiment is adapted to operate on embedded board, and right
The hardware environment of embedded board does not specially require.The embedded board of operation this specification embodiment may be mounted at
Near video camera, stream of people's data that real-time statistics go out are sent to the clothes for being responsible for acquisition data on flows by the communication module of itself
Business device can obtain essence under conditions of not invading the privacy of pedestrian without uploading the video or image of video camera shooting
True stream of people's statistical data.
It is above-mentioned that this specification specific embodiment is described.Other embodiments are in the scope of the appended claims
It is interior.In some cases, the movement recorded in detail in the claims or step can be come according to the sequence being different from embodiment
It executes and desired result still may be implemented.In addition, process depicted in the drawing not necessarily require show it is specific suitable
Sequence or consecutive order are just able to achieve desired result.In some embodiments, multitasking and parallel processing be also can
With or may be advantageous.
In an application example of this specification, need to stream of people's quantity by specific region in indoor open scene
It is counted.RGB (Red Green Blue, RGB) camera is mounted on higher place on wall, can be with obliquely
Visual angle indoor open scene is shot, statistical regions (i.e. specific region) are located at the center portion of coverage, with
The boundary of coverage is separated by a distance.
Stream of people's quantity statistics are carried out by the program operated on NVIDIA (tall and handsome to reach) embedded board, embedded development
Plate is mounted near camera, is included thereon communication unit, can be connect with camera by near radio mode, from camera shooting
Head obtains the video data of its shooting.Stream of people's data that embedded board can also be obtained statistics by communication unit upload
To scheduled server.
The structure of the stream of people's quantity statistics software run on embedded board is as shown in Figure 3.
RGB camera persistently shoots the image of open scene with 25 frames/second speed, forms video flowing.Embedded development
Plate extracts a frame RGB image every the frame of fixed quantity from the video flowing of shooting, as the currently active frame.
Stream of people's quantity statistics software uses YOLO algorithm of target detection, identifies each pedestrian in the currently active frame, really
Partial region occupied by fixed position coordinates (a kind of position feature) and each pedestrian of each pedestrian in image coordinate system
Image.
For each pedestrian in the currently active frame, pedestrian's weight recognizer occupies in the image in region from the pedestrian to be extracted
The external appearance characteristic of the pedestrian, using the external appearance characteristic of the pedestrian and position coordinates as foundation, pedestrian's weight recognizer uses Hungary
Algorithm judges whether the pedestrian match with each pedestrian in 3 valid frames before the currently active frame, identifies the row
Whether people occurred in 3 valid frames before, if do not occurred, generates new character recognition and label for the pedestrian and comes only
One represents the pedestrian, and marks the pedestrian with new character recognition and label;If once occurred, the existing personage of the pedestrian is used
Mark is to mark the pedestrian.
Character recognition and label and the inspection of YOLO target based on pedestrian's weight recognizer to pedestrian's label each in the currently active frame
The position coordinates of each pedestrian of method of determining and calculating output, if some pedestrian appears in statistical regions in 3 valid frames before
When appearing in statistical regions and being appeared in statistical regions in the currently active frame outside and never, which is labeled as
Enter the pedestrian of statistical regions in the currently active frame;If some pedestrian once appears in Statistical Area in 3 valid frames before
When never appearing in domain outside statistical regions and being appeared in outside statistical regions in the currently active frame, which is marked
For the pedestrian for leaving statistical regions in the currently active frame.
Stream of people's quantity statistics software is using scheduled statistical time section as the period, cumulative all valid frames in one cycle
In leave pedestrian's quantity of statistical regions as stream of people's quantity, and single pedestrian is left to valid frame and the entrance of statistical regions
Time interval between the valid frame of statistical regions is as the pedestrian in the stay time of statistical regions, and add up all these pedestrians
Stay time.After a cycle, stream of people's quantity statistics software sends stream of people's number in the period to scheduled server
The stop total duration of amount and these stream of peoples.
