WO2021043089A1 - 一种客流量统计方法、装置、设备及计算机可读存储介质 - Google Patents

一种客流量统计方法、装置、设备及计算机可读存储介质 Download PDF

Info

Publication number
WO2021043089A1
WO2021043089A1 PCT/CN2020/112338 CN2020112338W WO2021043089A1 WO 2021043089 A1 WO2021043089 A1 WO 2021043089A1 CN 2020112338 W CN2020112338 W CN 2020112338W WO 2021043089 A1 WO2021043089 A1 WO 2021043089A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
humanoid
images
preset
passenger flow
Prior art date
Application number
PCT/CN2020/112338
Other languages
English (en)
French (fr)
Inventor
陈思静
Original Assignee
平安科技(深圳)有限公司
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 平安科技(深圳)有限公司 filed Critical 平安科技(深圳)有限公司
Publication of WO2021043089A1 publication Critical patent/WO2021043089A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00Scenes; Scene-specific elements
    • G06V20/50Context or environment of the image
    • G06V20/52Surveillance or monitoring of activities, e.g. for recognising suspicious objects
    • G06V20/53Recognition of crowd images, e.g. recognition of crowd congestion
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00Scenes; Scene-specific elements
    • G06V20/40Scenes; Scene-specific elements in video content
    • G06V20/41Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
    • G06V20/42Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items of sport video content
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16Human faces, e.g. facial parts, sketches or expressions
    • G06V40/172Classification, e.g. identification

