CN112329635B - Method and device for counting store passenger flow - Google Patents

Method and device for counting store passenger flow Download PDF

Info

Publication number
CN112329635B
CN112329635B CN202011227756.1A CN202011227756A CN112329635B CN 112329635 B CN112329635 B CN 112329635B CN 202011227756 A CN202011227756 A CN 202011227756A CN 112329635 B CN112329635 B CN 112329635B
Authority
CN
China
Prior art keywords
candidate
face
store
salesclerk
data information
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.)
Active
Application number
CN202011227756.1A
Other languages
Chinese (zh)
Other versions
CN112329635A (en
Inventor
毛丽
曹松
吴浩
徐子豪
宋君
陶海
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Vion Intelligent Technology Co ltd
Original Assignee
Beijing Vion Intelligent Technology Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Beijing Vion Intelligent Technology Co ltd filed Critical Beijing Vion Intelligent Technology Co ltd
Priority to CN202011227756.1A priority Critical patent/CN112329635B/en
Publication of CN112329635A publication Critical patent/CN112329635A/en
Application granted granted Critical
Publication of CN112329635B publication Critical patent/CN112329635B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • 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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0201—Market modelling; Market analysis; Collecting market data
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G06V40/161—Detection; Localisation; Normalisation
    • G06V40/166—Detection; Localisation; Normalisation using acquisition arrangements
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G06V40/168—Feature extraction; Face representation

Abstract

The invention provides a method and a device for counting the passenger flow of a shop, wherein the method for counting the passenger flow of the shop comprises the following steps: continuously tracking and detecting dynamic faces in a video scene of a shop, acquiring a face feature data packet, and acquiring people stream data information of the shop based on the face feature data packet; performing secondary clustering analysis on the human face data information to extract a plurality of candidate salesclerk data sets from the human face feature data packet; extracting an actual salesclerk data set from the candidate salesclerk data sets based on the statistical analysis of the time domain characteristics, and acquiring actual salesclerk data information of the shop based on the actual salesclerk data set; actual store employee data information in the people flow data information is filtered out, so that passenger flow data information of a shop laid within a first preset time length is obtained. The invention solves the problem that in the prior art, when the passenger flow volume of the store is counted, store personnel entering and exiting the store flow cannot be filtered out quickly and efficiently, so that the passenger flow volume counting precision of the store is influenced.

