WO2020048376A1 - 一种货架分析方法、装置、系统及电子设备 - Google Patents

一种货架分析方法、装置、系统及电子设备 Download PDF

Info

Publication number
WO2020048376A1
WO2020048376A1 PCT/CN2019/103291 CN2019103291W WO2020048376A1 WO 2020048376 A1 WO2020048376 A1 WO 2020048376A1 CN 2019103291 W CN2019103291 W CN 2019103291W WO 2020048376 A1 WO2020048376 A1 WO 2020048376A1
Authority
WO
WIPO (PCT)
Prior art keywords
shelf
product
label
target
target object
Prior art date
Application number
PCT/CN2019/103291
Other languages
English (en)
French (fr)
Inventor
朱皓
童俊艳
任烨
Original Assignee
杭州海康威视数字技术股份有限公司
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 杭州海康威视数字技术股份有限公司 filed Critical 杭州海康威视数字技术股份有限公司
Publication of WO2020048376A1 publication Critical patent/WO2020048376A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • 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
    • G06Q30/06—Buying, selling or leasing transactions

Definitions

  • the present application relates to the field of image processing technology, and in particular, to a shelf analysis method, device, system, and electronic device.
  • Shelves are widely used in retail stores to display and display goods. For example, in a large supermarket, all goods are placed on shelves to facilitate customers' purchase of goods.
  • the shelves need to be analyzed. For example, it is necessary to analyze whether the placement area of various goods on the shelf is accurate, whether the various goods on the shelf are out of stock, and whether the paste area of the various product labels pasted on the shelf is accurate.
  • shelves are usually analyzed manually. For example, for each type of goods placed on the shelves, the staff manually checks whether the actual placement area of the goods on the shelves is the same as the predetermined placement area of the goods. Obviously, the manual analysis of the shelves by the staff is inefficient. So, how to analyze shelves quickly and effectively is an urgent problem.
  • the purpose of the embodiments of the present application is to provide a shelf analysis method, device, system, and electronic equipment, so as to realize rapid and effective analysis of shelves. Specific technical solutions are as follows:
  • an embodiment of the present application provides a shelf analysis method, where the method includes:
  • the target shelf image includes an image area corresponding to the shelf to be analyzed
  • a shelf analysis result corresponding to the shelf to be analyzed is determined.
  • an embodiment of the present application provides a shelf analysis device, where the device includes:
  • An image acquisition module configured to obtain a target shelf image; wherein the target shelf image includes an image area corresponding to a shelf to be analyzed;
  • An object recognition module is configured to identify each target object belonging to the target object category in the target shelf image and obtain attribute information of each target object, wherein the target object category is: a type of a shelf analysis result to be determined The corresponding object category;
  • a shelf analysis module is configured to determine a shelf analysis result corresponding to the shelf to be analyzed based on the attribute information of each target object.
  • an embodiment of the present application provides a shelf analysis system, where the system includes:
  • the image acquisition device is configured to collect a target shelf image and send the collected target shelf image to a server; wherein the target shelf image includes an image area corresponding to a shelf to be analyzed;
  • a server is configured to obtain a target shelf image from an image acquisition device; identify each target object belonging to the target object category in the target shelf image, and obtain attribute information of each target object, wherein the target object category is: An object category corresponding to the type of the determined shelf analysis result; and based on the attribute information of each target object, a shelf analysis result corresponding to the shelf is determined.
  • an embodiment of the present application provides an electronic device including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory complete communication with each other through the communication bus;
  • the processor is configured to implement the shelf analysis method according to the first aspect when executing a program stored in the memory.
  • an embodiment of the present application provides a computer-readable storage medium.
  • the computer-readable storage medium stores a computer program, and the computer program implements the shelf analysis method according to the first aspect when executed by a processor. .
  • an embodiment of the present application provides an executable program code, where the executable program code is used to be executed to execute the shelf analysis method according to the first aspect.
  • the technical solution provided in the embodiment of the present application obtains a target shelf image; wherein the target shelf image includes an image area corresponding to the shelf to be analyzed; identifying each target object belonging to the target object category in the target shelf image and obtaining attribute information of each target object
  • the target object category is: an object category corresponding to the type of shelf analysis result to be determined; and based on attribute information of each target object, a shelf analysis result corresponding to the shelf to be analyzed is determined.
  • FIG. 1 is a flowchart of a shelf analysis method according to an embodiment of the present application
  • FIG. 2 is a schematic diagram of a target shelf image according to an embodiment of the present application.
  • FIG. 3 is a schematic diagram of an installation manner of an image acquisition device that collects a target shelf image according to an embodiment of the present application
  • FIG. 4 is a schematic diagram of a target trellis diagram provided by an embodiment of the present application.
  • FIG. 5 is a schematic diagram of a target shelf image including a promotion label according to an embodiment of the present application.
  • FIG. 6 is a schematic structural diagram of a shelf analysis device according to an embodiment of the present application.
  • FIG. 7 is a schematic diagram of a shelf analysis system according to an embodiment of the present application.
  • FIG. 8 is a schematic diagram of an electronic device according to an embodiment of the present application.
  • embodiments of the present application provide a shelf analysis method, device, system and electronic equipment.
  • the execution subject of a shelf analysis method may be a shelf analysis device, and the shelf analysis device may be run in a shelf analysis system for data processing equipment.
  • the rack analysis system may include a server and an image acquisition device for acquiring a target shelf image of a shelf to be analyzed.
  • the rack analysis device may be run on the server to take pictures based on the image acquisition device.
  • the shelf analysis system can also include only: image acquisition equipment, at this time, the shelf analysis device can be run on the image acquisition equipment, then the image acquisition equipment is shooting the shelf to be analyzed After the target shelf image is obtained, a shelf analysis result can be obtained based on the target shelf image.
  • the image acquisition device may be a camera in the form of a camera.
  • a shelf analysis method provided in an embodiment of the present application may include the following steps:
  • the target shelf image includes an image area corresponding to the shelf to be analyzed.
  • a target shelf image captured by an image acquisition device may be acquired, and the target shelf image includes an image area corresponding to the shelf to be analyzed.
  • any shelf image captured by the image acquisition device may be the target shelf image described in the embodiment of the present application.
  • the target shelf image may be the shelf image shown in FIG. 2.
  • the shelf image may include: the products displayed on the shelf, the product labels pasted on the shelf, etc.
  • FIG. 2 is only The shelf image is schematically shown by way of example, and the content included in the shelf image is not specifically limited in this application.
  • the image acquisition device can be installed in various manners. In specific applications, as shown in FIG. 3, the image acquisition device can be installed in a manner of being hoisted or embedded in a shelf, which is not limited to this.
  • An image acquisition device installed in a ceiling manner may be referred to as a suspended image acquisition device, and a camera installed in a shelf-mounted manner may be referred to as an image acquisition device embedded in a shelf.
  • the image acquisition device When the installation mode of the image acquisition device is hoisting, the image acquisition device can be suspended on the roof of the room where the shelf to be analyzed is located, and the image acquisition device can collect the target shelf image from top to bottom.
  • the image acquisition device can collect the target shelf image from top to bottom.
  • the image acquisition device When the installation mode of the image acquisition device is embedded in a shelf, the image acquisition device is embedded in a floor of a certain shelf. At this time, the image acquisition device can also capture the target shelf image. There are also many advantages to installing the image acquisition device embedded in the shelf. For example, the perspective distortion of the shelf image captured by the image acquisition device is small; and because the image acquisition device is embedded in the shelf, the image acquisition device does not affect the room where the shelf is located Appearance.
  • the embodiment of the present application does not specifically limit the installation manner of the image acquisition equipment; and the number of image acquisition equipment may be determined according to the actual situation, and the embodiment of the application does not specifically limit the number of image acquisition equipment.
  • the image acquisition device can capture the target shelf image in real time, and can also capture the target shelf image at a preset sampling interval, which is not specifically limited in the implementation of this application.