Definitions

  • This application relates to the technical field of retail data processing, and in particular to a method, device, device, and computer-readable storage medium for passenger flow statistics.
  • This application provides a method, device, device, and computer-readable storage medium for passenger flow statistics to solve the technical problem of lack of accuracy in passenger flow statistics.
  • a passenger flow statistics method includes the following steps:
  • Recognizing the humanoid image by adopting the pedestrian re-recognition technology, and defining the humanoid image containing the preset clothing image as the clerk image, so as to distinguish the clerk image and the customer image from the plurality of humanoid images;
  • the number of the customer images is counted to obtain the passenger flow corresponding to the video data.
  • a passenger flow statistics device which includes:
  • the data acquisition module is used to acquire video data for passenger flow statistics
  • the human figure recognition module is used to perform human figure recognition on each frame of image in the video data to identify multiple human figure images
  • An image distinguishing module for recognizing the humanoid image using pedestrian re-recognition technology, and defining a humanoid image containing a preset dress image as a clerk image, so as to distinguish a clerk image and a customer image from the plurality of humanoid images ;
  • the passenger flow statistics module is used to count the number of the customer images to obtain the passenger flow corresponding to the video data.
  • An electronic device comprising: a memory and a processor, the memory storing a passenger flow statistics program, and the following steps are implemented when the passenger flow statistics program is executed by the processor:
  • Recognizing the humanoid image by adopting the pedestrian re-recognition technology, and defining the humanoid image containing the preset clothing image as the clerk image, so as to distinguish the clerk image and the customer image from the plurality of humanoid images;
  • the number of the customer images is counted to obtain the passenger flow corresponding to the video data.
  • a computer-readable storage medium having a passenger flow statistics program stored on the computer-readable storage medium, and the passenger flow statistics program may be executed by one or more processors, and the following steps are implemented:
  • Recognizing the humanoid image by adopting the pedestrian re-recognition technology, and defining the humanoid image containing the preset clothing image as the clerk image, so as to distinguish the clerk image and the customer image from the plurality of humanoid images;
  • the number of the customer images is counted to obtain the passenger flow corresponding to the video data.
  • FIG. 1 is a schematic flowchart of a passenger flow statistics method provided by an embodiment of this application
  • FIG. 2 is a schematic diagram of the internal structure of an electronic device provided by an embodiment of the application.
  • FIG. 3 is a schematic diagram of modules of a passenger flow statistics device provided by an embodiment of the application.
  • This application provides a method for counting passenger flow.
  • FIG. 1 it is a schematic flowchart of a passenger flow statistics method provided by an embodiment of this application.
  • the method can be executed by a device, and the device can be implemented by software and/or hardware.
  • the passenger flow statistics method includes step S10-step S40, specifically:
  • Step S10 Obtain video data used for passenger flow statistics.
  • the video data is preferably surveillance video data, which can be directly captured by monitoring equipment (such as a camera, etc.) arranged in a physical store in a passenger flow statistics area (such as a physical store).
  • monitoring equipment such as a camera, etc.
  • passenger flow statistics area such as a physical store.
  • the duration of the video data is not limited. It is determined according to the statistical demand of actual passenger flow.
  • the video data should preferably be omni-directional surveillance video data in the passenger flow statistics area, so as to better avoid the phenomenon of missing statistics.
  • Step S20 Perform humanoid recognition on each frame of image in the video data to identify multiple humanoid images.
  • each frame of image may be recognized through a human figure recognition technology to identify the plurality of human figure images.
  • humanoid recognition technology refers to the use of certain characteristics of human imaging, through the processing of graphics images, and finally discover the technology of identifying and positioning humanoid targets in the imaging space.
  • Step S30 Recognizing the humanoid image using the pedestrian re-recognition technology, and defining the humanoid image containing the preset clothing image as the clerk image, so as to distinguish the clerk image and the customer image from the plurality of humanoid images.
  • pedestrian re-identification is a technology that uses computer vision technology to identify whether there is a specific pedestrian in an image or video sequence.
  • the preset dress is used as the input of the pedestrian re-identification technology to identify a specific pedestrian wearing the preset dress in the video data.
  • the preset dress is the work clothes of the preset store clerk, so that the specific pedestrian wearing the preset dress identified in the video data is the store clerk, so as to realize from the plurality of humanoid images Distinguish the image of the clerk and the image of the customer.
  • Step S40 Count the number of the customer images to obtain the passenger flow corresponding to the video data.
  • the passenger flow statistics method proposed in this embodiment uses surveillance video data to count the number of customers, with high accuracy, and uses pedestrian re-identification technology to distinguish between shop assistants wearing preset dresses, so as to distinguish between shop assistants and customers, and avoid Staff statistics are included to ensure the high accuracy of passenger flow statistics.
  • the step of performing humanoid recognition on each frame of image in the video data to identify multiple humanoid images includes:
  • Step S21 Perform moving target detection on each frame of the video image to obtain a moving target area
  • Step S22 performing blob processing, area filtering, and aspect ratio filtering on the moving target area to obtain a humanoid candidate area
  • Step S23 Use a human shape classifier model to classify the human shape candidate regions to classify and recognize the plurality of human shape images.
  • the moving target detection in step S21 can be implemented by using existing moving target detection or moving target detection methods, including but not limited to one or a combination of the following methods: moving target detection based on background modeling, frame-based Moving target detection based on difference method, moving target detection based on optical flow method, etc.
  • moving target detection based on background modeling including but not limited to one or a combination of the following methods: moving target detection based on background modeling, frame-based Moving target detection based on difference method, moving target detection based on optical flow method, etc.
  • a video image is input, and a moving target detection method based on mixed Gaussian background modeling is adopted, and moving target detection is performed from each frame of the video image to obtain a moving target area.
  • a clump processing algorithm may be used to perform clumping and filtering processing on the moving target area, with the purpose of continuously converging the moving target area to make the movement
  • the target area tends to be human.
  • the humanoid classifier model can be trained in the following ways: first, sample images are randomly sampled, and the convolutional neural network is trained for the first time, and then according to the classifier model trained for the first time, the difficult negative samples are obtained, preferably According to the positive samples and the difficult negative samples, the first training classifier model is trained for the second time, so as to train the humanoid classifier model.
  • the step of using the pedestrian re-recognition technology to recognize the humanoid image, and defining the humanoid image containing the preset clothing image as the clerk image includes:
  • the current humanoid image is defined as the clerk image.
  • the step of matching and recognizing the humanoid image by using the preset dress image can be specifically implemented by the following steps:
  • the step of defining the current humanoid image as the clerk image can be specifically implemented by the following steps:
  • the current humanoid image is defined as the clerk image.
  • the threshold is greater than 90%, preferably 98%.
  • the method may further include:
  • the current humanoid image is defined as the clerk image
  • the preset dress image is not recognized in all the target humanoid images, then the next humanoid image is matched and recognized.
  • the preset number of frames is preferably 5 frames, that is, the same humanoid image wearing the preset clothing image appears in 5 consecutive frames, then the humanoid image is defined as an image of a clerk.
  • this embodiment can better avoid probabilistic risks and improve the accuracy of distinguishing the store clerk.
  • the method may further include:
  • the type includes one or more of age, gender, VIP membership, new customers and old customers. This can make the statistical data diversified, more referential and representative.
  • the step of classifying and recognizing the customer image can be specifically implemented through the following steps:
  • the face area is recognized, and the customer image is classified according to the recognition result. That is, the face can be used to distinguish age, gender, VIP members, new customers and regular customers.
  • the step of obtaining video data for passenger flow statistics includes:
  • the preset time is preferably 24 hours, that is, the passenger flow is counted once a day.
  • the method may further include:
  • the grayscale processing algorithm is:
  • F(i,j) 0.30*f R (i,j)+0.59*f G (i,j)+0.11*f B (i,j), F(i,j) is the grayscale processed
  • the pixel values, f R (i, j), f G (i, j), and f B (i, j) are the values of the R component, the G component, and the B component in the image before the grayscale processing, respectively.
  • the median filter algorithm is used to denoise each frame of image.
  • the principle of median filter is to replace the value of a pixel in the image with the median value of each pixel in a neighborhood of the pixel. , So that the surrounding pixel values are closer to the true value, thereby eliminating isolated noise points.
  • the method is to select the pixel area with the target pixel as the center, sort the pixel values of all the pixels in the pixel area in the order from largest to smallest or from smallest to largest, and select a value in the middle of the sorted sequence ( That is, the median) as the new pixel value of the target pixel.
  • the median filtering algorithm is:
  • g(x,y) med ⁇ f(xk,yi),(k,i ⁇ W), f(x,y) and g(x,y) are the pixel values of the image before and after filtering, respectively, med represents the median of multiple values, W is the area size of the pixel area selected with the pixel (x, y) as the center, k, i is the positional relationship of a pixel with respect to the pixel (x, y) , F(xk, yi) represents the pixel value of the pixel (xk, yi) in the pixel area.
  • the size of the pixel area is usually 3*3, or 5*5.
  • the application also provides an electronic device.
  • FIG. 2 it is a schematic diagram of the internal structure of an electronic device provided by an embodiment of this application.