Description

Method and device for counting store passenger flow
Technical Field
The invention relates to the technical field of image processing, in particular to a shop passenger flow statistical method and a shop passenger flow statistical device.
Background
The shop passenger flow statistics is to accurately count the number of people who pass in and out of each exit or entrance in real time by installing passenger flow statistics equipment in a shop operation area, so that shops are scientifically managed according to data, and a targeted marketing means is adopted according to the passenger flow of each shop.
Effective business management has become an important factor in the success or failure of business marketing today when business competition is becoming increasingly intense. The business model gradually changes from the traditional seat business to the extremely initiative business, and puts higher requirements on business managers: the system has the advantages that the system can quickly respond to the weak change of the market in the shortest time, has market predictability, saves the commercial operation cost to the maximum extent, and improves the scientificity of daily operation decision of a market, the comfort of shopping environment, the rationality of human resource allocation and the like.
Through the passenger flow statistics of different floors and different regions, managers can count the attraction rate and the busyness of each region, so that the berths are reasonably distributed, and the sales volume is increased. In a customer flow volume statistics scene of a store, the existence of store personnel often affects the statistics result of the customer flow volume, and particularly for the store with more store personnel and fewer customers, the influence can greatly reduce the customer flow volume statistics precision of the store, and is not beneficial to timely and accurate operation strategy adjustment of the store aiming at the customer flow volume change, so that the operation cost of the store is improved. Therefore, when the passenger flow volume of the store is counted, how to quickly and efficiently filter out the store staff who enter and exit the store passenger flow to ensure the statistical accuracy of the passenger flow volume of the store becomes a problem to be solved in the prior art.
Disclosure of Invention
The invention mainly aims to provide a store passenger flow statistical method and a store passenger flow statistical device, which are used for solving the problem that in the prior art, when the passenger flow of a store is unified, store personnel entering and exiting the store passenger flow cannot be filtered out quickly and efficiently, and the passenger flow statistical precision of the store is influenced.
In order to achieve the above object, according to an aspect of the present invention, there is provided a store passenger flow statistic method including: step S1, continuously tracking and detecting a dynamic face in a video scene of a shop within a first preset time length, acquiring a face feature data packet, and acquiring people stream data information of the shop based on the face feature data packet; step S2, carrying out secondary clustering analysis on the human flow data information to extract a plurality of candidate salesclerk data sets from the human face feature data packet; step S3, extracting an actual clerk data set containing actual clerk face features from the candidate clerk data sets based on the statistical analysis of the time domain features, and acquiring actual clerk data information of the shop based on the actual clerk data set; step S4, filtering out actual clerk data information in the people flow data information to obtain the passenger flow data information when the shop is laid within the first preset time.
Further, in step S2, performing quadratic clustering analysis on the stream data information includes: step S21, local clustering calculation, including: step S211, carrying out time interval segmentation on the first preset time length to form a plurality of sub-time intervals; step S212, constructing a kd-tree for all the face features in the sub-time period, performing first K nearest neighbor analysis on all the face features in the sub-time period, performing first judgment to obtain a local cluster set, and recording the face features in the local cluster set; and step S213, repeating the step S212 until all the sub-time periods are covered so as to obtain a plurality of local cluster sets.
Further, in step S2, performing quadratic clustering analysis on the stream data information further includes: step S22, global cluster computation, including: step S221, calculating the average value F of the human face features in each local clustering set respectively; and step S222, constructing a kd-tree for all the average values F, performing repeated K nearest neighbor analysis on all the average values F, and performing second judgment to obtain a global clustering set.
Further, in step S2, performing quadratic clustering analysis on the stream data information further includes:
step S23, extracting candidate frequent items, including: and counting all the global cluster sets, and carrying out third judgment to obtain a candidate salesclerk data set.
Further, the first determination includes: setting K1 in the first K nearest neighbor analysis to be 100; selecting a distance formula, and setting a distance threshold th1 to be 0.5; and combining part of the face features meeting the distance threshold constraint in all the face features in the sub-time period to form a local clustering set.