  • S120 Identify each target object belonging to the target object category in the target shelf image to obtain attribute information of each target object, where the target object category is: an object category corresponding to the type of shelf analysis result to be determined.
  • a pre-trained algorithm model can be used to identify each target object belonging to the target object category in the target shelf image, and obtain attribute information of each target object.
  • the target object category may be: product tag category, promotion label category, goods category, personnel category, etc.
  • the target object may be product tag, promotion label, goods, personnel, etc.
  • the attributes of each target object may be: position information of the target object, area size of the target object, and the like. The embodiment of the present application does not specifically limit the target object category, the target object, and the attribute information of the target object.
  • the target object category is the object category corresponding to the type of the shelf analysis result to be determined, that is, the type of the product analysis result determines the object category that needs to be identified, and when the type of the shelf analysis result is determined Next, the type of object to be identified is determined.
  • the required attribute information may be different or the same during shelf analysis.
  • the target object category may be: the product label category.
  • the target object Each target object corresponding to the category is a goods label, and the attribute information of each target object may be: location information of the goods label.
  • the target object categories can be: promotion label category and goods label category.
  • each target object corresponding to the target object category is : Promotion labels and product labels.
  • the attribute information of each target object can be: location information and promotion information of each promotional label, and position information of each product label and the product identification of the product indicated by each product label.
  • the target object category can be: the product category and the product label category.
  • each target object corresponding to the target object category is: the product and the product label.
  • the attribute information of each target object It can be: position information of each item and position information of each item label.
  • the target object categories may be: goods categories and goods label categories.
  • each target object corresponding to the target object categories is: goods and product tags, and attribute information of each target object. It can be: position information of each item and position information of each item label.
  • the target object category may be: the personnel category.
  • each target object corresponding to the target object category is: personnel, and the attribute information of each target object may be: number of personnel.
  • S130 Determine a shelf analysis result corresponding to the shelf to be analyzed based on the attribute information of each target object.
  • the target object category is an object category corresponding to the type of shelf analysis result to be determined
  • the shelf analysis result corresponding to the shelf to be analyzed can be determined based on the attribute information of each target object.
  • the specific process of determining the shelf analysis result corresponding to the shelf to be analyzed is different based on the attribute information of each target object.
  • the technical solution provided in the embodiment of the present application obtains a target shelf image; wherein the target shelf image includes an image area corresponding to the shelf to be analyzed; identifying each target object belonging to the target object category in the target shelf image and obtaining attribute information of each target object
  • the target object category is: an object category corresponding to the type of shelf analysis result to be determined; and based on attribute information of each target object, a shelf analysis result corresponding to the shelf to be analyzed is determined.
  • the type of shelf analysis result to be determined may be fixed, and at this time, the target object category may be fixed.
  • the type of the shelf analysis result to be determined can be manually specified, or the shelf analysis device determines the type of the shelf analysis result to be determined according to a preset rule.
  • the preset rule may be: according to a correspondence between a time point / time period and a type of a shelf analysis result, which is not limited to this, of course.
  • a target object category corresponding to the type of the shelf analysis result to be determined is determined.
  • the type of the shelf analysis result to be determined may be obtained; and, based on a preset mapping relationship between the type of the shelf analysis result type and the object category, it is determined that Determine the target object category corresponding to the type of shelf analysis result.
  • the mapping relationship between the types of shelf analysis results and object categories can be: whether the product display accurately corresponds to: the product label category; whether the promotional label accurately corresponds to: the promotional label category and the product label category; whether the product is out of stock In: Goods category and goods label category; the heat information of the goods corresponds to: goods category and goods label category; the personnel's heat information corresponds to: personnel category.
  • this merely describes the mapping relationship between the type of the shelf analysis result and the object category by way of example, and should not constitute a limitation on the embodiment of the present application.
  • the target object category may be determined as: the product label category. At this time, You can identify only the product labels in the target shelf image.
  • the following describes the specific process of identifying each target object belonging to the target object category in the target shelf image, obtaining the attribute information of each target object, and the attributes based on each target object in combination with each type of shelf analysis results to be determined. Information to determine the specific process of shelf analysis results corresponding to the shelf to be analyzed.
  • the type of the shelf analysis result to be determined includes: whether the product display is accurate, and correspondingly, the target object category includes: a product label category.
  • the step of identifying each target object belonging to the target object category in the target shelf image and obtaining attribute information of each target object may include:
  • the above product labels may be: paper display boards or electronic display boards used to display information such as product names and prices;
  • a pre-trained neural network model for identifying the position information of the product labels can be used to identify each product label in the target shelf image and obtain the position information of each product label.
  • the type, structure and training process of the neural network model are not limited here.
  • step S130 based on the attribute information of each target object, the step of determining a shelf analysis result corresponding to the shelf to be analyzed may include the following steps a1-a4:
  • the step of determining the layer number and column number of each item label in the shelf to be analyzed based on the position information of each item label may include:
  • the position information of each product label corresponding to each layer number is horizontally projected from small to large or from large to small according to the abscissa information to obtain the column number of each product label in the shelf to be analyzed.
  • the step of determining the product identification of the product indicated by the product label may include:
  • the so-called label content for identifying the product label may specifically be: identifying the text or barcode in the image area where the product label is located, and obtaining the product identification of the product indicated by the product label.
  • the step of determining the item identification of the item indicated by the item label may include:
  • the first position information that meets the first screening condition is determined, and the product identifier of the product corresponding to the first position information is used as the product identifier of the product indicated by the product label, where ,
  • the first filtering condition is: the corresponding area is closest to the area corresponding to the position information of the product label.
  • the area corresponding to the first position information is closest to the area corresponding to the position information of the product label, indicating that the product corresponding to the first position information is the product indicated by the product label. Therefore, the first position
  • the product identification of the product corresponding to the information is the same as the product identification of the product indicated by the product label, and the product identification of the product corresponding to the first position information may be used as the product identification of the product indicated by the product label.
  • the preset generation method is: each product label corresponds to a shed in the target trellis diagram, and the identification of the shed corresponding to any product label is the product identification of the product indicated by the product label, and any of the product identifications
  • the layer number and column number of the corresponding shelf in the shelf chart are the same as the layer number and column number of the goods label in the shelf to be analyzed.
  • the target shelf image of the target shelf image can be generated.
  • the trellis diagram of any shelf refers to the topology of the shelf layout, which is used to indicate the product category and position relationship of each layer and each column.
  • the target shelf image is the shelf image shown in FIG. 2.
  • the topmost layer (layer number 1) of the shelf image has three types of products. From left to right, the product identifiers of these three types of products are in order: A, B, C; There are two types of products displayed in the middle layer (layer number 2) of the shelf image. From left to right, the product identifiers of these two types of products are D and E in turn. There are three types of products displayed on the layer (layer number 3). From left to right, the product identifiers of these three types of products are F, G, and H in this order.
  • the target shelf image includes eight product tags.
  • the position information of the eight product tags and the product identifiers of the products indicated by the eight product tags are respectively A, B, C, D, E, F, G, and H.
  • the position information of the eight product tags is vertically projected from small to large (from top to bottom) according to the vertical coordinate to obtain the layer number of each product tag in the shelf to be analyzed. It can be understood that among the eight product tags, , The layer number of 3 product labels is 1, the layer number of 2 product labels is 2, and the layer numbers of the remaining 3 product labels are 3.
  • the three product labels with the layer number 1 are horizontally projected from small to large (from left to right) according to the abscissa information, and the 3 with the layer number 1 is obtained.
  • the column numbers of each product label are 1, 2, and 3; the two product labels with the layer number 2 are horizontally projected from small to large according to the abscissa information, and the column numbers of the two product labels with the layer number 2 are respectively Is 1, 2; the three item labels with the layer number 3 are horizontally projected from small to large according to the abscissa information, and the column numbers of the three item labels with the layer number 3 are 1, 2, and 3, respectively.