  • the electronic device 1 may be a PC (Personal Computer, personal computer), or a terminal device such as a smart phone, a tablet computer, or a portable computer.
  • the electronic device 1 at least includes a memory 11, a processor 12, a network interface 13, and a communication bus 14.
  • the memory 11 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), magnetic memory, magnetic disk, optical disk, and the like.
  • the memory 11 may be an internal storage unit of the electronic device 1 in some embodiments, such as a hard disk of the electronic device 1.
  • the memory 11 may also be an external storage device of the electronic device 1, such as a plug-in hard disk equipped on the electronic device 1, a smart memory card (Smart Media Card, SMC), and a Secure Digital (SD) Card, Flash Card, etc.
  • the memory 11 may also include both an internal storage unit of the electronic device 1 and an external storage device.
  • the memory 11 can be used not only to store application software and various data installed in the electronic device 1, such as the code of the passenger flow counting program 01, etc., but also to temporarily store data that has been output or will be output.
  • the processor 12 may be a central processing unit (CPU), controller, microcontroller, microprocessor, or other data processing chip, for running program codes or processing stored in the memory 11 Data, such as the execution of passenger flow statistics program 01, etc.
  • CPU central processing unit
  • controller microcontroller
  • microprocessor microprocessor
  • other data processing chip for running program codes or processing stored in the memory 11 Data, such as the execution of passenger flow statistics program 01, etc.
  • the network interface 13 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface), and is usually used to establish a communication connection between the device 1 and other electronic equipment such as monitoring equipment.
  • a standard wired interface and a wireless interface such as a WI-FI interface
  • the communication bus 14 is used to realize the connection and communication between these components.
  • the electronic device 1 may also include a user interface.
  • the user interface may include a display (Display) and an input unit such as a keyboard (Keyboard).
  • the optional user interface may also include a standard wired interface and a wireless interface.
  • the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode, organic light-emitting diode) touch device, etc.
  • the display can also be appropriately called a display screen or a display unit, which is used to display the information processed in the electronic device 1 and to display a visualized user interface.
  • FIG. 2 only shows the electronic device 1 with components 11-14 and the passenger flow statistics program 01.
  • FIG. 1 does not constitute a limitation on the electronic device 1, and may include ratios Fewer or more parts are shown, or some parts are combined, or different parts are arranged.
  • the memory 11 stores the passenger flow statistics program 01; when the processor 12 executes the passenger flow statistics program 01 stored in the memory 11, the following steps are implemented:
  • Recognizing the humanoid image by adopting the pedestrian re-recognition technology, and defining the humanoid image containing the preset clothing image as the clerk image, so as to distinguish the clerk image and the customer image from the plurality of humanoid images;
  • the number of the customer images is counted to obtain the passenger flow corresponding to the video data.
  • the electronic device proposed in this embodiment uses surveillance video data to count the number of customers, with high accuracy, and uses pedestrian re-identification technology to distinguish between shop assistants wearing preset dresses to distinguish between shop assistants and customers, and avoid counting shop assistants Inside, to ensure the high accuracy of passenger flow statistics.
  • the passenger flow statistics program 01 may also be called by the processor to implement the following steps:
  • the human shape classifier model is used to classify the human shape candidate regions to classify and recognize the plurality of human shape images.
  • moving target detection you can use existing moving target detection or moving target detection methods to achieve, including but not limited to one or a combination of the following methods: moving target detection based on background modeling, motion based on frame difference method Target detection, moving target detection based on optical flow method, etc.
  • moving target detection based on background modeling motion based on frame difference method
  • Target detection moving target detection based on optical flow method
  • a video image is input, and a moving target detection method based on mixed Gaussian background modeling is adopted, and moving target detection is performed from each frame of the video image to obtain a moving target area.
  • a clump processing algorithm can be used to perform clumping and filtering processing on the moving target area, with the purpose of continuously converging the moving target area so that the moving target area Tend to human form.
  • the humanoid classifier model can be trained in the following ways: first, sample images are randomly sampled, and the convolutional neural network is trained for the first time, and then according to the classifier model trained for the first time, the difficult negative samples are obtained, preferably According to the positive samples and the difficult negative samples, the first training classifier model is trained for the second time, so as to train the humanoid classifier model.
  • the passenger flow statistics program 01 may also be called by the processor to implement the following steps:
  • the current humanoid image is defined as the clerk image.
  • the passenger flow statistics program 01 may also be called by the processor to implement the following steps:
  • the current humanoid image is defined as the clerk image.
  • the threshold is greater than 90%, preferably 98%.
  • the passenger flow statistics program 01 may also be called by the processor to implement the following steps:
  • the current humanoid image is defined as the clerk image
  • the preset dress image is not recognized in all the target humanoid images, then the next humanoid image is matched and recognized.
  • the preset number of frames is 5 frames, that is, the same humanoid image wearing the preset clothing image appears in the consecutive 5 frames of pictures, then the humanoid image is defined as a clerk image, compared to
  • this embodiment can better avoid probabilistic risks and improve the accuracy of distinguishing shop assistants.
  • the passenger flow statistics program 01 may also be called by the processor to implement the following steps:
  • the type includes one or more of age, gender, VIP membership, new customers and old customers.
  • the passenger flow statistics program 01 may also be called by the processor to implement the following steps:
  • the face area is recognized, and the customer image is classified according to the recognition result, that is, the face can be used to distinguish age, gender, VIP members, new customers and old customers.
  • the passenger flow statistics program 01 may also be called by the processor to implement the following steps:
  • the preset time is preferably 24 hours, that is, the passenger flow is counted once a day.
  • the passenger flow statistics program 01 may also be called by the processor to implement the following steps:
  • the grayscale processing algorithm is:
  • F(i,j) 0.30*f R (i,j)+0.59*f G (i,j)+0.11*f B (i,j), F(i,j) is the grayscale processed Pixel values, f R (i,j), f G (i,j), and f B (i,j) are the values of the R component, G component, and B component in the image before grayscale processing;
  • the median filter algorithm is used to denoise each frame of image.
  • the principle of median filter is to replace the value of a pixel in the image with the median value of each pixel in a neighborhood of the pixel. , To make the surrounding pixel values closer to the true value, thereby eliminating isolated noise points.
  • the method is to select the pixel area as the center of the target pixel, and sort the pixel values of all pixels in the pixel area in the order from largest to smallest or from smallest to largest, and select a value in the middle of the sequence ( That is, the median) as the new pixel value of the target pixel.
  • the median filter algorithm is:
  • g(x,y) med ⁇ f(xk,yi),(k,i ⁇ W), f(x,y) and g(x,y) are the pixel values of the image before and after filtering, respectively, med represents the median of multiple values, W is the area size of the pixel area selected with the pixel (x, y) as the center, k, i is the positional relationship of a pixel with respect to the pixel (x, y) , F(xk, yi) represents the pixel value of the pixel (xk, yi) in the pixel area.
  • the size of the pixel area is usually 3*3, or 5*5.
  • the passenger flow statistics program 01 may also be divided into one or more modules, and the one or more modules are stored in the memory 11 and run by one or more processors (this embodiment It is executed by the processor 12) to complete the present application.
  • the module referred to in the present application refers to a series of computer program instruction segments that can complete specific functions, and is used to describe the execution process of the passenger flow statistics program in the electronic device 1.
  • FIG. 3 this is a schematic diagram of modules in an embodiment of the passenger flow statistics apparatus 100 of this application.
  • the passenger flow statistics apparatus 100 can be divided into a data acquisition module 10, a human figure recognition module 20, and an image distinguishing module.
  • Module 30 and passenger flow statistics module 40 exemplarily:
  • the data acquisition module 10 is used for: acquiring video data for passenger flow statistics;
  • the human figure recognition module 20 is configured to: perform human figure recognition on each frame of image in the video data to identify multiple human figure images;
  • the image distinguishing module 30 is used for recognizing the humanoid image using the pedestrian re-recognition technology, and defining the humanoid image containing the preset dress image as the clerk image, so as to distinguish the clerk image and the customer from the plurality of humanoid images image;
  • the passenger flow statistics module 40 is configured to count the number of the customer images to obtain the passenger flow corresponding to the video data.
  • the embodiment of the present application also proposes a computer-readable storage medium.
  • the computer-readable storage medium may be non-volatile or volatile, and a passenger flow statistics program is stored on the computer-readable storage medium.
  • the passenger flow statistics program can be executed by one or more processors to achieve the following operations:
  • Recognizing the humanoid image by adopting the pedestrian re-recognition technology, and defining the humanoid image containing the preset clothing image as the clerk image, so as to distinguish the clerk image and the customer image from the plurality of humanoid images;
  • the number of the customer images is counted to obtain the passenger flow corresponding to the video data.
  • the passenger flow statistics method provided by this application further ensures the privacy and security of all the above-mentioned data
  • all the above-mentioned data can also be stored in a node of a blockchain.
  • video data, human images, etc. these data can be stored in the blockchain node.
  • the blockchain referred to in this application is a new application mode of computer technology such as distributed data storage, point-to-point transmission, consensus mechanism, and encryption algorithm.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • Human Computer Interaction (AREA)
  • Computational Linguistics (AREA)
  • Software Systems (AREA)
  • Image Analysis (AREA)