Corresponding with the realization of above-mentioned process, the embodiment of this specification additionally provides a kind of real-time stream of people statistics of open scene
Device.The device can also be realized by software realization by way of hardware or software and hardware combining.It is implemented in software
For, it is CPU (Central Process Unit, the central processing by place equipment as the device on logical meaning
Device) by corresponding computer program instructions be read into memory operation formed.For hardware view, in addition to shown in Fig. 4
Except CPU, memory and memory, the equipment where real-time stream of people's statistic device of open scene is also typically included for carrying out
Other hardware such as the chip of wireless signal transmitting-receiving, and/or for realizing other hardware such as board of network communicating function.
Fig. 5 show a kind of real-time stream of people's statistic device of open scene of this specification embodiment offer, including effective
Frame extraction unit, pedestrian detection unit, pedestrian's weight recognition unit and flow rate calculation unit, in which: valid frame extraction unit is used for
The currently active frame is extracted from the live video stream for shooting the open scene;The video flowing is by being placed in above open scene
Camera shooting;Pedestrian detection unit is used to detect each pedestrian in the currently active frame;Pedestrian's weight recognition unit is for adopting
With pedestrian's weight recognizer, identical pedestrian at least one valid frame in the currently active frame and before is identified;Flow rate calculation
Unit is used for according to pedestrian's weight recognizer as a result, calculating stream of people's quantity Jing Guo the open scene.
Optionally, stream of people's quantity by open scene includes: by scheduled Statistical Area in the open scene
Stream of people's quantity in domain;The flow rate calculation unit is specifically used at least one of following: when some pedestrian goes out in valid frame before
When never appearing in statistical regions and appeared in statistical regions in the currently active frame outside present statistical regions, recognize
Enter statistical regions for the pedestrian;Stream of people's quantity Jing Guo the statistical regions is calculated according to the number for entering statistical regions;When
Some pedestrian once appears in statistical regions in valid frame before and never appears in outside statistical regions and the currently active
When being appeared in outside statistical regions in frame, it is believed that the pedestrian leaves statistical regions;It is calculated and is passed through according to the number for leaving statistical regions
Stream of people's quantity of the statistical regions.
In a kind of implementation, pedestrian's weight recognition unit is specifically used for: obtaining each pedestrian in the currently active frame
External appearance characteristic or external appearance characteristic and position feature;According to the external appearance characteristic of pedestrian or external appearance characteristic and position feature, determine to work as
Each pedestrian in preceding valid frame whether be before at least one valid frame in already present pedestrian, it is new if not then generating
Character recognition and label mark the pedestrian, the pedestrian is if it is marked with existing character recognition and label.
In above-mentioned implementation, pedestrian's weight recognition unit is according to the external appearance characteristic or external appearance characteristic of pedestrian and position
Feature, determine some pedestrian in the currently active frame whether be before already present pedestrian in valid frame, comprising: use Hungary
Algorithm carries out the currently active frame and the row in valid frame before according to the external appearance characteristic of pedestrian or external appearance characteristic and position feature
People's matching.
Optionally, the pedestrian detection unit is specifically used for: using YOLO object detection method, mentions from the currently active frame
Take the image range and location information of each pedestrian.
Optionally, the camera is RGB RGB camera.
Optionally, described device operates on embedded board.
The embodiment of this specification provides a kind of computer equipment, which includes memory and processor.
Wherein, the computer program that can be run by processor is stored on memory;Computer program of the processor in operation storage
When, execute each step of real-time stream of people's statistical method of open scene in this specification embodiment.To the real-time of open scene
The detailed description of each step of stream of people's statistical method refer to before content, be not repeated.
The embodiment of this specification provides a kind of computer readable storage medium, is stored with computer on the storage medium
Program, these computer programs execute the real-time stream of people system of open scene in this specification embodiment when being run by processor
Each step of meter method.Before being referred to the detailed description of each step of real-time stream of people's statistical method of open scene
Content is not repeated.
The foregoing is merely the preferred embodiments of this specification, all the application's not to limit the application
Within spirit and principle, any modification, equivalent substitution, improvement and etc. done be should be included within the scope of the application protection.
In a typical configuration, calculating equipment includes one or more processors (CPU), input/output interface, net
Network interface and memory.
Memory may include the non-volatile memory in computer-readable medium, random access memory (RAM) and/or
The forms such as Nonvolatile memory, such as read-only memory (ROM) or flash memory (flash RAM).Memory is computer-readable medium
Example.