  • the target shelf map corresponding to the target shelf image can be drawn, and the drawn target shelf
  • the figure may be a trellis diagram as shown in FIG. 4.
  • the target shed grid After the target shed grid is drawn, the target shed grid can be compared with a preset standard shed grid to obtain a comparison result, and based on the obtained comparison result, it is determined whether the display of the goods on the shelf to be analyzed is accurate.
  • the step of determining whether the display of the goods on the shelf to be analyzed is accurate based on a comparison result of the target shelf chart and a preset standard shelf chart may include:
  • the standard shed corresponding to the shed is: the shed in the standard shed grid with the same floor number and column number as the shed.
  • the identification of a shed in the target shed graph is A
  • the standard shed corresponding to the shed is B, indicating that the display of the goods on the shelf to be analyzed is not accurate.
  • the product labels on the shelf to be analyzed can also be determined. Whether it is lost. For example, if the comparison result between the target shed chart and the preset standard shed chart is: the number of layers in the target shed chart is different from the number of layers in the standard shed chart; or the number of columns in the target shed chart is in line with the standard shed chart If the number of columns is different, it can be determined that the product labels on the shelves to be analyzed are missing.
  • the terminal sends an alarm message.
  • the terminal associated with the worker may be a mobile phone, a computer, etc.
  • the content of the alarm information may be a simple alarm sound, or it may be information that the display of the goods carrying the shelves to be analyzed is inaccurate or the goods labels on the shelves to be analyzed are missing;
  • the form of the alarm information may be voice, text message, email, etc. The content and form of the alarm information are not specifically limited in the embodiment of the present application.
  • the type of the shelf analysis result to be determined includes: whether the promotion label is accurate; the target object category includes: a promotion label category and a product label category.
  • the step of identifying each target object belonging to the target object category in the target shelf image and obtaining attribute information of each target object may include:
  • Identify each promotional label and each product label in the target shelf image to obtain the position information and promotional information of each promotional label, as well as the position information of each product label and the product identification of the product indicated by each product label.
  • a pre-trained neural network model for identifying the position information of the product labels can be used to identify each promotional label and each product label in the target shelf image, to obtain the position information and promotional information of each promotional label, and each product label. Location information and the item identification of each item indicated on the item label.
  • the type, structure and training process of the neural network model are not limited here.
  • step S130 based on the attribute information of each target object, the step of determining the shelf analysis result corresponding to the shelf to be analyzed may include the following two steps, which are b1 and b2, respectively:
  • For each promotional label determine, from the position information of each product label, the second position information that meets the second screening condition, and use the product identifier of the product label corresponding to the second position information as the product corresponding to the promotional label. Identification; wherein the second screening condition is: the corresponding area is closest to the area corresponding to the position information of the promotion tag;
  • the promotion information specified by the goal is: The specified promotion information associated with the product identification corresponding to the promotion label.
  • the product identification of the product label can be used as the product identification corresponding to the promotion label. Any implementation manner capable of calculating the distance between the areas corresponding to the two position information can be applied to the embodiments of the present application.
  • promotion information of the promotion label is 20%. It is assumed that the identified product corresponding to the promotion label is A, but The designated promotional information associated with the identifier A is 30%. It can be seen that 20% is different from 30%, that is, the promotional information of the promotional label does not match the formulated promotional information associated with the product identification corresponding to the promotional label, so it can be determined The promotional label is inaccurate.
  • the promotional label is missing or not. For example, if the specified promotion information associated with a certain product identifier is 50%, and the promotion information corresponding to the product identifier is not recognized in the target shelf image, at this time, the promotion label may be judged to be missing.
  • an alarm message can be sent to the terminal associated with the staff, and it will not be repeated here. To repeat.
  • the type of shelf analysis result to be determined includes: whether the product is out of stock;
  • Audience categories include: product categories and product label categories.
  • the step of identifying each target object belonging to the target object category in the target shelf image and obtaining attribute information of each target object may include:
  • a pre-trained neural network model for identifying the position information of the product labels can be used to identify each product and each product label in the target shelf image to obtain the position information of each product and the position information of each product label.
  • the type, structure and training process of the neural network model are not limited here.
  • step S130 based on the attribute information of each target object, the step of determining a shelf analysis result corresponding to the shelf to be analyzed may include the following steps c1 and c2:
  • the step of calculating the designated storage area of the goods indicated by each goods label based on the position information of each goods label may include:
  • the bottom left vertex and the bottom right vertex of the designated storage area of the product indicated by the item label are calculated. Apex and the predetermined height value of each layer to determine the designated storage area of the goods indicated by the goods label;
  • the reference object corresponding to any product label is: an adjacent product label located at the same horizontal position as the product label, or an area on the shelf edge of the shelf to be analyzed at the same horizontal position with the product label.
  • the position information of the product label and its adjacent product label can be used Information to determine the lower left vertex and the lower right vertex of the designated storage area of the product indicated by the product label.
  • the top left vertex of the product label can be used as the bottom left vertex of the designated storage area, and the product adjacent to the product label will be The top left vertex of the label is used as the bottom right vertex of the specified storage area.
  • the position information of the product label and the position information of the reference object can also be used to determine The lower left vertex and the lower right vertex of the designated storage area of the goods indicated by the product label.
  • the position information of the product label is different, and the vertex at the lower left and the vertex at the lower right of the designated storage area indicated by the determined product label are also different. This embodiment of the present application does not specifically limit this.
  • the third position information that meets the third screening condition from the position information of each product, calculate the sum of the areas of the areas corresponding to the third position information, and calculate the sum of the area and the product label.
  • the ratio of the area of the corresponding designated storage area is used as the storage ratio of the product corresponding to the product label, and it is determined whether the storage ratio is less than the preset storage ratio. If so, determine that the product corresponding to the product label is out of stock, otherwise, determine that The product corresponding to the product label is not out of stock; the third screening condition is: the corresponding area is located in the designated storage area of the product indicated by the product label.
  • the sum of the area occupied by the goods in the designated storage area can be calculated, and the sum of the area occupied by the goods and the designated area can be calculated.
  • the ratio of the area of the storage area is used as the storage ratio of the product corresponding to the product label; and whether the storage ratio is less than the preset storage ratio. If the storage ratio is less than the preset storage ratio, the product label indicates There are fewer products in the designated storage area of the product. At this time, it can be determined that the product corresponding to the product label is out of stock. If the storage ratio is not less than the preset storage ratio, it indicates that the designated storage area of the product indicated by the product label is There are many more products. At this time, it can be determined that the product corresponding to the product label is not out of stock.
  • the size of the preset storage ratio can be set according to actual conditions, such as 0%, 10%, etc.
  • the embodiment of the present application does not specifically limit the size of the preset storage ratio.
  • an alarm message may be sent to a terminal associated with the staff, which is not repeated here.
  • the type of the shelf analysis result to be determined includes: hotness information of the goods;
  • the target object categories include: product categories and product label categories;
  • the step of obtaining the target shelf image may include:
  • the step of identifying each target object belonging to the target object category in the target shelf image and obtaining attribute information of each target object may include:
  • For each target shelf image identify each product and each product label in the target shelf image, and obtain position information of each product and position information of each product label.
  • the preset time period may be minutes, hours, days, etc.
  • the embodiment of the present application does not specifically limit the preset time period.
  • step S130 based on the attribute information of each target object, determining the shelf analysis result corresponding to the shelf to be analyzed may include the following steps d1-d3:
  • step c2 how to determine the storage ratio of the goods corresponding to each product tag based on the position information of each product and the position information of each product label in the target shelf image has been described in detail, and will not be repeated here. To repeat.
  • the sum of the differences corresponding to the product label is used as the product heat of the product corresponding to the product label.
  • the three target shelf images collected within one hour are target shelf image 1, target shelf image 2, and target shelf image 3.