Abstract

一种客流量统计方法,该方法包括:获取用于客流量统计的视频数据(S10);对所述视频数据中的每一帧图像进行人形识别,以识别出多个人形图像(S20);采用行人重识别技术对所述人形图像进行识别,并将包含预设着装图像的人形图像定义为店员图像,以从所述多个人形图像中区分出店员图像和顾客图像(S30)。通过采用监控视频数据来统计顾客数,准确性高,并采用行人重识别技术对身穿预设着装的店员进行区分,以区分开店员和顾客,避免将店员统计在内,保证客流量统计的高准确性。

Description

一种客流量统计方法、装置、设备及计算机可读存储介质
本申请要求于2019年9月2日提交中国专利局、申请号为CN201910823327.1,发明名称为“一种客流量统计方法、装置及计算机可读存储介质”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
技术领域
本申请涉及零售数据处理技术领域,尤其涉及一种客流量统计方法、装置、设备及计算机可读存储介质。
背景技术
在零售中,对客流量的分析至关重要。具体而言,在实体店投资、创业等商业行为中,客流量与购买力均为非常重要的参数。然而,发明人意识到在现有的对客流量的统计方法中,往往仅能对少量参数进行笼统估计,例如通过购买记录推知进店顾客数,或者通过门店口处的进出感应器来统计顾客数,导致缺乏准确性。
发明内容
本申请提供一种客流量统计方法、装置、设备及计算机可读存储介质,以解决现有客流量统计缺乏准确性的技术问题。
一种客流量统计方法,所述方法包括如下步骤:
获取用于客流量统计的视频数据;
对所述视频数据中的每一帧图像进行人形识别,以识别出多个人形图像;
采用行人重识别技术对所述人形图像进行识别,并将包含预设着装图像的人形图像定义为店员图像,以从所述多个人形图像中区分出店员图像和顾客图像;
统计所述顾客图像的数量,以得到所述视频数据对应的客流量。
一种客流量统计装置,该装置包括:
数据获取模块,用于获取用于客流量统计的视频数据;
人形识别模块,用于对所述视频数据中的每一帧图像进行人形识别,以识别出多个人形图像;
图像区分模块,用于采用行人重识别技术对所述人形图像进行识别,并将包含预设着装图像的人形图像定义为店员图像,以从所述多个人形图像中区分出店员图像和顾客图像;
客流统计模块,用于统计所述顾客图像的数量,以得到所述视频数据对应的客流量。
一种电子设备,该电子设备包括:存储器、处理器,所述存储器中存储有客流量统计程序,所述客流量统计程序被所述处理器执行时实现如下步骤:
获取用于客流量统计的视频数据;
对所述视频数据中的每一帧图像进行人形识别,以识别出多个人形图像;
采用行人重识别技术对所述人形图像进行识别,并将包含预设着装图像的人形图像定义为店员图像,以从所述多个人形图像中区分出店员图像和顾客图像;
统计所述顾客图像的数量,以得到所述视频数据对应的客流量。
一种计算机可读存储介质,所述计算机可读存储介质上存储有客流量统计程序,所述客流量统计程序可被一个或者多个处理器执行,实现如下步骤:
获取用于客流量统计的视频数据;
对所述视频数据中的每一帧图像进行人形识别,以识别出多个人形图像;
采用行人重识别技术对所述人形图像进行识别,并将包含预设着装图像的人形图像定义为店员图像,以从所述多个人形图像中区分出店员图像和顾客图像;
统计所述顾客图像的数量,以得到所述视频数据对应的客流量。
附图说明
图1为本申请一实施例提供的客流量统计方法的流程示意图;
图2为本申请一实施例提供的电子设备的内部结构示意图;
图3为本申请一实施例提供的客流量统计装置的模块示意图。
本申请目的的实现、功能特点及优点将结合实施例,参照附图做进一步说明。
具体实施方式
应当理解,此处所描述的具体实施例仅仅用以解释本申请,并不用于限定本申请。
本申请提供一种客流量统计方法。参照图1所示,为本申请一实施例提供的客流量统计方法的流程示意图。该方法可以由一个装置执行,该装置可以由软件和/或硬件实现。
在本实施例中,客流量统计方法包括步骤S10-步骤S40,具体地:
步骤S10,获取用于客流量统计的视频数据。
其中,所述的视频数据优选为监控视频数据,可以通过布置在客流量统计区域(如实体店)实体店的监控设备(如摄像头等)直接拍摄得到,该视频数据的时长不做限制,可根据实际客流量的统计需求情况来定。此外,所述的视频数据最好为客流量统计区域的全方位监控视频数据,这样可以较好的避免出现漏统计的现象。
步骤S20,对所述视频数据中的每一帧图像进行人形识别,以识别出多个人形图像。
具体地,在本实施例当中,步骤S20可以通过人形识别技术来对每一帧图像进行识别,以识别出所述多个人形图像。其中,人形识别技术是指利用人体成像的一定特征,通过对图形图像的处理,最终在成像空间中发现识别和定位人形目标的技术。
步骤S30,采用行人重识别技术对所述人形图像进行识别,并将包含预设着装图像的人形图像定义为店员图像,以从所述多个人形图像中区分出店员图像和顾客图像。
其中,行人重识别技术(Person re-identification,简称reid)也称行人再识别,是利用计算机视觉技术识别图像或者视频序列中是否存在特定行人的技术。本实施例采用预设着装作为行人重识别技术的输入,以在所述的视频数据中识别出穿着预设着装的特定行人。在本实施例当中,所述预设着装为预设的店员的工作服,这样在所述视频数据中识别出的穿着预设着装的特定行人即为店员,进而实现从所述多个人形图像中区分出店员图像和顾客图像。
步骤S40,统计所述顾客图像的数量,以得到所述视频数据对应的客流量。
本实施例提出的客流量统计方法,通过采用监控视频数据来统计顾客数,准确性高,并采用行人重识别技术对身穿预设着装的店员进行区分,以区分开店员和顾客,避免将店员统计在内,保证客流量统计的高准确性。
进一步地,在本申请方法的另一实施例中,所述的对所述视频数据中的每一帧图像进行人形识别,以识别出多个人形图像的步骤包括:
步骤S21,对所述视频图像中的每一帧图像进行运动目标检测,获取运动目标区域;
步骤S22,对所述运动目标区域进行团块处理、面积过滤和宽高比过滤,获取人形候选区域;
步骤S23,采用人形分类器模型对所述人形候选区域进行分类,以分类识别出所述多个人形图像。
在步骤S21中的运动目标检测,可以采用现有的运动目标检测或者移动目标检测方法来实现,包括但不限于以下一种或者多种方法的组合:基于背景建模的运动目标检测、基于帧差法的运动目标检测、基于光流法的运动目标检测等。示例性地,输入视频图像,采用基于混合高斯背景建模的运动目标检测方法,从所述视频图像中的每一帧图像进行运动目标检测,以获取运动目标区域。
在步骤S22中,可采用团块处理算法、面积过滤算法和宽高比过滤算法来对所述运动目标区域进行团块及过滤处理,目的在于不断收敛所述运动目标区域,以使所述运动目标区域趋于人形。
其中,所述人形分类器模型可以通过以下方式来训练:首先对样本图像进行随机采样,并对卷积神经网络进行初次训练,然后根据初次训练的分类器模型,获取难例负样本,最好根据正样本和难例负样本对初次训练的分类器模型进行二次训练,从而训练出所述人形分类器模型。
进一步地,在本申请方法的另一实施例中,所述的采用行人重识别技术对所述人形图像进行识别,并将包含预设着装图像的人形图像定义为店员图像的步骤包括:
采用所述预设着装图像对所述人形图像进行匹配识别;
当在任一当前人形图像中识别到所述预设着装图像时,将所述当前人形图像定义为所述店员图像。