Further, the second determination includes: setting K2 in the repeated K nearest neighbor analysis to 50; selecting a distance formula, and setting a distance threshold th2 to be 0.5; and merging the partial average values F meeting the distance threshold constraint in all the average values F to form a global cluster set.
Further, the distance formula is:
Figure GDA0003528913950000021
wherein the content of the first and second substances,
Figure GDA0003528913950000022
wherein f isa(i) The ith feature in the K nearest neighbor list, O, representing face ab(f) Representing the position of the face feature f in the K nearest neighbor list of the face b;
Figure GDA0003528913950000023
wherein f isb(i) The ith feature in the K nearest neighbor list, O, representing face ba(f) Indicating the position of the face feature f in the K nearest neighbor list of the face a.
Further, the third determination includes: when the number of the human face features in the global clustering set is larger than or equal to a first preset value, the global clustering set is reserved to serve as a candidate salesclerk data set; the first preset value is greater than or equal to 20 and less than or equal to 200.
Further, in step S3, calculating a weight value S of each candidate clerk data set according to a weighting formula including a time domain feature, and when the weight value S is greater than or equal to a preset weight value, determining that the candidate clerk data set is an actual clerk data set; wherein the time domain feature comprises a time feature StAnd frequency characteristic SpAnd a stability characteristic σ.
Further, the preset weight value is greater than or equal to 0.7 and less than or equal to 0.9; the weighting formula is: s ═ wtSt+wpSp×wσσ，
Wherein w is more than or equal to 0.1t≤0.5，0.3≤wp≤0.7，0.3≤wσ≤0.7；StAccording to a time characteristic algorithmS is more than or equal to 0.1t≤2.1；SpObtained according to a frequency characteristic algorithm, Sp0 or Sp1 is ═ 1; and sigma is obtained according to a stability characteristic algorithm, and sigma is more than or equal to 0.
According to another aspect of the present invention, there is provided a store passenger flow statistic apparatus including: the image acquisition unit is used for acquiring a video scene of a shop; the image extraction unit is used for continuously tracking and detecting a dynamic face in a video scene, acquiring a face characteristic data packet, and acquiring people stream data information of a shop based on the face characteristic data packet; the image analysis unit is used for carrying out secondary clustering analysis on the human flow data information so as to extract a plurality of candidate salesclerk data sets from the human face feature data packet; the image processing unit is used for extracting an actual salesclerk data set containing the face features of the actual salesclerk from the candidate salesclerk data sets based on the statistical analysis of the time domain features and acquiring the actual salesclerk data information of the shop based on the actual salesclerk data set; and the passenger flow statistical unit is used for filtering actual store employee data information in the people flow data information to acquire the passenger flow data information of the shop laid in a first preset time.
By applying the technical scheme of the invention, a plurality of candidate salesclerk data sets can be effectively extracted from the face feature data packet of the store through secondary cluster analysis, and then the actual salesclerk data set containing the face features of the actual salesclerk can be accurately judged from the plurality of candidate salesclerk data sets through the statistical analysis of the time domain features, so that the actual salesclerk data information can be filtered from the people flow data information of the store, and the passenger flow data information of the store can be finally obtained. According to the technical scheme, under the condition that no person participates in collecting the store clerk data information, the store clerk data information in the people flow data information can be automatically mined, so that the personal privacy and the data security of the store clerk are protected, the intelligent degree of the passenger flow statistics of the store is improved, and the passenger flow statistics precision of the store is improved.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this application, illustrate embodiments of the invention and, together with the description, serve to explain the invention and not to limit the invention. In the drawings:
FIG. 1 illustrates a flow diagram of a store airflow statistics method in accordance with an alternative embodiment of the present invention;
fig. 2 shows a flowchart of the local cluster calculation in step S2 in fig. 1;
fig. 3 shows a flowchart of the global cluster calculation in step S2 in fig. 1.
Detailed Description
It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict. The present invention will be described in detail below with reference to the embodiments with reference to the attached drawings.
In order to make the technical solutions of the present invention better understood, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