  • the product label 1 in the target shelf image 1 The storage ratio of the corresponding product is 90%; the storage ratio of the product corresponding to the product label 1 in the target shelf image 2 is 70%; the storage ratio of the product corresponding to the product label 1 in the target shelf image 3 is 60%; the target shelf image
  • the difference between the storage ratio of the product corresponding to the product label 1 in 1 and the storage ratio of the product corresponding to the product label 1 in the target shelf image 2 is 20%; the storage ratio of the product corresponding to the product label 1 in the target shelf image 2 and the target shelf image
  • the difference between the storage ratios of the products corresponding to the product label 1 in 3 is 10%; the difference between the two storage ratios obtained by the calculation is added to obtain the product heat of the product corresponding to the product label 1 is 30%.
  • the type of shelf analysis result to be determined includes: personnel's heat information;
  • the target object categories include: people categories;
  • the steps to obtain the target shelf image include:
  • the steps of identifying each target object belonging to the target object category in the target shelf image and obtaining attribute information of each target object include:
  • For each target shelf image identify each person in the target shelf image, and obtain the number of persons in the target shelf image.
  • the preset time period may be minutes, hours, days, etc.
  • the embodiment of the present application does not specifically limit the preset time period.
  • S130 based on the attribute information of each target object, determining the shelf analysis result corresponding to the shelf to be analyzed may include the following two steps, which are e1 and e2, respectively:
  • the ratio of the sum of the number of persons obtained to the number of images of multiple target shelf images is calculated as the person's popularity information. For example, 3 target shelf images collected within one hour are obtained: target shelf image 1, target shelf image 2 and target shelf image 3, where the number of persons contained in target shelf image 1 is 6; target The number of persons contained in shelf image 2 is 9; the number of persons contained in target shelf image 3 is 3; the sum of the number of persons included in these three target shelf images is 18; the resulting person is calculated The ratio of the sum of the number to the number of images of the target shelf image is 6, then the person's popularity information is 6.
  • the ratio of the heat information of the goods to the heat information of the personnel can also be calculated, that is, the conversion rate of the goods can be obtained.
  • the types of shelf analysis results to be determined in the embodiments of the present application may include only one type, and may also include multiple types. That is, one or more of the following can be analyzed whether the display of the goods is accurate, whether the promotional labels are accurate, whether the goods are out of stock, the heat information of the goods, and the heat information of the personnel, which is not specifically limited in this embodiment of the present application. Moreover, it is reasonable to output the shelf analysis results in the form of a report.
  • an embodiment of the present application provides a shelf analysis device. As shown in FIG. 6, the device includes:
  • An image acquisition module 610 is configured to obtain a target shelf image, where the target shelf image includes an image area corresponding to a shelf to be analyzed;
  • An object recognition module 620 is configured to identify each target object belonging to the target object category in the target shelf image, and obtain attribute information of each target object, where the target object category is: The type of object corresponding to the type;
  • a shelf analysis module 630 is configured to determine a shelf analysis result corresponding to the shelf to be analyzed based on the attribute information of each target object.
  • the technical solution provided in the embodiment of the present application obtains a target shelf image; wherein the target shelf image includes an image area corresponding to the shelf to be analyzed; identifying each target object belonging to the target object category in the target shelf image and obtaining attribute information of each target object
  • the target object category is: an object category corresponding to the type of shelf analysis result to be determined; and based on attribute information of each target object, a shelf analysis result corresponding to the shelf to be analyzed is determined.
  • the device may further include:
  • a type determination module configured to identify each target object belonging to the target object category in the target shelf image in the target recognition module, and obtain the type of shelf analysis result to be determined before obtaining attribute information of each target object;
  • a target object category corresponding to the type of the shelf analysis result to be determined is determined.
  • the types of shelf analysis results to be determined include: whether the display of goods is accurate;
  • the target object category includes: a product tag category
  • the object recognition module is specifically configured to:
  • the shelf analysis module includes:
  • the layer number and column number determination submodule is used to determine the layer number and column number of each item label in the shelf to be analyzed based on the position information of each item label;
  • a first product identification determination sub-module configured to determine, for each product label, the product identification of the product indicated by the product label
  • Shed chart generation submodule is used to generate the target shelf image based on the preset generation method based on the layer number and column number of each product label in the shelf to be analyzed, and the product identification of the product indicated by each product label.
  • the target trellis diagram wherein the preset generation method is: each product label corresponds to a trellis in the target trellis diagram, and the identification of the trellis corresponding to any product label is the product indicated by the product label
  • the layer and column numbers of the shelf in the shelf chart corresponding to any of the labels are the same as the layer and column numbers of the label on the shelf to be analyzed;
  • the goods display determination sub-module is configured to determine whether the goods display of the shelf to be analyzed is accurate based on a comparison result of the target shed chart and a preset standard shed chart.
  • the product display determination sub-module is specifically used for:
  • the standard shelf corresponding to the shelf is: the shelf in the standard shelf map with the same floor number and column number as the shelf.
  • the layer number and column number determine the submodule, which are specifically used for:
  • the position information of each product label corresponding to each layer number is horizontally projected from small to large or from large to small according to the abscissa information to obtain the column number of each product label in the shelf to be analyzed.
  • the goods identification determining sub-module is specifically used for:
  • For each product label identify the label content of the product label, and obtain the product identification of the product indicated by the product label.
  • the goods identification determining sub-module is specifically used for:
  • the first position information that meets the first screening condition is determined, and the product identifier of the product corresponding to the first position information is used as the product identifier of the product indicated by the product label.
  • the first screening condition is: the corresponding area is closest to the area corresponding to the position information of the product tag.
  • the type of the shelf analysis result to be determined includes: whether the promotion label is accurate;
  • the target object categories include: promotion tag categories and product tag categories;
  • the object recognition module is specifically configured to:
  • Identify each promotional label and each product label in the target shelf image obtain position information and promotional information of each promotional label, and position information of each product label and a product identifier of the product indicated by each product label.
  • the shelf analysis module includes:
  • a second product identification determining sub-module for each promotional label, determining, from the position information of each product label, second position information that meets a second screening condition, and identifying the product identification of the product label corresponding to the second position information, As the product identification corresponding to the promotion label; wherein the second filtering condition is: the corresponding area is closest to the area corresponding to the position information of the promotion label;
  • a promotion label determination sub-module is used to determine, for each promotion label, whether the promotion information of the promotion label matches the target specified promotion information. If it matches, it is determined that the promotion label is accurate; otherwise, it is determined that the promotion label is inaccurate;
  • the target specified promotion information is specified promotion information associated with the product identifier corresponding to the promotion label.
  • the type of the shelf analysis result to be determined includes: whether the product is out of stock;
  • the target object categories include: goods categories and goods label categories;
  • the object recognition module is specifically configured to:
  • the shelf analysis module includes:
  • a storage area determination sub-module for calculating a specified storage area of the goods indicated by each of the goods labels based on the position information of each of the goods labels;
  • the out-of-stock determination sub-module is used to determine, for each product label, the third position information that meets the third screening condition from the position information of each product, calculates the sum of the areas of the areas corresponding to the third position information, and calculates The ratio of the sum of the area to the area of the designated storage area corresponding to the product label is used as the storage ratio of the product corresponding to the product label, and it is determined whether the storage ratio is less than a preset storage ratio. If so, determine the product label The corresponding product is out of stock, otherwise, it is determined that the product corresponding to the product label is not out of stock; wherein the third screening condition is that the corresponding area is located in the designated storage area of the product indicated by the product label.
  • the storage area determining submodule is specifically configured to:
  • the bottom left vertex and the bottom right vertex of the designated storage area of the product indicated by the item label are calculated. Apex and the predetermined height value of each layer to determine the designated storage area of the goods indicated by the goods label;
  • the reference object corresponding to any product label is: an adjacent product label located at the same horizontal position as the product label, or an area on the shelf edge of the shelf at the same horizontal position as the product label.
  • the type of the shelf analysis result to be determined includes: heat information of the goods;
  • the target object categories include: goods categories and goods label categories;
  • the image acquisition module is specifically configured to:
  • the object recognition module is specifically configured to:
  • For each target shelf image identify each product and each product label in the target shelf image, and obtain position information of each product and position information of each product label.