在本实施例当中,所述的采用所述预设着装图像对所述人形图像进行匹配识别的步骤具体可以通过以下步骤来实现:
将所述预设着装图像与所述人形图像的每一区域或预设区域进行相似度匹配;则
所述的当在任一当前人形图像中识别到所述预设着装图像时,将所述当前人形图像定义为所述店员图像的步骤具体可以通过以下步骤来实现:
当在任一当前人形图像中存在图像区域与预设着装图像的相似度大于阈值,将所述当前人形图像定义为所述店员图像。
其中,所述阈值大于90%,优选为98%。
进一步地,在本申请方法的另一实施例中,在所述的将所述当前人形图像定义为所述店员图像的步骤之前,所述方法还可以包括:
基于行人追踪技术,确定出所述当前人形图像分别在连续的预设帧数的图像中对应的目标人形图像;
判断所有所述目标人形图像中是否均识别到所述预设着装图像;
若判断到所有所述目标人形图像中均识别到所述预设着装图像,则将所述当前人形图像定义为所述店员图像;
若判断到所有所述目标人形图像中未均识别到所述预设着装图像,则对下一个人形图像进行匹配识别。
其中,所述预设帧数优选为5帧,即在连续的5帧图片中都出现了同一个身穿所述预设着装图像的人形图像,则将该人形图像定义为店员图像,相比于通过单帧图片来判定店员图像的方案,本实施例可以较好的规避概率风险,提高店员区分的准确性。
进一步地,在本申请方法的另一实施例中,所述方法还可以包括:
对所述顾客图像进行分类识别,并统计每一类型的顾客图像的数量;
其中,所述类型包括年龄、性别、VIP会员、新顾客及老顾客当中的一种或多种。这样可以使统计数据多样化,更具有参考性和代表性。
在本实施例当中,所述的对所述顾客图像进行分类识别的步骤具体可以通过以下步骤来实现:
获取所述顾客图像的人脸区域
对所述人脸区域进行识别,并根据识别结果对所述顾客图像进行分类。即可以通过人脸来区分年龄、性别、VIP会员、新顾客及老顾客。
进一步地,在本申请方法的另一实施例中,所述的获取用于客流量统计的视频数据的步骤包括:
每隔预设时间,从视频采集设备中获取一次所述视频数据。其中,所述预设时间优选为24h,即每天统计一次客流量。
进一步地,为了降低数据处理量,在所述的获取用于客流量统计的视频数据的步骤之后,所述方法还可以包括:
对所述视频数据中的每一帧图像进行灰度化和去噪处理。
其中,灰度化处理算法为:
F(i,j)=0.30*f R(i,j)+0.59*f G(i,j)+0.11*f B(i,j),F(i,j)为灰度化处理后的像素值,f R(i,j)、f G(i,j)、f B(i,j)分别为灰度化处理前的图像中的R分量、G分量及B分量的值。
其中,采用中值滤波算法对每一帧图像进行去噪处理,中值滤波的原理是把图像中一像素点的值用该像素点的一个邻域中各像素点的像素值的中值代替,让周围的像素值更接近真实值,从而消除孤立的噪声点。方法是以目标像素点为中心选取像素点区域,将该像素点区域内的所有像素点的像素值按照从大到小或者从小到大的顺序进行排序,选择排序得的序列中间的一个值(即中值)作为目标像素点的新的像素值。
其中,中值滤波算法为:
g(x,y)=med{f(x-k,y-i),(k,i∈W),f(x,y)及g(x,y)分别为滤波前和滤波后的图像的像素值,med表示取多个值的中值,W为以像素点(x,y)为中心选取的像素点区域的区域大小,k,i为一个像素点相对于像素点(x,y)的位置关系,f(x-k,y-i)表示以像素点区域内的像素点(x-k,y-i)的像素值。
其中,像素点区域的大小通常为3*3,或者5*5。
本申请还提供一种电子设备。参照图2所示,为本申请一实施例提供的电子设备的内部结构示意图。
在本实施例中,电子设备1可以是PC(Personal Computer,个人电脑),也可以是智能手机、平板电脑、便携计算机等终端设备。该电子设备1至少包括存储器11、处理器12,网络接口13,以及通信总线14。
其中,存储器11至少包括一种类型的可读存储介质,所述可读存储介质包括闪存、硬盘、多媒体卡、卡型存储器(例如,SD或DX存储器等)、磁性存储器、磁盘、光盘等。存储器11在一些实施例中可以是电子设备1的内部存储单元,例如该电子设备1的硬盘。存储器11在另一些实施例中也可以是电子设备1的外部存储设备,例如电子设备1上配备的插接式硬盘,智能存储卡(Smart Media Card,SMC),安全数字(Secure Digital,SD)卡,闪存卡(Flash Card)等。进一步地,存储器11还可以既包括电子设备1的内部存储单元也包括外部存储设备。存储器11不仅可以用于存储安装于电子设备1的应用软件及各类数据,例如客流量统计程序01的代码等,还可以用于暂时地存储已经输出或者将要输出的数据。
处理器12在一些实施例中可以是一中央处理器(Central Processing Unit,CPU)、控制器、微控制器、微处理器或其他数据处理芯片,用于运行存储器11中存储的程序代码或处理数据,例如执行客流量统计程序01等。
网络接口13可选的可以包括标准的有线接口、无线接口(如WI-FI接口),通常用于在该装置1与监控设备等其他电子设备之间建立通信连接。
通信总线14用于实现这些组件之间的连接通信。
可选地,该电子设备1还可以包括用户接口,用户接口可以包括显示器(Display)、输入单元比如键盘(Keyboard),可选的用户接口还可以包括标准的有线接口、无线接口。可选地,在一些实施例中,显示器可以是LED显示器、液晶显示器、触控式液晶显示器以及OLED(Organic Light-Emitting Diode,有机发光二极管)触摸器等。其中,显示器也可以适当的称为显示屏或显示单元,用于显示在电子设备1中处理的信息以及用于显示可视化的用户界面。
图2仅示出了具有组件11-14以及客流量统计程序01的电子设备1,本领域技术人员 可以理解的是,图1示出的结构并不构成对电子设备1的限定,可以包括比图示更少或者更多的部件,或者组合某些部件,或者不同的部件布置。
在图2所示的电子设备1实施例中,存储器11中存储有客流量统计程序01;处理器12执行存储器11中存储的客流量统计程序01时实现如下步骤:
获取用于客流量统计的视频数据;
对所述视频数据中的每一帧图像进行人形识别,以识别出多个人形图像;
采用行人重识别技术对所述人形图像进行识别,并将包含预设着装图像的人形图像定义为店员图像,以从所述多个人形图像中区分出店员图像和顾客图像;
统计所述顾客图像的数量,以得到所述视频数据对应的客流量。
本实施例提出的电子设备,通过采用监控视频数据来统计顾客数,准确性高,并采用行人重识别技术对身穿预设着装的店员进行区分,以区分开店员和顾客,避免将店员统计在内,保证客流量统计的高准确性。
进一步地,在本申请电子设备的另一实施例中,客流量统计程序01还可被处理器调用,以实现如下步骤:
对所述视频图像中的每一帧图像进行运动目标检测,获取运动目标区域;
对所述运动目标区域进行团块处理、面积过滤和宽高比过滤,获取人形候选区域;
采用人形分类器模型对所述人形候选区域进行分类,以分类识别出所述多个人形图像。
对于运动目标检测,可以采用现有的运动目标检测或者移动目标检测方法来实现,包括但不限于以下一种或者多种方法的组合:基于背景建模的运动目标检测、基于帧差法的运动目标检测、基于光流法的运动目标检测等。示例性地,输入视频图像,采用基于混合高斯背景建模的运动目标检测方法,从所述视频图像中的每一帧图像进行运动目标检测,以获取运动目标区域。
具体地,可采用团块处理算法、面积过滤算法和宽高比过滤算法来对所述运动目标区域进行团块及过滤处理,目的在于不断收敛所述运动目标区域,以使所述运动目标区域趋于人形。