It should be noted that the terms "first," "second," and the like in the description and claims of the present invention and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged under appropriate circumstances in order to facilitate the description of the embodiments of the invention herein. Furthermore, the terms "comprises," "comprising," "includes," "including," "has," "having," and any variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements explicitly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
The invention provides a shop passenger flow statistical method and a shop passenger flow statistical device, aiming at solving the problem that in the prior art, when the shop passenger flow statistics is carried out, shop personnel entering and exiting the shop can not be filtered out quickly and efficiently, and the statistical accuracy of the shop passenger flow is influenced.
FIG. 1 is a flow chart of a store airflow statistics method according to an embodiment of the invention. As shown in fig. 1, the method for counting the number of the shop passengers comprises the following steps:
step S1, continuously tracking and detecting a dynamic face in a video scene of a shop within a first preset time length, acquiring a face feature data packet, and acquiring people stream data information of the shop based on the face feature data packet;
step S2, carrying out secondary clustering analysis on the human flow data information to extract a plurality of candidate salesclerk data sets from the human face feature data packet;
step S3, extracting an actual clerk data set containing actual clerk face features from the candidate clerk data sets based on the statistical analysis of the time domain features, and acquiring actual clerk data information of the shop based on the actual clerk data set;
step S4, filtering out actual clerk data information in the people flow data information to obtain the passenger flow data information when the shop is laid within the first preset time.
A plurality of candidate salesclerk data sets can be effectively extracted from the face feature data packet of the store through secondary cluster analysis, and then the actual salesclerk data set containing the face features of the actual salesclerk can be accurately judged from the plurality of candidate salesclerk data sets through the statistical analysis of the time domain features, so that the actual salesclerk data information can be favorably filtered from the people flow data information of the store, and the passenger flow data information of the store can be finally obtained. According to the technical scheme, under the condition that no person participates in collecting the store clerk data information, the store clerk data information in the people flow data information can be automatically mined, so that the personal privacy and the data security of the store clerk are protected, the intelligent degree of the passenger flow statistics of the store is improved, and the passenger flow statistics precision of the store is improved.
It should be noted that, in the present internet communication field, data security is a very important element. If facial features of store personnel, such as facial images, need to be acquired, privacy disclosure is possible. Under the current situation that data security is increasingly important and the background of the era that personal privacy is seriously considered, by using the shop passenger flow statistical method and the shop passenger flow statistical device, under the condition that data information of store personnel, such as face images, is not collected manually, the store personnel can be accurately and quickly identified in a large sample of face features according to a filtering algorithm, and finally the store personnel are filtered from the passenger flow so as to achieve the purpose of passenger flow statistics.
Corresponding to the above-mentioned shop passenger flow statistical method, the present application also provides a shop passenger flow statistical device, comprising: the system comprises an image acquisition unit, an image extraction unit, an image analysis unit, an image processing unit and a passenger flow statistics unit, wherein the image acquisition unit is used for acquiring a video scene of a shop; the image extraction unit is used for continuously tracking and detecting a dynamic face in a video scene, acquiring a face characteristic data packet, and acquiring people stream data information of a shop based on the face characteristic data packet; the image analysis unit is used for performing secondary clustering analysis on the human face data information so as to extract a plurality of candidate salesclerk data sets from the human face feature data packet; the image processing unit is used for extracting an actual salesclerk data set containing the face features of an actual salesclerk from the plurality of candidate salesclerk data sets based on the statistical analysis of the time domain features, and acquiring the actual salesclerk data information of the shop based on the actual salesclerk data set; the passenger flow statistical unit is used for filtering actual store employee data information in the people flow data information to obtain passenger flow data information of the shop spread within a first preset time.