  • the shelf analysis module includes:
  • a storage ratio determination submodule configured to determine, for each target shelf image, a storage ratio of each product label based on the position information of each item in the target shelf image and the position information of each item label;
  • a storage ratio difference determination sub-module for calculating a difference between storage ratios of goods corresponding to a same product label for two adjacent target shelf images
  • the product heat determination sub-module is configured to, for each product label, sum the respective differences corresponding to the product label as the product heat of the product corresponding to the product label.
  • the type of the shelf analysis result to be determined includes: personnel's heat information;
  • the target object category includes: a personnel category
  • the image acquisition module is specifically configured to:
  • the object recognition module is specifically configured to:
  • For each target shelf image identify each person in the target shelf image, and obtain the number of persons in the target shelf image.
  • the shelf analysis module includes:
  • the number of persons calculation sub-module is used to calculate the sum of the number of persons of the persons included in each target shelf image
  • the personnel heat determination sub-module is configured to use the ratio of the sum of the calculated number of people and the number of images of multiple target shelf images as the personnel's heat information.
  • an embodiment of the present application provides a shelf analysis system. As shown in FIG. 7, the system includes:
  • the image acquisition device 710 is configured to collect a target shelf image and send the collected target shelf image to a server; wherein the target shelf image includes an image area corresponding to a shelf to be analyzed;
  • the server 720 is configured to obtain a target shelf image from an image acquisition device; identify each target object belonging to the target object category in the target shelf image, and obtain attribute information of each target object, where the target object category is: and The object category corresponding to the type of the shelf analysis result to be determined; and based on the attribute information of each target object, a shelf analysis result corresponding to the shelf to be analyzed is determined.
  • the technical solution provided by the embodiment of the present invention obtains a target shelf image; wherein the target shelf image includes an image area corresponding to the shelf to be analyzed; identifying each target object belonging to the target object category in the target shelf image and obtaining attribute information of each target object
  • the target object category is: an object category corresponding to the type of shelf analysis result to be determined; and based on attribute information of each target object, a shelf analysis result corresponding to the shelf to be analyzed is determined.
  • the server is further configured to:
  • the object recognition module identifies each target object belonging to the target object category in the target shelf image and obtains attribute information of each target object, obtaining a type of shelf analysis result to be determined;
  • a target object category corresponding to the type of the shelf analysis result to be determined is determined.
  • the type of the shelf analysis result to be determined includes: whether the display of the goods is accurate;
  • the target object category includes: a product tag category
  • the server identifying each target object belonging to the target object category in the target shelf image, and obtaining attribute information of each target object is specifically:
  • the determining a shelf analysis result corresponding to the shelf to be analyzed based on the attribute information of each target object is specifically:
  • a target shelf image of the target shelf image is generated according to a preset generation method, where the The preset generation method is as follows: each product label corresponds to a shed in the target trellis diagram, and the identifier of the shed corresponding to any of the product labels is the product identifier of the product indicated by the product label, and any of the product identifiers
  • the floor and column numbers of the corresponding shed in the shed graph are the same as the layer and column numbers of the goods label in the shelf to be analyzed;
  • the server determines whether the display of the goods on the shelf to be analyzed is accurate based on a comparison result between the target trellis diagram and a preset standard trellis diagram, specifically:
  • the standard shelf corresponding to the shelf is: the shelf in the standard shelf map with the same floor number and column number as the shelf.
  • the server determines the layer number and column number of each item label in the shelf to be analyzed based on the position information of each item label, specifically:
  • the position information of each product label corresponding to each layer number is horizontally projected from small to large or from large to small according to the abscissa information to obtain the column number of each product label in the shelf to be analyzed.
  • the server determines, for each product label, the product identifier of the product indicated by the product label, which is specifically:
  • For each product label identify the label content of the product label, and obtain the product identification of the product indicated by the product label.
  • the server determines, for each product label, the product identifier of the product indicated by the product label, which is specifically:
  • the first position information that meets the first screening condition is determined, and the product identifier of the product corresponding to the first position information is used as the product identifier of the product indicated by the product label.
  • the first screening condition is: the corresponding area is closest to the area corresponding to the position information of the product tag.
  • the type of the shelf analysis result to be determined includes: whether the promotion label is accurate;
  • the target object categories include: promotion tag categories and product tag categories.
  • the server identifying each target object belonging to the target object category in the target shelf image, and obtaining attribute information of each target object is specifically:
  • Identify each promotional label and each product label in the target shelf image obtain position information and promotional information of each promotional label, and position information of each product label and a product identifier of the product indicated by each product label.
  • the server determining a shelf analysis result corresponding to the shelf to be analyzed based on the attribute information of each target object is specifically:
  • the second screening condition is: the corresponding area is closest to the area corresponding to the position information of the promotion tag;
  • For each promotion tag determine whether the promotion information of the promotion tag matches the promotion information specified by the target. If they match, determine that the promotion label is accurate; otherwise, determine that the promotion label is inaccurate; wherein the promotion information specified by the target is: The specified promotion information associated with the product identification corresponding to the promotion label.
  • the type of the shelf analysis result to be determined includes: whether the product is out of stock;
  • the target object categories include: a product category and a product label category.
  • the identifying each target object belonging to the target object category in the target shelf image, and obtaining attribute information of each target object is specifically:
  • the server determining a shelf analysis result corresponding to the shelf to be analyzed based on the attribute information of each target object is specifically:
  • the third position information For each product label, from the position information of each product, determine the third position information that meets the third screening condition, calculate the sum of the area of the area corresponding to each third position information, and calculate the sum of the area and the product label
  • the ratio of the area of the corresponding designated storage area is used as the storage ratio of the product corresponding to the product label, and it is determined whether the storage ratio is less than the preset storage ratio. If so, it is determined that the product corresponding to the product label is out of stock, otherwise, It is determined that the product corresponding to the product label is not out of stock; wherein the third screening condition is that the corresponding area is located in a designated storage area of the product indicated by the product label.
  • the server calculates the designated storage area of the goods indicated by each goods label based on the location information of each goods label, which is specifically:
  • the bottom left vertex and the bottom right vertex of the designated storage area of the product indicated by the item label are calculated. Apex and the predetermined height value of each layer to determine the designated storage area of the goods indicated by the goods label;
  • the reference object corresponding to any product label is: an adjacent product label located at the same horizontal position as the product label, or an area on the shelf edge of the shelf at the same horizontal position as the product label.
  • the type of the shelf analysis result to be determined includes: heat information of the goods;
  • the target object categories include: goods categories and goods label categories;
  • the obtaining the target shelf image is specifically:
  • the server identifying each target object belonging to the target object category in the target shelf image, and obtaining attribute information of each target object is specifically:
  • For each target shelf image identify each product and each product label in the target shelf image, and obtain position information of each product and position information of each product label.
  • the server determining a shelf analysis result corresponding to the shelf to be analyzed based on the attribute information of each target object is specifically:
  • each target shelf image For each target shelf image, based on the position information of each item in the target shelf image and the position information of each item label, determining a storage ratio of the item corresponding to each item label;
  • the sum of the differences corresponding to the product label is used as the product heat of the product corresponding to the product label.
  • the type of the shelf analysis result to be determined includes: personnel's heat information;
  • the target object category includes: a personnel category
  • the server obtaining the target shelf image is specifically:
  • the server identifying each target object belonging to the target object category in the target shelf image, and obtaining attribute information of each target object is specifically:
  • For each target shelf image identify each person in the target shelf image, and obtain the number of persons in the target shelf image.
  • the server determining a shelf analysis result corresponding to the shelf to be analyzed based on the attribute information of each target object is specifically:
  • the ratio of the sum of the number of persons obtained to the number of images of multiple target shelf images is calculated as the person's popularity information.
  • an embodiment of the present application further provides an electronic device.