其中,所述人形分类器模型可以通过以下方式来训练:首先对样本图像进行随机采样,并对卷积神经网络进行初次训练,然后根据初次训练的分类器模型,获取难例负样本,最好根据正样本和难例负样本对初次训练的分类器模型进行二次训练,从而训练出所述人形分类器模型。
进一步地,在本申请电子设备的另一实施例中,客流量统计程序01还可被处理器调用,以实现如下步骤:
采用所述预设着装图像对所述人形图像进行匹配识别;
当在任一当前人形图像中识别到所述预设着装图像时,将所述当前人形图像定义为所述店员图像。
进一步地,在本申请电子设备的另一实施例中,客流量统计程序01还可被处理器调用,以实现如下步骤:
将所述预设着装图像与所述人形图像的每一区域或预设区域进行相似度匹配;以及
当在任一当前人形图像中存在图像区域与预设着装图像的相似度大于阈值,将所述当前人形图像定义为所述店员图像。
其中,所述阈值大于90%,优选为98%。
进一步地,在本申请电子设备的另一实施例中,客流量统计程序01还可被处理器调用,以实现如下步骤:
基于行人追踪技术,确定出所述当前人形图像分别在连续的预设帧数的图像中对应的目标人形图像;
判断所有所述目标人形图像中是否均识别到所述预设着装图像;
若判断到所有所述目标人形图像中均识别到所述预设着装图像,则将所述当前人形图像定义为所述店员图像;
若判断到所有所述目标人形图像中未均识别到所述预设着装图像,则对下一个人形图像进行匹配识别。
其中,所述预设帧数为5帧,即在连续的5帧图片中都出现了同一个身穿所述预设着装图像的人形图像,则将该人形图像定义为店员图像,相比于通过单帧图片来判定店员图像的方案,本实施例可以较好的规避概率风险,提高店员区分的准确性。
进一步地,在本申请电子设备的另一实施例中,客流量统计程序01还可被处理器调用,以实现如下步骤:
对所述顾客图像进行分类识别,并统计每一类型的顾客图像的数量;
其中,所述类型包括年龄、性别、VIP会员、新顾客及老顾客当中的一种或多种。
进一步地,在本申请电子设备的另一实施例中,客流量统计程序01还可被处理器调用,以实现如下步骤:
获取所述顾客图像的人脸区域
对所述人脸区域进行识别,并根据识别结果对所述顾客图像进行分类,即可以通过人脸来区分年龄、性别、VIP会员、新顾客及老顾客。
进一步地,在本申请电子设备的另一实施例中,客流量统计程序01还可被处理器调用,以实现如下步骤:
每隔预设时间,从视频采集设备中获取一次所述视频数据。其中,所述预设时间优选为24h,即每天统计一次客流量。
进一步地,在本申请电子设备的另一实施例中,客流量统计程序01还可被处理器调用,以实现如下步骤:
对所述视频数据中的每一帧图像进行灰度化和去噪处理。
其中,灰度化处理算法为:
F(i,j)=0.30*f R(i,j)+0.59*f G(i,j)+0.11*f B(i,j),F(i,j)为灰度化处理后的像素值,f R(i,j)、f G(i,j)、f B(i,j)分别为灰度化处理前的图像中的R分量、G分量及B分量的值;
其中,采用中值滤波算法对每一帧图像进行去噪处理,中值滤波的原理是把图像中一像素点的值用该像素点的一个邻域中各像素点的像素值的中值代替,让周围的像素值更接近真实值,从而消除孤立的噪声点。方法是以目标像素点为中心选取像素点区域,将该像素点区域内的所有像素点的像素值按照从大到小或者从小到大的顺序进行排序,选择排序得的序列中间的一个值(即中值)作为目标像素点的新的像素值。
中值滤波算法为:
g(x,y)=med{f(x-k,y-i),(k,i∈W),f(x,y)及g(x,y)分别为滤波前和滤波后的图像的像素值,med表示取多个值的中值,W为以像素点(x,y)为中心选取的像素点区域的区域大小,k,i为一个像素点相对于像素点(x,y)的位置关系,f(x-k,y-i)表示以像素点区域内的像素点(x-k,y-i)的像素值。
其中,像素点区域的大小通常为3*3,或者5*5。
可选地,在其他实施例中,客流量统计程序01还可以被分割为一个或者多个模块,一个或者多个模块被存储于存储器11中,并由一个或多个处理器(本实施例为处理器12)所执行以完成本申请,本申请所称的模块是指能够完成特定功能的一系列计算机程序指令段,用于描述客流量统计程序在电子设备1中的执行过程。
例如,参照图3所示,为本申请客流量统计装置100一实施例中的模块示意图,该实施例中,客流量统计装置100可以被分割为数据获取模块10、人形识别模块20、图像区分模块30及客流统计模块40,示例性地:
数据获取模块10用于:获取用于客流量统计的视频数据;
人形识别模块20用于:对所述视频数据中的每一帧图像进行人形识别,以识别出多个人形图像;
图像区分模块30用于:采用行人重识别技术对所述人形图像进行识别,并将包含预设着装图像的人形图像定义为店员图像,以从所述多个人形图像中区分出店员图像和顾客图像;
客流统计模块40用于:统计所述顾客图像的数量,以得到所述视频数据对应的客流量。
上述数据获取模块10、人形识别模块20、图像区分模块30及客流统计模块40等程序模块被执行时所实现的功能或操作步骤与上述实施例大体相同,在此不再赘述。
此外,本申请实施例还提出一种计算机可读存储介质,所述计算机可读存储介质可以是非易失性,也可以是易失性,所述计算机可读存储介质上存储有客流量统计程序,所述客流量统计程序可被一个或多个处理器执行,以实现如下操作:
获取用于客流量统计的视频数据;
对所述视频数据中的每一帧图像进行人形识别,以识别出多个人形图像;
采用行人重识别技术对所述人形图像进行识别,并将包含预设着装图像的人形图像定义为店员图像,以从所述多个人形图像中区分出店员图像和顾客图像;
统计所述顾客图像的数量,以得到所述视频数据对应的客流量。
本申请计算机可读存储介质具体实施方式与上述客流量统计装置和方法各实施例基本相同,在此不作累述。
在另一实施例中,本申请所提供的客流量统计方法,为进一步保证上述所有出现的数据的私密和安全性,上述所有数据还可以存储于一区块链的节点中。例如视频数据、人形图像等等,这些数据均可存储在区块链节点中。
需要说明的是,本申请所指区块链是分布式数据存储、点对点传输、共识机制、加密算法等计算机技术的新型应用模式。
需要说明的是,上述本申请实施例序号仅仅为了描述,不代表实施例的优劣。并且本文中的术语“包括”或者其任何其他变体意在涵盖非排他性的包含,从而使得包括一系列要素的过程、装置、物品或者方法不仅包括那些要素,而且还包括没有明确列出的其他要素,或者是还包括为这种过程、装置、物品或者方法所固有的要素。在没有更多限制的情况下,由语句“包括一……”限定的要素,并不排除在包括该要素的过程、装置、物品或者方法中还存在另外的相同要素。
通过以上的实施方式的描述,本领域的技术人员可以清楚地了解到上述实施例方法可借助软件加必需的通用硬件平台的方式来实现,当然也可以通过硬件,但很多情况下前者是更佳的实施方式。基于这样的理解,本申请的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来,该计算机软件产品存储在如上所述的一个存储介质(如ROM/RAM、磁碟、光盘)中,包括若干指令用以使得一台终端设备(可以是手机,计算机,服务器,或者网络设备等)执行本申请各个实施例所述的方法。
以上仅为本申请的优选实施例,并非因此限制本申请的专利范围,凡是利用本申请说明书及附图内容所作的等效结构或等效流程变换,或直接或间接运用在其他相关的技术领域,均同理包括在本申请的专利保护范围内。