Computer-readable medium includes permanent and non-permanent, removable and non-removable media can be by any method
Or technology come realize information store.Information can be computer readable instructions, data structure, the module of program or other data.
The example of the storage medium of computer includes, but are not limited to phase change memory (PRAM), static random access memory (SRAM), moves
State random access memory (DRAM), other kinds of random access memory (RAM), read-only memory (ROM), electric erasable
Programmable read only memory (EEPROM), flash memory or other memory techniques, read-only disc read only memory (CD-ROM) (CD-ROM),
Digital versatile disc (DVD) or other optical storage, magnetic cassettes, tape magnetic disk storage or other magnetic storage devices
Or any other non-transmission medium, can be used for storage can be accessed by a computing device information.As defined in this article, it calculates
Machine readable medium does not include temporary computer readable media (transitory media), such as the data-signal and carrier wave of modulation.
It should also be noted that, the terms "include", "comprise" or its any other variant are intended to nonexcludability
It include so that the process, method, commodity or the equipment that include a series of elements not only include those elements, but also to wrap
Include other elements that are not explicitly listed, or further include for this process, method, commodity or equipment intrinsic want
Element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including described want
There is also other identical elements in the process, method of element, commodity or equipment.
It will be understood by those skilled in the art that the embodiment of this specification can provide as the production of method, system or computer program
Product.Therefore, the embodiment of this specification can be used complete hardware embodiment, complete software embodiment or combine software and hardware side
The form of the embodiment in face.Moreover, it wherein includes that computer is available that the embodiment of this specification, which can be used in one or more,
It is real in the computer-usable storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.) of program code
The form for the computer program product applied.
Claims (16)
1. a kind of real-time stream of people's statistical method of open scene, comprising:
The currently active frame is extracted from the live video stream for shooting the open scene;The video flowing is by being placed in open scene
The camera of top is shot;
Detect each pedestrian in the currently active frame;
Using pedestrian's weight recognizer, identical pedestrian at least one valid frame in the currently active frame and before is identified;
According to pedestrian's weight recognizer as a result, calculating stream of people's quantity Jing Guo the open scene.
2. according to the method described in claim 1, stream of people's quantity by open scene includes: by the open scene
In scheduled statistical regions stream of people's quantity;
It is described according to pedestrian's weight recognizer as a result, calculate stream of people's quantity Jing Guo the open scene, including it is following at least
One:
It appears in when some pedestrian appears in outside statistical regions in valid frame before and never in statistical regions and current
When being appeared in statistical regions in valid frame, it is believed that the pedestrian enters statistical regions;It is calculated according to the number for entering statistical regions
Stream of people's quantity by the statistical regions;
When some pedestrian was once appeared in valid frame before in statistical regions and is never appeared in outside statistical regions and is being worked as
When being appeared in outside statistical regions in preceding valid frame, it is believed that the pedestrian leaves statistical regions;According to the population number meter for leaving statistical regions
Calculate stream of people's quantity Jing Guo the statistical regions.
3. being identified in the currently active frame and before according to the method described in claim 1, described use pedestrian's weight recognizer
Identical pedestrian at least one valid frame, comprising: obtain the external appearance characteristic or external appearance characteristic of each pedestrian in the currently active frame
And position feature;According to the external appearance characteristic of pedestrian or external appearance characteristic and position feature, each pedestrian in the currently active frame is determined
Whether be before at least one valid frame in already present pedestrian, mark the row if not new character recognition and label is then generated
People if it is marks the pedestrian with existing character recognition and label.
4. according to the method described in claim 3, the external appearance characteristic or external appearance characteristic and position feature according to pedestrian, sentences
Some pedestrian in settled preceding valid frame whether be before already present pedestrian in valid frame, comprising: use Hungary Algorithm, root
The currently active frame is carried out according to the external appearance characteristic or external appearance characteristic and position feature of pedestrian to match with the pedestrian in valid frame before.
5. according to the method described in claim 1, each pedestrian in the currently active frame of detection, comprising: use YOLO mesh
Detection method is marked, the image range and location information of each pedestrian are extracted from the currently active frame.
6. according to the method described in claim 1, the camera is RGB RGB camera.
7. according to the method described in claim 1, the method operates on embedded board.