  • the electronic device includes a processor 801, a communication interface 802, a memory 803, and a communication bus 804.
  • the processor 801, the communication interface 802, and the memory 803 Complete communication with each other through the communication bus 804,
  • the processor 801 is configured to implement the shelf analysis method according to the first aspect when executing a program stored in the memory 803.
  • the technical solution provided in the embodiment of the present application obtains a target shelf image; wherein the target shelf image includes an image area corresponding to the shelf to be analyzed; identifying each target object belonging to the target object category in the target shelf image and obtaining attribute information of each target object
  • the target object category is: an object category corresponding to the type of shelf analysis result to be determined; and based on attribute information of each target object, a shelf analysis result corresponding to the shelf is determined.
  • the communication bus mentioned in the above electronic device may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc.
  • PCI Peripheral Component Interconnect
  • EISA Extended Industry Standard Architecture
  • the communication bus can be divided into an address bus, a data bus, a control bus, and the like.
  • the figure only uses a thick line to represent, but it does not mean that there is only one bus or one type of bus.
  • the communication interface is used for communication between the aforementioned electronic device and other devices.
  • the memory may include Random Access Memory (RAM), and may also include Non-Volatile Memory (NVM), such as at least one disk memory.
  • RAM Random Access Memory
  • NVM Non-Volatile Memory
  • the memory may also be at least one storage device located far from the foregoing processor.
  • the aforementioned processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc .; it may also be a digital signal processor (Digital Signal Processing, DSP), special integration Circuit (Application Specific Integrated Circuit, ASIC), Field-Programmable Gate Array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.
  • CPU central processing unit
  • NP network processor
  • DSP Digital Signal Processing
  • ASIC Application Specific Integrated Circuit
  • FPGA Field-Programmable Gate Array
  • an embodiment of the present application provides a computer-readable storage medium.
  • the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the shelf analysis according to the first aspect is implemented. method.
  • an embodiment of the present application provides an executable program code, where the executable program code is used to be executed to execute the shelf analysis method according to the first aspect.
  • the technical solution provided in the embodiment of the present application obtains a target shelf image; wherein the target shelf image includes an image area corresponding to the shelf to be analyzed; identifying each target object belonging to the target object category in the target shelf image and obtaining attribute information of each target object
  • the target object category is: an object category corresponding to the type of shelf analysis result to be determined; and based on attribute information of each target object, a shelf analysis result corresponding to the shelf to be analyzed is determined.

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Accounting & Taxation (AREA)
  • Finance (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Development Economics (AREA)
  • Data Mining & Analysis (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

本申请实施例提供了一种货架分析方法、装置、系统及电子设备，所述方法包括：获得目标货架图像；其中，所述目标货架图像中包含待分析货架对应的图像区域；识别所述目标货架图像中属于目标对象类别的各个目标对象，得到所述各个目标对象的属性信息，其中，所述目标对象类别为：与货架分析结果的类型相对应的对象类别；基于所述各个目标对象的属性信息，确定所述待分析货架对应的货架分析结果。可见，本申请实施例所提供的货架分析方法可以自动化地对货架进行分析，而无需通过工作人员人工对货架进行分析，从而提高了货架分析的效率。

Description

一种货架分析方法、装置、系统及电子设备
本申请要求于2018年9月7日提交中国专利局、申请号为201811045439.0发明名称为“一种货架分析方法、装置、系统及电子设备”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