Claims (20)

  1. 一种客流量统计方法,其中,所述方法包括如下步骤:
    获取用于客流量统计的视频数据;
    对所述视频数据中的每一帧图像进行人形识别,以识别出多个人形图像;
    采用行人重识别技术对所述人形图像进行识别,并将包含预设着装图像的人形图像定义为店员图像,以从所述多个人形图像中区分出店员图像和顾客图像;
    统计所述顾客图像的数量,以得到所述视频数据对应的客流量。
  2. 如权利要求1所述的客流量统计方法,其中,所述的对所述视频数据中的每一帧图像进行人形识别,以识别出多个人形图像的步骤包括:
    对所述视频图像中的每一帧图像进行运动目标检测,获取运动目标区域;
    对所述运动目标区域进行团块处理、面积过滤和宽高比过滤,获取人形候选区域;
    采用人形分类器模型对所述人形候选区域进行分类,以分类识别出所述多个人形图像。
  3. 如权利要求1所述的客流量统计方法,其中,所述的采用行人重识别技术对所述人形图像进行识别,并将包含预设着装图像的人形图像定义为店员图像的步骤包括:
    采用所述预设着装图像对所述人形图像进行匹配识别;
    当在任一当前人形图像中识别到所述预设着装图像时,将所述当前人形图像定义为所述店员图像。
  4. 如权利要求3所述的客流量统计方法,其中,所述的采用所述预设着装图像对所述人形图像进行匹配识别的步骤包括:
    将所述预设着装图像与所述人形图像的每一区域或预设区域进行相似度匹配;则
    所述的当在任一当前人形图像中识别到所述预设着装图像时,将所述当前人形图像定义为所述店员图像的步骤包括:
    当在任一当前人形图像中存在图像区域与预设着装图像的相似度大于阈值,将所述当前人形图像定义为所述店员图像。
  5. 如权利要求3或4所述的客流量统计方法,其中,在所述的将所述当前人形图像定义为所述店员图像的步骤之前,还包括:
    基于行人追踪技术,确定出所述当前人形图像分别在连续的预设帧数的图像中对应的目标人形图像;
    判断所有所述目标人形图像中是否均识别到所述预设着装图像;
    若判断到所有所述目标人形图像中均识别到所述预设着装图像,则将所述当前人形图像定义为所述店员图像。
  6. 如权利要求5所述的客流量统计方法,其中,所述预设帧数为5帧。
  7. 如权利要求1所述的客流量统计方法,其中,所述方法还包括:
    对所述顾客图像进行分类识别,并统计每一类型的顾客图像的数量;
    其中,所述类型包括年龄、性别、VIP会员、新顾客及老顾客当中的一种或多种。
  8. 如权利要求7所述的客流量统计方法,其中,所述的对所述顾客图像进行分类识别的步骤包括:
    获取所述顾客图像的人脸区域
    对所述人脸区域进行识别,并根据识别结果对所述顾客图像进行分类。
  9. 一种客流量统计装置,其中,所述装置包括:
    数据获取模块,用于获取用于客流量统计的视频数据;
    人形识别模块,用于对所述视频数据中的每一帧图像进行人形识别,以识别出多个人形图像;
    图像区分模块,用于采用行人重识别技术对所述人形图像进行识别,并将包含预设着 装图像的人形图像定义为店员图像,以从所述多个人形图像中区分出店员图像和顾客图像;
    客流统计模块,用于统计所述顾客图像的数量,以得到所述视频数据对应的客流量。
  10. 一种电子设备,其中,所述电子设备包括:存储器、处理器,所述存储器中存储有客流量统计程序,所述客流量统计程序被所述处理器执行时实现如下步骤:
    获取用于客流量统计的视频数据;
    对所述视频数据中的每一帧图像进行人形识别,以识别出多个人形图像;
    采用行人重识别技术对所述人形图像进行识别,并将包含预设着装图像的人形图像定义为店员图像,以从所述多个人形图像中区分出店员图像和顾客图像;
    统计所述顾客图像的数量,以得到所述视频数据对应的客流量。
  11. 如权利要求10所述的电子设备,其中,所述的对所述视频数据中的每一帧图像进行人形识别,以识别出多个人形图像的步骤包括:
    对所述视频图像中的每一帧图像进行运动目标检测,获取运动目标区域;
    对所述运动目标区域进行团块处理、面积过滤和宽高比过滤,获取人形候选区域;
    采用人形分类器模型对所述人形候选区域进行分类,以分类识别出所述多个人形图像。
  12. 如权利要求10所述的电子设备,其中,所述的采用行人重识别技术对所述人形图像进行识别,并将包含预设着装图像的人形图像定义为店员图像的步骤包括:
    采用所述预设着装图像对所述人形图像进行匹配识别;
    当在任一当前人形图像中识别到所述预设着装图像时,将所述当前人形图像定义为所述店员图像。
  13. 如权利要求12所述的电子设备,其中,所述的采用所述预设着装图像对所述人形图像进行匹配识别的步骤包括:
    将所述预设着装图像与所述人形图像的每一区域或预设区域进行相似度匹配;则
    所述的当在任一当前人形图像中识别到所述预设着装图像时,将所述当前人形图像定义为所述店员图像的步骤包括:
    当在任一当前人形图像中存在图像区域与预设着装图像的相似度大于阈值,将所述当前人形图像定义为所述店员图像。
  14. 如权利要求12或13所述的电子设备,其中,在所述的将所述当前人形图像定义为所述店员图像的步骤之前,所述客流量统计程序被所述处理器执行时还实现如下步骤:
    基于行人追踪技术,确定出所述当前人形图像分别在连续的预设帧数的图像中对应的目标人形图像;
    判断所有所述目标人形图像中是否均识别到所述预设着装图像;
    若判断到所有所述目标人形图像中均识别到所述预设着装图像,则将所述当前人形图像定义为所述店员图像。
  15. 如权利要求14所述的电子设备,其中,所述预设帧数为5帧。
  16. 一种计算机可读存储介质,其中,所述计算机可读存储介质上存储有客流量统计程序,所述客流量统计程序可被一个或者多个处理器执行,实现如下步骤:
    获取用于客流量统计的视频数据;
    对所述视频数据中的每一帧图像进行人形识别,以识别出多个人形图像;
    采用行人重识别技术对所述人形图像进行识别,并将包含预设着装图像的人形图像定义为店员图像,以从所述多个人形图像中区分出店员图像和顾客图像;
    统计所述顾客图像的数量,以得到所述视频数据对应的客流量。
  17. 如权利要求16所述的计算机可读存储介质,其中,所述的对所述视频数据中的每一帧图像进行人形识别,以识别出多个人形图像的步骤包括:
    对所述视频图像中的每一帧图像进行运动目标检测,获取运动目标区域;
    对所述运动目标区域进行团块处理、面积过滤和宽高比过滤,获取人形候选区域;
    采用人形分类器模型对所述人形候选区域进行分类,以分类识别出所述多个人形图像。
  18. 如权利要求16所述的计算机可读存储介质,其中,所述的采用行人重识别技术对所述人形图像进行识别,并将包含预设着装图像的人形图像定义为店员图像的步骤包括:
    采用所述预设着装图像对所述人形图像进行匹配识别;
    当在任一当前人形图像中识别到所述预设着装图像时,将所述当前人形图像定义为所述店员图像。
  19. 如权利要求18所述的计算机可读存储介质,其中,所述的采用所述预设着装图像对所述人形图像进行匹配识别的步骤包括:
    将所述预设着装图像与所述人形图像的每一区域或预设区域进行相似度匹配;则
    所述的当在任一当前人形图像中识别到所述预设着装图像时,将所述当前人形图像定义为所述店员图像的步骤包括:
    当在任一当前人形图像中存在图像区域与预设着装图像的相似度大于阈值,将所述当前人形图像定义为所述店员图像。
  20. 如权利要求18或19所述的计算机可读存储介质,其中,在所述的将所述当前人形图像定义为所述店员图像的步骤之前,所述客流量统计程序可被一个或者多个处理器执行,还实现如下步骤:
    基于行人追踪技术,确定出所述当前人形图像分别在连续的预设帧数的图像中对应的目标人形图像;
    判断所有所述目标人形图像中是否均识别到所述预设着装图像;
    若判断到所有所述目标人形图像中均识别到所述预设着装图像,则将所述当前人形图像定义为所述店员图像。
PCT/CN2020/112338 2019-09-02 2020-08-30 一种客流量统计方法、装置、设备及计算机可读存储介质 WO2021043089A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910823327.1A CN110659588A (zh) 2019-09-02 2019-09-02 一种客流量统计方法、装置及计算机可读存储介质
CN201910823327.1 2019-09-02