In this optional embodiment, a plurality of functional units of the store passenger flow statistics apparatus may be integrally configured or separately configured, wherein the image acquisition unit and the image extraction unit may employ image capturing devices such as a camera and a video camera, a height of the image capturing device is 2.5m to 3.5m from the ground, an image capturing center line of a camera of the image capturing device is preferably inclined at an angle of 45 degrees with respect to the ground, and the image capturing device captures an image toward a face of a person. The method comprises the steps of extracting human face features at the front end of the image capturing device in real time, transmitting all the human face features to a server end for subsequent operation, and integrating an image analysis unit, an image processing unit and a passenger flow statistics unit at the server end. One specific method is to use a deep learning algorithm to identify the human face in the image shot by the image capturing equipment, and extract the human face features after registration. The face features extracted by the algorithm model are 256-dimensional vectors and are marked as f. And simultaneously, recording the face features f and the snapshot time t.
Fig. 2 shows a flowchart of the local cluster calculation in the above step S2; in step S2, performing quadratic clustering analysis on the stream data information includes: step S21, the local clustering calculation, specifically, the local clustering calculation includes: step S211, carrying out time interval segmentation on the first preset time length to form a plurality of sub-time intervals; step S212, constructing a kd-tree for all the face features in the sub-time period, performing first K nearest neighbor analysis on all the face features in the sub-time period, performing first judgment to obtain a local cluster set, and recording the face features in the local cluster set; and step S213, repeating the step S212 until all the sub-time periods are covered so as to obtain a plurality of local cluster sets. Wherein the first judging comprises: setting K1 in the first K nearest neighbor analysis to be 100; selecting a distance formula, and setting a distance threshold th1 to be 0.5; and combining part of the face features meeting the distance threshold constraint in all the face features in the sub-time period to form a local clustering set.
As a specific embodiment of the application, store passenger flow statistics is carried out on stores by taking the first preset time length as all weather, and considering that the data volume of the face features needing to be collected by the stores all weather is huge, and each sub-time period is 2 hours, and a kd-tree is constructed for all the face features in the sub-time period.
Fig. 3 shows a flowchart of the global cluster calculation in the above step S2; in step S2, performing quadratic clustering analysis on the stream data information further includes: step S22, global cluster calculation, specifically, for all-weather data of facial features, the global cluster calculation includes: step S221, calculating the average value F of the human face features in each local clustering set respectively; and step S222, constructing a kd-tree for all the average values F, performing repeated K nearest neighbor analysis on all the average values F, and performing second judgment to obtain a global clustering set. Wherein the second determination includes: setting K2 in the repeated K nearest neighbor analysis to 50; selecting a distance formula, and setting a distance threshold th2 to be 0.5; and merging the partial average values F meeting the distance threshold constraint in all the average values F to form a global cluster set.
It should be noted that, the distance formula selected in the local clustering calculation and the global clustering calculation is the same, and specifically:
Figure GDA0003528913950000051
wherein the content of the first and second substances,
Figure GDA0003528913950000052
wherein f isa(i) The ith feature in the K nearest neighbor list, O, representing face ab(f) Representing the position of the face feature f in the K nearest neighbor list of the face b;
Figure GDA0003528913950000061
wherein f isb(i) The ith feature in the K nearest neighbor list, O, representing face ba(f) Indicating the position of the face feature f in the K nearest neighbor list of the face a.
Furthermore, in step S2, performing quadratic clustering analysis on the stream data information further includes: step S23, extracting candidate frequent items, including: and counting all the global cluster sets, and carrying out third judgment to obtain a candidate salesclerk data set. Wherein the third judging comprises: when the number of the human face features in the global clustering set is larger than or equal to a first preset value, the global clustering set is reserved to serve as a candidate salesclerk data set; the first preset value is greater than or equal to 20 and less than or equal to 200.
Preferably, the first preset value is 50.
In conclusion, the technical scheme of the application is suitable for operation under the condition that the number of the snap-shot images is large, the memory occupation is greatly reduced through two times of hierarchical clustering calculation, and the operation amount is reduced on the basis of keeping high precision by using an exquisite distance algorithm.
Alternatively, when k nearest neighbors of each face feature are obtained, other specific implementations of k nearest neighbor algorithms can be used, and are not limited to the kd-tree.