本申请涉及图像处理技术领域，特别是涉及一种货架分析方法、装置、系统及电子设备。
背景技术
货架在零售店中被广泛采用，用来展示及陈列货品。例如在大型超市中，所有货品都被置于货架上，以方便顾客选购货品。
为了确保货品正常有序地销售，需要对货架进行分析。例如，需要分析货架上的各类货品的摆放区域是否准确，货架上的各类货品是否缺货，以及货架上粘贴的各类货品标签的粘贴区域是否准确等。
现有技术中，通常通过人工来对货架进行分析。例如，对于货架上摆放的每一类货品，工作人员人工检测该类货品在货架上的实际摆放区域与预先确定的该类货品的摆放区域是否相同。很显然，工作人员人工分析货架的效率较低。那么，如何快速有效地对货架进行分析，是一个亟待解决的问题。
发明内容
本申请实施例的目的在于提供一种货架分析方法、装置、系统及电子设备，以实现快速有效地对货架进行分析。具体技术方案如下：
第一方面，本申请实施例提供了一种货架分析方法，所述方法包括：
获得目标货架图像；其中，所述目标货架图像中包含待分析货架对应的图像区域；
识别所述目标货架图像中属于目标对象类别的各个目标对象，得到所述各个目标对象的属性信息，其中，所述目标对象类别为：与待确定的货架分析结果的类型相对应的对象类别；
基于所述各个目标对象的属性信息，确定所述待分析货架对应的货架分析结果。
第二方面，本申请实施例提供了一种货架分析装置，所述装置包括：
图像获取模块，用于获得目标货架图像；其中，所述目标货架图像中包含待分析货架对应的图像区域；
对象识别模块，用于识别所述目标货架图像中属于目标对象类别的各个目标对象，得到所述各个目标对象的属性信息，其中，所述目标对象类别为：与待确定的货架分析结果的类型相对应的对象类别；
货架分析模块，用于基于所述各个目标对象的属性信息，确定所述待分析货架对应的货架分析结果。
第三方面，本申请实施例提供了一种货架分析系统，所述系统包括：
图像采集设备和服务器；
其中，所述图像采集设备，用于采集目标货架图像，并将所采集的目标货架图像发送至服务器；其中，所述目标货架图像中包含待分析货架对应的图像区域；
服务器，用于从图像采集设备获得目标货架图像；识别所述目标货架图像中属于目标对象类别的各个目标对象，得到所述各个目标对象的属性信息，其中，所述目标对象类别为：与待确定的货架分析结果的类型相对应的对象类别；基于所述各个目标对象的属性信息，确定所述货架对应的货架分析结果。
第四方面，本申请实施例提供了一种电子设备，包括处理器、通信接口、存储器和通信总线，其中，处理器，通信接口，存储器通过通信总线完成相互间的通信；
存储器，用于存放计算机程序；
处理器，用于执行存储器上所存放的程序时，实现第一方面所述的货架分析方法。
第五方面，本申请实施例提供了一种计算机可读存储介质，所述计算机可读存储介质内存储有计算机程序，所述计算机程序被处理器执行时实现第一方面所述的货架分析方法。
第六方面，本申请实施例提供了一种可执行程序代码，所述可执行程序代码用于被运行以执行第一方面所述的货架分析方法。
本申请实施例提供的技术方案，获得目标货架图像；其中，目标货架图像中包含待分析货架对应的图像区域；识别目标货架图像中属于目标对象类别的各个目标对象，得到各个目标对象的属性信息，其中，目标对象类别为：与待确定的货架分析结果的类型相对应的对象类别；基于各个目标对象的属性信息，确定待分析货架对应的货架分析结果。可见，本申请实施例所提供的货架分析方法可以自动化地对货架进行分析，而无需通过工作人员人工对货架进行分析，从而提高了货架分析的效率。
附图说明
为了更清楚地说明本申请实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1为本申请实施例所提供的一种货架分析方法的流程图；
图2为本申请实施例所提供的一种目标货架图像的示意图；
图3为本申请实施例所提供的一种采集目标货架图像的图像采集设备的安装方式示意图；
图4为本申请实施例所提供的一种目标棚格图的示意图；
图5为本申请实施例所提供的一种包含促销标签的目标货架图像的示意图；
图6为本申请实施例所提供的一种货架分析装置的结构示意图；
图7为本申请实施例所提供的一种货架分析系统的示意图；
图8为本申请实施例所提供的一种电子设备的示意图。
具体实施方式
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
为了解决现有技术存在的工作人员分析货架的效率较低的技术问题，本申请实施例提供了一种货架分析方法、装置、系统及电子设备。
第一方面，下面首先对本申请实施例所提供的一种货架分析方法进行介绍。
需要说明的是，本申请实施例所提供的一种货架分析方法的执行主体可以为一种货架分析装置，该货架分析装置可以运行于一种货架分析系统中用于数据处理的设备。
在具体应用中，该货架分析系统可以包括：服务器，以及用于采集待分析货架的目标货架图像的图像采集设备，此时，该货架分析装置可以运行于该服务器，以基于图像采集设备所拍摄的目标货架图像来得到货架分析结果；当然，该货架分析系统也可以仅仅包括：图像采集设备，此时，该货架分析装置可以运行于图像采集设备，那么，该图像采集设备在拍摄待分析货架的目标货架图像后，可以基于该目标货架图像来得到货架分析结果。其中，该图像采集设备可以为摄像头形式的相机等。
如图1所示，本申请实施例所提供的一种货架分析方法，可以包括如下步骤：
S110，获得目标货架图像；其中，目标货架图像中包含待分析货架对应的图像区域；
在对货架进行分析时，可以获取图像采集设备拍摄的目标货架图像，该目标货架图像中包括待分析货架对应的图像区域。需要说明的是，图像采集设备拍摄的任一张货架图像均可以是本申请实施例所述的目标货架图像。举例而言：目标货架图像可以是如图2所示的货架图像，由图2可知，货架图像中可以包括：陈列在货架上的货品，粘贴在货架上的货品标签等，当然，图2只是以举例的方式示意性地展示了货架图像，本申请对货架图像中所包括的内容不做具体限定。
其中，图像采集设备的安装方式可以有多种，在具体应用中，如图3所示，图像采集设备的安装方式可以为：吊装或内嵌于货架，当然并不局限于此。以吊装方式安装的图像采集设备可以称为吊装图像采集设备，以内嵌于货架方式安装的相机可以称为内嵌于货架的图像采集设备。
在图像采集设备的安装方式为吊装时，图像采集设备可以悬吊于待分析货架所在房间的屋顶上，图像采集设备可以由上而下地采集目标货架图像。以吊装的方式安装图像采集设备有诸多优点，例如，方便为图像采集设备供 电；对图像采集设备体积和外观的约束较小；图像采集设备拍摄货架图像时，受到遮挡以及干扰的概率较小。
在图像采集设备的安装方式为内嵌于货架时，图像采集设备内嵌于某货架某层的底板之中。此时，图像采集设备也可以拍摄目标货架图像。以内嵌于货架的方式安装图像采集设备也有诸多优点，例如，图像采集设备所拍摄的货架图像的透视畸变小；而且，由于图像采集设备内嵌于货架中，因此图像采集设备不影响货架所在房间的外观。
可以理解的是，本申请实施例对图像采集设备的安装方式不做具体限定；且图像采集设备的数量可以根据实际情况来确定，本申请实施例对图像采集设备的数量不做具体限定。而且，图像采集设备可以实时地拍摄目标货架图像，也可以按照预设的采样间隔拍摄目标货架图像，本申请实施对此也不做具体限定。
S120，识别目标货架图像中属于目标对象类别的各个目标对象，得到各个目标对象的属性信息，其中，目标对象类别为：与待确定的货架分析结果的类型相对应的对象类别。
在获得了目标货架图像之后，可以利用预先训练好的算法模型识别目标货架图像中属于目标对象类别的各个目标对象，并得到各个目标对象的属性信息。其中，目标对象类别可以为：货品标签类别，促销标签类别，货品类别，人员类别等；相应的，目标对象可以为货品标签，促销标签，货品，人员等。并且，各个目标对象的属性可以为：目标对象的位置信息，目标对象的区域大小等。本申请实施例对目标对象类别、目标对象及目标对象的属性信息不做具体限定。
需要说明的是，目标对象类别为与待确定的货架分析结果的类型相对应的对象类别，也就是说，货品分析结果的类型决定所需识别的对象类别，在货架分析结果的类型确定的情况下，待识别的对象类别是确定的。另外，对于不同对象类别的目标对象，在货架分析时，所需的属性信息可以不同或相同。
可选地，在一种实现方式中，在货品分析结果的类型为货品陈列是否准确，即货品在货架上的陈列区域是否准确时，目标对象类别可以为：货品标签类别，此时，目标对象类别对应的各个目标对象为货品标签，各个目标对 象的属性信息可以为：货品标签的位置信息。
在货品分析结果的类型为促销标签是否准确，即促销标签在货架上的区域是否准确时，目标对象类别可以为：促销标签类别和货品标签类别，此时，目标对象类别对应的各个目标对象为：促销标签和货品标签，各个目标对象的属性信息可以为：各个促销标签的位置信息和促销信息，以及各个货品标签的位置信息和各个货品标签所指示货品的货品标识。
在货品分析结果的类型为货品是否缺货时，目标对象类别可以为：货品类别和货品标签类别，此时，目标对象类别对应的各个目标对象为：货品和货品标签，各个目标对象的属性信息可以为：各个货品的位置信息和各个货品标签的位置信息。
在货品分析结果的类型为货品的热度信息时，目标对象类别可以为：货品类别和货品标签类别，此时，目标对象类别对应的各个目标对象为：货品和货品标签，各个目标对象的属性信息可以为：各个货品的位置信息和各个货品标签的位置信息。
在货品分析结果的类型为人员的热度信息时，目标对象类别可以为：人员类别，此时，目标对象类别对应的各个目标对象为：人员，各个目标对象的属性信息可以为：人员数量。