Publications (1)

Publication Number Publication Date
WO2021043089A1 true WO2021043089A1 (zh) 2021-03-11

Family

ID=69037900

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/112338 WO2021043089A1 (zh) 2019-09-02 2020-08-30 一种客流量统计方法、装置、设备及计算机可读存储介质

Country Status (2)

Country Link
CN (1) CN110659588A (zh)
WO (1) WO2021043089A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114937241A (zh) * 2022-06-01 2022-08-23 北京凯利时科技有限公司 基于过渡区的客流统计方法和系统以及计算机程序产品
WO2023029574A1 (zh) * 2021-08-30 2023-03-09 上海商汤智能科技有限公司 获取客流信息的方法、装置、计算机设备以及存储介质

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110659588A (zh) * 2019-09-02 2020-01-07 平安科技(深圳)有限公司 一种客流量统计方法、装置及计算机可读存储介质
CN112232262A (zh) * 2020-10-27 2021-01-15 上海依图网络科技有限公司 一种客流量统计方法及装置

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106845620A (zh) * 2016-12-19 2017-06-13 江苏慧眼数据科技股份有限公司 一种基于队列状态分析的客流计数方法
CN108846724A (zh) * 2018-06-06 2018-11-20 北京京东尚科信息技术有限公司 数据分析方法和系统
US20190019019A1 (en) * 2017-07-14 2019-01-17 Panasonic Corporation People stream analysis method, people stream analysis apparatus, and people stream analysis system
CN109272347A (zh) * 2018-08-16 2019-01-25 苏宁易购集团股份有限公司 一种门店客流量的统计分析方法及系统
CN109508586A (zh) * 2017-09-15 2019-03-22 杭州海康威视数字技术股份有限公司 一种客流统计方法、装置及设备
CN109902551A (zh) * 2018-11-09 2019-06-18 阿里巴巴集团控股有限公司 开放场景的实时人流统计方法和装置
CN110659588A (zh) * 2019-09-02 2020-01-07 平安科技(深圳)有限公司 一种客流量统计方法、装置及计算机可读存储介质

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106295788B (zh) * 2015-05-12 2019-01-08 杭州海康威视数字技术股份有限公司 客流量的统计方法及装置
CN108597065A (zh) * 2018-03-12 2018-09-28 南京甄视智能科技有限公司 基于人脸识别的客流统计方法
CN108764181B (zh) * 2018-05-31 2021-06-08 北京学之途网络科技有限公司 一种客流统计方法和装置、计算机可读存储介质
CN109448026A (zh) * 2018-11-16 2019-03-08 南京甄视智能科技有限公司 基于头肩检测的客流统计方法和系统

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106845620A (zh) * 2016-12-19 2017-06-13 江苏慧眼数据科技股份有限公司 一种基于队列状态分析的客流计数方法
US20190019019A1 (en) * 2017-07-14 2019-01-17 Panasonic Corporation People stream analysis method, people stream analysis apparatus, and people stream analysis system
CN109508586A (zh) * 2017-09-15 2019-03-22 杭州海康威视数字技术股份有限公司 一种客流统计方法、装置及设备
CN108846724A (zh) * 2018-06-06 2018-11-20 北京京东尚科信息技术有限公司 数据分析方法和系统
CN109272347A (zh) * 2018-08-16 2019-01-25 苏宁易购集团股份有限公司 一种门店客流量的统计分析方法及系统
CN109902551A (zh) * 2018-11-09 2019-06-18 阿里巴巴集团控股有限公司 开放场景的实时人流统计方法和装置
CN110659588A (zh) * 2019-09-02 2020-01-07 平安科技(深圳)有限公司 一种客流量统计方法、装置及计算机可读存储介质

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2023029574A1 (zh) * 2021-08-30 2023-03-09 上海商汤智能科技有限公司 获取客流信息的方法、装置、计算机设备以及存储介质
CN114937241A (zh) * 2022-06-01 2022-08-23 北京凯利时科技有限公司 基于过渡区的客流统计方法和系统以及计算机程序产品
CN114937241B (zh) * 2022-06-01 2024-03-26 北京凯利时科技有限公司 基于过渡区的客流统计方法和系统以及计算机程序产品

Also Published As

Publication number Publication date
CN110659588A (zh) 2020-01-07

Similar Documents

Publication Publication Date Title
WO2021043089A1 (zh) 一种客流量统计方法、装置、设备及计算机可读存储介质
US9996737B2 (en) Method and system for automatically recognizing facial expressions via algorithmic periocular localization
CN108229314B (zh) 目标人物的搜索方法、装置和电子设备
Hu et al. Dense crowd counting from still images with convolutional neural networks
Radman et al. Automated segmentation of iris images acquired in an unconstrained environment using HOG-SVM and GrowCut
Günther et al. Unconstrained face detection and open-set face recognition challenge
US9536153B2 (en) Methods and systems for goods received gesture recognition
Tran et al. Activity analysis in crowded environments using social cues for group discovery and human interaction modeling
TW201737134A (zh) 用於藉由機器學習訓練物件分類器之系統及方法
US20190258866A1 (en) Human presence detection in edge devices
WO2019033569A1 (zh) 眼球动作分析方法、装置及存储介质
WO2018210047A1 (zh) 数据处理方法、数据处理装置、电子设备及存储介质
WO2021043090A1 (zh) 顾客数量的统计方法及装置、电子设备及可读存储介质
Lee et al. Context and profile based cascade classifier for efficient people detection and safety care system
WO2019033570A1 (zh) 嘴唇动作分析方法、装置及存储介质
CN110222582B (zh) 一种图像处理方法和相机
WO2021051547A1 (zh) 暴力行为检测方法及系统
US20160078323A1 (en) Method and apparatus for counting person
WO2022156317A1 (zh) 视频帧处理方法及装置、电子设备和存储介质
CN111860309A (zh) 一种人脸识别方法及系统
US10592687B2 (en) Method and system of enforcing privacy policies for mobile sensory devices
CN111340569A (zh) 基于跨境追踪的门店人流分析方法、装置、系统、终端和介质
Ryan et al. Scene invariant crowd counting
JP5552946B2 (ja) 顔画像サンプル採取装置、顔画像サンプル採取方法、プログラム
Yanakova et al. Facial recognition technology on ELcore semantic processors for smart cameras

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20861598

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20861598

Country of ref document: EP

Kind code of ref document: A1