In step S3, the statistical analysis based on the time domain features specifically includes calculating a weight value S of each candidate clerk data set according to a weighting formula including the time domain features, and when the weight value S is greater than or equal to a preset weight value, determining that the candidate clerk data set is an actual clerk data set; wherein the time domain feature comprises a time feature StAnd frequency characteristic SpAnd a stability characteristic σ.
Further, in this embodiment, the preset weight value is greater than or equal to 0.7 and less than or equal to 0.9; the weighting formula is: s ═ wtSt+wpSp×wσSigma, wherein, 0.1 is less than or equal to wt≤0.5，0.3≤wp≤0.7，0.3≤wσ≤0.7；StObtained according to a time characteristic algorithm, S is more than or equal to 0.1t≤2.1；SpObtained according to a frequency characteristic algorithm, Sp1 or Sp0; and sigma is obtained according to a stability characteristic algorithm, and sigma is more than or equal to 0.
Preferably, wt＝0.3，wp＝0.5，wσThe preset weight value is 0.8, which is 0.5.
Specifically, StThe specific steps obtained according to the time characteristic algorithm are as follows: performing histogram statistics on the snapshot time t aiming at all the face features f and the snapshot time t of a certain candidate salesman data set, wherein in the first step: the histogram bins is {0: 00-10: 00,10: 00-11: 00,11: 00-12: 00,12: 00-13: 00,13: 00-14: 00,14: 00-15: 00,15: 00-16: 0,16: 00-17: 00,17: 00-18: 00,18: 00-19: 00,19: 00-24: 00}, and b [ i ] is used below]The ith statistic in the table is shown, i being 0,1, … 10. The second step is that: if b [0 ]]>0 or b 10]>0, then StAnd 0.5. The third step: then for each i ═ 0,1, … 10, if b [ i [ ]]>0, then St0.1. To sum up: s is more than or equal to 0.1t≤2.1。
The second step is designed to take into account the working hours of the store clerks, which tend to arrive earlier in the store (occurring during the first time period) and leave later (occurring during the last time period). The third step is designed in consideration of the snapshot time distribution of the store clerk, and since the store clerk is always in the store all day long, the snapshot time distribution is very wide. And customers are often only snapshotted in one or two hours.
Specifically, SpThe method comprises the following specific steps of: counting the face feature number of each candidate salesman data set for all-weather candidate salesman data sets, and solving the quartile Q of the face feature number3(i.e., 75% of the candidate clerk data set has fewer items than Q)3) And for each candidate salesclerk data set, if the number of the face features contained in the candidate salesclerk data set is more than or equal to Q3Then S isp1 is ═ 1; if the number of the face features contained in the image is less than Q3Then Sp＝0。
Specifically, the specific steps of σ obtaining according to the stability feature algorithm are as follows: for the data of the face features of two consecutive days, the distance threshold th3 is specified to be 0.95. If the cosine similarity between the average feature M of the face feature of a certain candidate clerk data set and the candidate clerk data set of the previous day is greater than or equal to th3, the candidate clerk data set is considered to appear in the previous day and is marked as i (occur), if so, i (occur) is 1, otherwise, i (occur) is 0. The stability characteristic σ is calculated according to the following formula:
σT＝I(occur)+λ×σT-1where the update rate λ is 0.5, σTFor stability characteristics at day T, σ in the above iterative formula1Is 0.
It should be added that, the present application also provides an electronic device, including: the device comprises a shell, a processor, a memory, a circuit board and a power circuit, wherein the circuit board is arranged in a space enclosed by the shell, and the processor and the memory are arranged on the circuit board; a power supply circuit for supplying power to each circuit or device of the electronic apparatus; the memory is used for storing executable program codes; the processor executes a program corresponding to the executable program code by reading the executable program code stored in the memory, and is used for executing the store passenger flow statistical method.
The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments.
The integrated unit in the above embodiments, if implemented in the form of a software functional unit and sold or used as a separate product, may be stored in the above computer-readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for causing one or more computer devices (which may be personal computers, servers, network devices, etc.) to execute all or part of the steps of the method according to the embodiments of the present invention.
In the above embodiments of the present invention, the descriptions of the respective embodiments have respective emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to related descriptions of other embodiments.
In the several embodiments provided in the present application, it should be understood that the disclosed client may be implemented in other manners. The above-described embodiments of the apparatus are merely illustrative, and for example, the division of the units is only one type of division of logical functions, and there may be other divisions when actually implemented, for example, a plurality of units or components may be combined or may be integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, units or modules, and may be in an electrical or other form.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (7)