8. a kind of real-time stream of people's statistic device of open scene, comprising:
Valid frame extraction unit, for extracting the currently active frame from the live video stream for shooting the open scene;The view
Frequency stream is shot by being placed in the camera above open scene;
Pedestrian detection unit, for detecting each pedestrian in the currently active frame;
Pedestrian's weight recognition unit, for using pedestrian's weight recognizer, identifying in the currently active frame to have with before at least one
Imitate identical pedestrian in frame;
Flow rate calculation unit, for according to pedestrian's weight recognizer as a result, calculating stream of people's quantity Jing Guo the open scene.
9. device according to claim 8, stream of people's quantity by open scene includes: by the open scene
In scheduled statistical regions stream of people's quantity;
The flow rate calculation unit is specifically used at least one of following:
It appears in when some pedestrian appears in outside statistical regions in valid frame before and never in statistical regions and current
When being appeared in statistical regions in valid frame, it is believed that the pedestrian enters statistical regions;It is calculated according to the number for entering statistical regions
Stream of people's quantity by the statistical regions;
When some pedestrian was once appeared in valid frame before in statistical regions and is never appeared in outside statistical regions and is being worked as
When being appeared in outside statistical regions in preceding valid frame, it is believed that the pedestrian leaves statistical regions;According to the population number meter for leaving statistical regions
Calculate stream of people's quantity Jing Guo the statistical regions.
10. device according to claim 8, pedestrian's weight recognition unit is specifically used for: obtaining every in the currently active frame
The external appearance characteristic or external appearance characteristic and position feature of a pedestrian;It is special according to the external appearance characteristic of pedestrian or external appearance characteristic and position
Sign, determine each pedestrian in the currently active frame whether be before at least one valid frame in already present pedestrian, if not
It is to generate new character recognition and label to mark the pedestrian, the pedestrian is if it is marked with existing character recognition and label.
11. device according to claim 10, pedestrian's weight recognition unit is special according to the external appearance characteristic or appearance of pedestrian
Seek peace position feature, determine some pedestrian in the currently active frame whether be before already present pedestrian in valid frame, comprising: adopt
With Hungary Algorithm, according to the external appearance characteristic of pedestrian or external appearance characteristic and position feature carry out the currently active frame with before effectively
Pedestrian's matching in frame.
12. device according to claim 8, the pedestrian detection unit is specifically used for: YOLO object detection method is used,
The image range and location information of each pedestrian are extracted from the currently active frame.
13. device according to claim 8, the camera is RGB RGB camera.
14. device according to claim 8, described device operate on embedded board.
15. a kind of computer equipment, comprising: memory and processor;Being stored on the memory can be by processor operation
Computer program;When the processor runs the computer program, the step as described in claims 1 to 7 any one is executed
Suddenly.
16. a kind of computer readable storage medium, is stored thereon with computer program, the computer program is run by processor
When, execute the step as described in claims 1 to 7 any one.
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811331786.XA CN109902551A (en) | 2018-11-09 | 2018-11-09 | The real-time stream of people's statistical method and device of open scene |
TW108128880A TWI729454B (en) | 2018-11-09 | 2019-08-14 | Open scene real-time crowd flow statistics method and device, computer equipment and computer readable storage medium |
PCT/CN2019/110014 WO2020093829A1 (en) | 2018-11-09 | 2019-10-08 | Method and device for real-time statistical analysis of pedestrian flow in open space |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811331786.XA CN109902551A (en) | 2018-11-09 | 2018-11-09 | The real-time stream of people's statistical method and device of open scene |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109902551A true CN109902551A (en) | 2019-06-18 |