1. A store passenger flow statistics method, comprising:
step S1, continuously tracking and detecting a dynamic face in a video scene of the shop within a first preset time length, acquiring a face feature data packet, and acquiring people stream data information of the shop based on the face feature data packet;
step S2, carrying out secondary clustering analysis on the people flow data information to extract a plurality of candidate salesclerk data sets from the human face feature data packet; performing secondary clustering analysis on the people flow data information comprises:
step S21, local clustering calculation, including: step S211, carrying out time interval segmentation on the first preset time length to form a plurality of sub-time intervals; step S212, constructing a kd-tree for all the face features in the sub-time period, performing first K nearest neighbor analysis on all the face features in the sub-time period, performing first judgment to obtain a local cluster set, and recording the face features in the local cluster set; step S213, repeating step S212 until all the sub-time periods are covered, so as to obtain a plurality of local cluster sets;
step S22, global cluster computation, including: step S221, calculating the average value F of the human face features in each local clustering set respectively; step S222, constructing a kd-tree for all the average values F, performing repeated K nearest neighbor analysis on all the average values F, and performing second judgment to obtain a global clustering set;
step S23, extracting candidate frequent items, including: counting all the global cluster sets, and performing third judgment to obtain the candidate salesclerk data set;
step S3, extracting an actual clerk data set containing actual clerk face features from the candidate clerk data sets based on the statistical analysis of the time domain features, and acquiring actual clerk data information of the shop based on the actual clerk data set;
step S4, filtering out the actual clerk data information in the people flow data information to obtain the customer flow data information of the store within the first preset time.
2. The store airflow statistic method according to claim 1, wherein said first determination includes:
setting K1 in the first K nearest neighbor analysis to be 100;
selecting a distance formula, and setting a distance threshold th1 to be 0.5;
and combining the face features meeting the distance threshold constraint in all the face features in the sub-time period to form the local clustering set.
3. The store airflow statistic method according to claim 1, wherein said second determination comprises:
setting K2 in the repeated K nearest neighbor analysis to 50;
selecting a distance formula, and setting a distance threshold th2 to be 0.5;
merging all of the average values F that satisfy a distance threshold constraint to form the global cluster set.
4. The store passenger flow statistical method according to claim 2 or 3,
the distance formula is:
Figure FDA0003528913940000021
wherein the content of the first and second substances,
Figure FDA0003528913940000022
wherein f isa(i) The ith feature in the K nearest neighbor list, O, representing face ab(f) Representing the position of the face feature f in the K nearest neighbor list of the face b;
Figure FDA0003528913940000023
wherein f isb(i) The ith feature in the K nearest neighbor list, O, representing face ba(f) Indicating the position of the face feature f in the K nearest neighbor list of the face a.
5. The store airflow statistic method according to claim 1, wherein said third determination includes:
when the number of the human face features in the global clustering set is larger than or equal to a first preset value, the global clustering set is reserved as the candidate clerk data set; the first preset value is greater than or equal to 20 and less than or equal to 200.
6. The store passenger flow statistical method according to claim 1, wherein in step S3, a weight value S of each candidate clerk data set is calculated according to a weighting formula including a time domain feature, and when the weight value S is greater than or equal to a preset weight value, the candidate clerk data set is determined to be an actual clerk data set; wherein the time domain feature comprises a time feature StAnd frequency characteristic SpAnd a stability characteristic σ;
the time characteristic StThe method is obtained according to a time characteristic algorithm, and comprises the following specific steps: performing histogram statistics on the snapshot time t aiming at all the face features f and the snapshot time t of a certain candidate salesman data set, wherein in the first step: obtaining histogram bins, using b [ i ]]Represents the ith statistic in the histogram bins, where i is 0,1, … 10; the second step is that: if b [0 ]]>0 or b 10]>0, then St0.5; the third step: and then, for each of i ═ 0,1, … 10,if b [ i ]]>0, then St+＝0.1；
The frequency characteristic SpThe method is obtained according to a frequency characteristic algorithm, and comprises the following specific steps: counting the face feature number of each candidate salesclerk data set and solving the quartile Q of the face feature number for all the candidate salesclerk data sets in all weather3For each candidate salesclerk data set, if the number of face features contained in the candidate salesclerk data set is greater than or equal to Q3Then S isp1 is ═ 1; if the number of the face features contained in the image is less than Q3Then Sp＝0；
The stability characteristic sigma is obtained according to a stability characteristic algorithm, and the method specifically comprises the following steps: for the face features of two consecutive days, a distance threshold th3 is specified to be 0.95; if the cosine similarity between the average feature M of the face feature of a certain candidate clerk data set and the candidate clerk data set of the previous day is greater than or equal to th3, the candidate clerk data set is considered to appear in the previous day and is marked as i (occur), if so, i (occur) is 1, otherwise, i (occur) is 0; the iterative formula of the stability characteristic σ is: sigmaT＝I(occur)+λ×σT-1Where the update rate λ is 0.5, σTFor stability characteristics at day T, σ in the above iterative formula1Is 0.
7. The store passenger flow statistical method according to claim 6, wherein the preset weight value is greater than or equal to 0.7 and less than or equal to 0.9;
the weighting formula is as follows: s ═ wtSt+wpSp×wσσ，
Wherein w is more than or equal to 0.1t≤0.5，0.3≤wp≤0.7，0.3≤wσ≤0.7；
Said StObtained according to a time characteristic algorithm, S is more than or equal to 0.1t≤2.1；
Said SpObtained according to a frequency characteristic algorithm, Sp0 or Sp＝1；
The sigma is obtained according to a stability characteristic algorithm, and the sigma is more than or equal to 0.
CN202011227756.1A 2020-11-06 2020-11-06 Method and device for counting store passenger flow Active CN112329635B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011227756.1A CN112329635B (en) 2020-11-06 2020-11-06 Method and device for counting store passenger flow

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011227756.1A CN112329635B (en) 2020-11-06 2020-11-06 Method and device for counting store passenger flow

Publications (2)

Publication Number Publication Date
CN112329635A CN112329635A (en) 2021-02-05
CN112329635B true CN112329635B (en) 2022-04-29

Family

ID=74316766

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011227756.1A Active CN112329635B (en) 2020-11-06 2020-11-06 Method and device for counting store passenger flow

Country Status (1)

Country Link
CN (1) CN112329635B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112906592A (en) * 2021-03-03 2021-06-04 新疆爱华盈通信息技术有限公司 Passenger flow volume analysis method and system and electronic equipment

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8005012B1 (en) * 2009-01-30 2011-08-23 Juniper Networks, Inc. Traffic analysis of data flows
US8570376B1 (en) * 2008-11-19 2013-10-29 Videomining Corporation Method and system for efficient sampling of videos using spatiotemporal constraints for statistical behavior analysis
CN105488478A (en) * 2015-12-02 2016-04-13 深圳市商汤科技有限公司 Face recognition system and method
CN110175549A (en) * 2019-05-20 2019-08-27 腾讯科技（深圳）有限公司 Face image processing process, device, equipment and storage medium
CN111784387A (en) * 2020-06-23 2020-10-16 大连中维世纪科技有限公司 Multi-dimensional big data-based consumer brand loyalty analysis method

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109325805B (en) * 2018-09-20 2022-03-11 北京工业大学 Unmanned supermarket passenger flow prediction method based on K-means clustering algorithm and discrete wavelet transform
CN110766474A (en) * 2019-10-30 2020-02-07 浙江易时科技股份有限公司 Sales exhibition room passenger flow batch statistics based on face recognition technology
CN111553288A (en) * 2020-04-29 2020-08-18 腾讯科技（深圳）有限公司 Passenger flow information processing method and device

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8570376B1 (en) * 2008-11-19 2013-10-29 Videomining Corporation Method and system for efficient sampling of videos using spatiotemporal constraints for statistical behavior analysis
US8005012B1 (en) * 2009-01-30 2011-08-23 Juniper Networks, Inc. Traffic analysis of data flows
CN105488478A (en) * 2015-12-02 2016-04-13 深圳市商汤科技有限公司 Face recognition system and method
CN110175549A (en) * 2019-05-20 2019-08-27 腾讯科技（深圳）有限公司 Face image processing process, device, equipment and storage medium
CN111784387A (en) * 2020-06-23 2020-10-16 大连中维世纪科技有限公司 Multi-dimensional big data-based consumer brand loyalty analysis method

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
"Tracking People Motion Based on Lloyd"s Clustering Algorithm";R.Rajeev等;《International Journal of Computer Science and Information Technologies》;20141231;第5卷(第2期);全文 *
"基于人脸检测的客流量统计研究";吴冰;《中国优秀博硕士学位论文全文数据库(硕士)·信息科技辑》;20140415;第2014年卷(第04期);全文 *

Also Published As

Publication number Publication date
CN112329635A (en) 2021-02-05

Similar Documents

Publication Publication Date Title
US7957565B1 (en) Method and system for recognizing employees in a physical space based on automatic behavior analysis
US8570376B1 (en) Method and system for efficient sampling of videos using spatiotemporal constraints for statistical behavior analysis
CN115272037B (en) Smart city based on Internet of things regional public security management early warning method and system
JPWO2017122258A1 (en) Congestion status monitoring system and congestion status monitoring method
CN109784274B (en) Method for identifying trailing and related product
CN106296879B (en) The methods, devices and systems that flow of the people based on wireless exploration obtains
US20080114633A1 (en) Method and Apparatus for Analyzing Activity in a Space
US20070253595A1 (en) Still Image Queue Analysis System and Method
Draughon et al. Implementation of a computer vision framework for tracking and visualizing face mask usage in urban environments
CN110334564A (en) A kind of permanent resident population's recognition methods and system based on target following
CN109145707A (en) Image processing method and device, electronic equipment and storage medium
CN111353338A (en) Energy efficiency improvement method based on business hall video monitoring
CN112329635B (en) Method and device for counting store passenger flow
CN113569659A (en) Passenger flow volume analysis method and system based on big data, intelligent terminal and storage medium
Chandran et al. Real-time identification of pedestrian meeting and split events from surveillance videos using motion similarity and its applications
CN110363180A (en) A kind of method and apparatus and equipment that statistics stranger's face repeats
Ryan Crowd monitoring using computer vision
CN113822367B (en) Regional behavior analysis method, system and medium based on human face
CN112699328A (en) Network point service data processing method, device, system, equipment and storage medium
CN108009530A (en) A kind of identity calibration system and method
CN114820239A (en) Community intelligent management method and system based on AI visual identification
CN110751046B (en) User identification method and device, readable storage medium and terminal equipment
CN112785462A (en) Scenic spot passenger flow volume statistics evaluation system based on big data
CN113065892A (en) Information pushing method, device, equipment and storage medium
CN106295597A (en) A kind of method and device obtaining passenger flow information

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