Family
ID=66943300
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811331786.XA Pending CN109902551A (en) | 2018-11-09 | 2018-11-09 | The real-time stream of people's statistical method and device of open scene |
Country Status (3)
Country | Link |
---|---|
CN (1) | CN109902551A (en) |
TW (1) | TWI729454B (en) |
WO (1) | WO2020093829A1 (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110659588A (en) * | 2019-09-02 | 2020-01-07 | 平安科技(深圳)有限公司 | Passenger flow volume statistical method and device and computer readable storage medium |
CN110705494A (en) * | 2019-10-10 | 2020-01-17 | 北京东软望海科技有限公司 | People flow monitoring method and device, electronic equipment and computer readable storage medium |
WO2020093829A1 (en) * | 2018-11-09 | 2020-05-14 | 阿里巴巴集团控股有限公司 | Method and device for real-time statistical analysis of pedestrian flow in open space |
CN112005247A (en) * | 2020-06-12 | 2020-11-27 | 深圳盈天下视觉科技有限公司 | People flow data monitoring system and people flow data display method and device thereof |
CN112149457A (en) * | 2019-06-27 | 2020-12-29 | 西安光启未来技术研究院 | People flow statistical method, device, server and computer readable storage medium |
CN112560765A (en) * | 2020-12-24 | 2021-03-26 | 上海明略人工智能(集团)有限公司 | Pedestrian flow statistical method, system, equipment and storage medium based on pedestrian re-identification |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113297888A (en) * | 2020-09-18 | 2021-08-24 | 阿里巴巴集团控股有限公司 | Method and device for checking image content detection result |
CN116665243A (en) * | 2021-03-23 | 2023-08-29 | 福建诺诚数字科技有限公司 | Method, device and equipment for measuring pedestrian distance based on TOF depth map |
TWI796033B (en) * | 2021-12-07 | 2023-03-11 | 巨鷗科技股份有限公司 | People flow analysis and identification system |
CN116385969B (en) * | 2023-04-07 | 2024-03-12 | 暨南大学 | Personnel gathering detection system based on multi-camera cooperation and human feedback |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103425967A (en) * | 2013-07-21 | 2013-12-04 | 浙江大学 | Pedestrian flow monitoring method based on pedestrian detection and tracking |
CN104318263A (en) * | 2014-09-24 | 2015-01-28 | 南京邮电大学 | Real-time high-precision people stream counting method |
CN106650695A (en) * | 2016-12-30 | 2017-05-10 | 苏州万店掌网络科技有限公司 | Video analysis technology-based people flow tracking statistics system |
CN107423708A (en) * | 2017-07-25 | 2017-12-01 | 成都通甲优博科技有限责任公司 | The method and its device of pedestrian's flow of the people in a kind of determination video |
US20180060669A1 (en) * | 2016-08-30 | 2018-03-01 | Canon Kabushiki Kaisha | Method, system and apparatus for processing an image |
CN108021848A (en) * | 2016-11-03 | 2018-05-11 | 浙江宇视科技有限公司 | Passenger flow volume statistical method and device |
Family Cites Families (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20070098303A1 (en) * | 2005-10-31 | 2007-05-03 | Eastman Kodak Company | Determining a particular person from a collection |
CN101231755B (en) * | 2007-01-25 | 2013-03-06 | 上海遥薇(集团)有限公司 | Moving target tracking and quantity statistics method |
TW201033908A (en) * | 2009-03-12 | 2010-09-16 | Micro Star Int Co Ltd | System and method for counting people flow |
CN105809178A (en) * | 2014-12-31 | 2016-07-27 | 中国科学院深圳先进技术研究院 | Population analyzing method based on human face attribute and device |
CN108388851A (en) * | 2018-02-09 | 2018-08-10 | 北京京东金融科技控股有限公司 | information statistical method, device, storage medium and electronic equipment |
CN108664946A (en) * | 2018-05-18 | 2018-10-16 | 上海极歌企业管理咨询中心(有限合伙) | Stream of people's characteristic-acquisition method based on image and device |
CN108764167B (en) * | 2018-05-30 | 2020-09-29 | 上海交通大学 | Space-time correlated target re-identification method and system |
CN109902551A (en) * | 2018-11-09 | 2019-06-18 | 阿里巴巴集团控股有限公司 | The real-time stream of people's statistical method and device of open scene |
-
2018
- 2018-11-09 CN CN201811331786.XA patent/CN109902551A/en active Pending
-
2019
- 2019-08-14 TW TW108128880A patent/TWI729454B/en active
- 2019-10-08 WO PCT/CN2019/110014 patent/WO2020093829A1/en active Application Filing
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103425967A (en) * | 2013-07-21 | 2013-12-04 | 浙江大学 | Pedestrian flow monitoring method based on pedestrian detection and tracking |
CN104318263A (en) * | 2014-09-24 | 2015-01-28 | 南京邮电大学 | Real-time high-precision people stream counting method |
US20180060669A1 (en) * | 2016-08-30 | 2018-03-01 | Canon Kabushiki Kaisha | Method, system and apparatus for processing an image |
CN108021848A (en) * | 2016-11-03 | 2018-05-11 | 浙江宇视科技有限公司 | Passenger flow volume statistical method and device |
CN106650695A (en) * | 2016-12-30 | 2017-05-10 | 苏州万店掌网络科技有限公司 | Video analysis technology-based people flow tracking statistics system |
CN107423708A (en) * | 2017-07-25 | 2017-12-01 | 成都通甲优博科技有限责任公司 | The method and its device of pedestrian's flow of the people in a kind of determination video |
Non-Patent Citations (1)
Title |
---|
朱延东: "复杂场景监控视频群体事件检测", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2020093829A1 (en) * | 2018-11-09 | 2020-05-14 | 阿里巴巴集团控股有限公司 | Method and device for real-time statistical analysis of pedestrian flow in open space |
CN112149457A (en) * | 2019-06-27 | 2020-12-29 | 西安光启未来技术研究院 | People flow statistical method, device, server and computer readable storage medium |
CN110659588A (en) * | 2019-09-02 | 2020-01-07 | 平安科技(深圳)有限公司 | Passenger flow volume statistical method and device and computer readable storage medium |
WO2021043089A1 (en) * | 2019-09-02 | 2021-03-11 | 平安科技(深圳)有限公司 | Customer traffic counting method and apparatus, device, and computer readable storage medium |
CN110705494A (en) * | 2019-10-10 | 2020-01-17 | 北京东软望海科技有限公司 | People flow monitoring method and device, electronic equipment and computer readable storage medium |
CN112005247A (en) * | 2020-06-12 | 2020-11-27 | 深圳盈天下视觉科技有限公司 | People flow data monitoring system and people flow data display method and device thereof |
CN112560765A (en) * | 2020-12-24 | 2021-03-26 | 上海明略人工智能(集团)有限公司 | Pedestrian flow statistical method, system, equipment and storage medium based on pedestrian re-identification |
Also Published As
Publication number | Publication date |
---|---|
TW202036375A (en) | 2020-10-01 |
TWI729454B (en) | 2021-06-01 |
WO2020093829A1 (en) | 2020-05-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109902551A (en) | The real-time stream of people's statistical method and device of open scene | |
CN110008799A (en) | Stream of people's situation evaluation method and device in specified region | |
CN106651916B (en) | A kind of positioning and tracing method and device of target | |
CN109697435B (en) | People flow monitoring method and device, storage medium and equipment | |
CN106096577B (en) | A kind of target tracking method in camera distribution map | |
CN105628951B (en) | The method and apparatus of speed for measurement object | |
US10529077B2 (en) | System and method for detecting interaction | |
CN102890781B (en) | A kind of Highlight recognition methods for badminton game video | |
WO2019083738A9 (en) | Methods and systems for applying complex object detection in a video analytics system | |
CN105405150B (en) | Anomaly detection method and device based on fusion feature | |
Zhang et al. | Social attribute-aware force model: exploiting richness of interaction for abnormal crowd detection | |
JP2020149704A (en) | System and method for activity monitoring using video data | |
CN108470353A (en) | A kind of method for tracking target, device and storage medium | |
Srivastava et al. | Crowd flow estimation using multiple visual features for scenes with changing crowd densities | |
CN108038176A (en) | A kind of method for building up, device, electronic equipment and the medium in passerby storehouse | |
CN110533013A (en) | A kind of track-detecting method and device | |
CN110111362A (en) | A kind of local feature block Similarity matching method for tracking target | |
CN113269091A (en) | Personnel trajectory analysis method, equipment and medium for intelligent park | |
CN109636828A (en) | Object tracking methods and device based on video image | |
CN207233038U (en) | Face is called the roll and number system | |
CN111899279A (en) | Method and device for detecting motion speed of target object | |
Khan et al. | Robust head detection in complex videos using two-stage deep convolution framework | |
Alotibi et al. | Cnn-based crowd counting through iot: Application for saudi public places | |
CN110059531A (en) | Behavioral value method and device of fighting based on video image | |
Tao et al. | Real-time detection and tracking of moving object |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190618 |