为了方案完整及描述清楚，后续结合具体实施例，对识别所述目标货架图像中属于目标对象类别的各个目标对象，得到所述各个目标对象的属性信息进行详细描述。需要强调的是，上述所给出的关于货架分析结果的类型与对象类别的映射关系，仅仅作为示例，并不应该构成对本申请实施例的限定。
S130，基于各个目标对象的属性信息，确定待分析货架对应的货架分析结果。
由于目标对象类别是与待确定的货架分析结果的类型相对应的对象类别，因此，在得到各个目标对象的属性信息后，可以基于各个目标对象的属性信息，确定待分析货架对应的货架分析结果。可以理解的是，对于不同类型的货架分析结果，基于各个目标对象的属性信息，确定待分析货架对应的货架分析结果的具体过程不同。为了方案完整及描述清楚，下面结合具体实施例，对基于各个目标对象的属性信息，确定待分析货架对应的货架分析结果的具体实现方式进行详细描述。
本申请实施例提供的技术方案，获得目标货架图像；其中，目标货架图像中包含待分析货架对应的图像区域；识别目标货架图像中属于目标对象类别的各个目标对象，得到各个目标对象的属性信息，其中，目标对象类别为：与待确定的货架分析结果的类型相对应的对象类别；基于各个目标对象的属性信息，确定待分析货架对应的货架分析结果。可见，本申请实施例所提供的货架分析方法可以自动化地对货架进行分析，而无需通过工作人员人工对货架进行分析，从而提高了货架分析的效率。
需要说明的是，待确定的货架分析结果的类型可以是固定的，此时，目标对象类别可以是固定的。当然，在每次货架分析时，可以由人工指定当前待确定的货架分析结果的类型，或者，该货架分析装置根据预先设定的规则，确定当前待确定的货架分析结果的类型，这样，可以提高货架分析的可控性，从而满足不同时刻下的货架分析需求。其中，该预先设定的规则可以为：根据时间点/时间段与货架分析结果的类型的对应关系，当然并不局限于此。那么，为了提高货架分析的可控性，在一种实施方式中，在识别目标货架图像中属于目标对象类别的各个目标对象，得到各个目标对象的属性信息的步骤之前，所述货架分析方法还可以包括：
获得待确定的货架分析结果的类型；
基于预设的关于货架分析结果的类型与对象类别的映射关系，确定与待确定的货架分析结果的类型相对应的目标对象类别。
在该实施方式中，在对目标货架图像的对象进行识别之前，可以获得待确定的货架分析结果的类型；并且，基于预设的关于货架分析结果的类型与对象类别的映射关系，确定与待确定的货架分析结果的类型相对应的目标对象类别。具体的，关于货架分析结果的类型与对象类别的映射关系，可以为：货品陈列是否准确对应于：货品标签类别；促销标签是否准确对应于：促销标签类别和货品标签类别；货品是否缺货对应于：货品类别和货品标签类别；货品的热度信息对应于：货品类别和货品标签类别；人员的热度信息对应于：人员类别。当然，这只是示例性地对关于货架分析结果的类型与对象类别的映射关系进行了描述，并不应该构成对本申请实施例的限定。
举例而言，当所获得的货品分析结果的类型为货品陈列是否准确时，基于预设的关于货架分析结果的类型与对象类别的映射关系，可以确定目标对 象类别为：货品标签类别，此时，可以只识别目标货架图像中的货品标签。
为了方案清楚，下面结合待确定的货架分析结果的各个类型，分别介绍识别目标货架图像中属于目标对象类别的各个目标对象，得到各个目标对象的属性信息的具体过程，以及基于各个目标对象的属性信息，确定待分析货架对应的货架分析结果的具体过程。
在一种实施方式中，待确定的货架分析结果的类型包括：货品陈列是否准确，相应的，目标对象类别包括：货品标签类别。
相应的，S120，识别目标货架图像中属于目标对象类别的各个目标对象，得到各个目标对象的属性信息的步骤，可以包括：
识别目标货架图像中各个货品标签，得到各个货品标签的位置信息。上述货品标签可以为：用于展示商品名称、价格等信息的纸质显示牌或电子显示牌；
其中，可以利用预先训练的、用于识别货品标签的位置信息的神经网络模型，来识别目标货架图像中各个货品标签，得到各个货品标签的位置信息。关于神经网络模型的类型、结构以及训练过程，在此不做限定。
相应的，S130，基于各个目标对象的属性信息，确定待分析货架对应的货架分析结果的步骤，可以包括如下步骤a1-a4：
a1，基于各个货品标签的位置信息，确定各个货品标签在待分析货架中的层号和列号；
其中，基于各个货品标签的位置信息，确定各个货品标签在待分析货架中的层号和列号的步骤的具体实现方式存在多种，任一种能够基于各个对象的位置信息来确定各个对象的层号和列号的实现方式均可以适用于本申请实施例。可选地，在一种示例中，基于各个货品标签的位置信息，确定各个货品标签在待分析货架中的层号和列号的步骤，可以包括：
将各个货品标签位置信息按照纵坐标信息由小到大或由大到小进行垂直投影，得到各个货品标签在待分析货架中的层号；
将每一层号对应的各个货品标签的位置信息按照横坐标信息由小到大或由大到小进行水平投影，得到各个货品标签在待分析货架中的列号。
a2，针对每一货品标签，确定该货品标签所指示货品的货品标识。
其中，关于所述针对每一货品标签，确定该货品标签所指示货品的货品标识的步骤的具体实现方式存在多种。可选地，针对每一货品标签，确定该货品标签所指示货品的货品标识的步骤，可以包括：
针对每一货品标签，识别该货品标签的标签内容，得到该货品标签所指示货品的货品标识。其中，所谓的识别该货品标签的标签内容具体可以为：对该货品标签所在的图像区域中的文字或者条形码进行识别，得到该货品标签所指示货品的货品标识。
可选地，针对每一货品标签，确定该货品标签所指示货品的货品标识的步骤，可以包括：
识别目标货架图像中各个货品，得到各个货品的位置信息和货品标识；
针对每一货品标签，从各个货品的位置信息中，确定符合第一筛选条件的第一位置信息，将对应第一位置信息的货品的货品标识，作为该货品标签所指示货品的货品标识，其中，第一筛选条件为：所对应的区域与该货品标签的位置信息所对应的区域最近。
在该实施方式中，第一位置信息所对应的区域与该货品标签的位置信息所对应的区域最近，说明第一位置信息所对应的货品是该货品标签所指示的货品，因此，第一位置信息所对应的货品的货品标识与该货品标签所指示货品的货品标识相同，可以将对应第一位置信息的货品的货品标识，作为该货品标签所指示货品的货品标识。
a3，基于各个货品标签在待分析货架中的层号和列号，以及各个货品标签所指示货品的货品标识，按照预设的生成方式，生成待分析货架的目标棚格图；
其中，预设的生成方式为：每一货品标签对应目标棚格图中的一个棚格，任一货品标签所对应棚格的标识为该货品标签所指示货品的货品标识，且任一货品标识所对应棚格在棚格图中的层号和列号，与该货品标签在待分析货架中的层号和列号相同。
在得到各个货品标签在待分析货架中的层号和列号以及各个货品标签所指示的货品标识后，即可以生成目标货架图像的目标棚格图。任一货架的棚格图指货架布置的拓扑图，用于标明每一层、每一列的商品类别和位置关系。
举例而言，目标货架图像为图2所示的货架图像，该货架图像最上面一层(层号为1)所陈列的商品有三类，从左到右，这三类商品的商品标识依次为A、B、C；该货架图像中间一层(层号为2)所陈列的商品有两类，从左到右，这两类商品的商品标识依次为D、E；该货架图像最下面一层(层号为3)所陈列的商品有三类，从左到右，这三类商品的商品标识依次为F、G、H。
由图2可知，目标货架图像中包含8个货品标签。分别识别8个货品标签的位置信息和8个货品标签所指示货品的货品标识，分别为A、B、C、D、E、F、G、H。
将8个货品标签的位置信息按照纵坐标由小到大(由上到下)进行垂直投影，得到各个货品标签在待分析货架中的层号，可以理解的是，8个货品标签中，其中，有3个货品标签的层号为1，有2个货品标签的层号为2，其余3个货品标签的层号为3。
在得到各个货品标签在待分析货架中的层号后，将层号为1的3个货品标签按照横坐标信息由小到大(由左到右)进行水平投影，得到层号为1的3个货品标签的列号分别为1、2、3；将层号为2的2个货品标签按照横坐标信息由小到大进行水平投影，得到层号为2的2个货品标签的列号分别为1、2；将层号为3的3个货品标签按照横坐标信息由小到大进行水平投影，得到层号为3的3个货品标签的列号分别为1、2、3。
在得到各个货品标签在待分析货架中的层号和列号之后，以及各个货品标签所指示的货品标识后，即可以绘制出目标货架图像对应的目标棚格图，所绘制出的目标棚格图可以为如图4所示的棚格图。
a4，基于目标棚格图和预设的标准棚格图的对比结果，确定待分析货架的货品陈列是否准确。
在绘制出目标棚格图之后，可以将目标棚格图和预设的标准棚格图进行比对，得到对比结果，并基于所得到的对比结果，确定待分析货架的货品陈列是否准确。
可选地，基于目标棚格图和预设的标准棚格图的对比结果，确定待分析货架的货品陈列是否准确的步骤，可以包括：
针对目标棚格图中的每一棚格，判断该棚格的标识与该棚格所对应的标准棚格的标识是否一致，如果是，确定待分析货架的货品陈列准确，否则，确定待分析货架的货品陈列不准确，该棚格所对应的标准棚格为：标准棚格图中与该棚格的层号和列号相同的棚格。
举例而言，目标棚格图中的一个棚格的标识为A，该棚格所对应的标准棚格的标识为B，说明待分析货架的货品陈列不准确。
并且，由于目标棚格图是基于待分析货架上的货品标签来生成的，因此，基于目标棚格图和预设的标准棚格图的对比结果，还可以判断出待分析货架上的货品标签是否丢失。例如，如果目标棚格图和预设的标准棚格图的对比结果为：目标棚格图的层数与标准棚格图的层数不同；或者目标棚格图的列数与标准棚格图的列数不同，则可以判断出待分析货架上的货品标签存在丢失的情况。
在确定出待分析货架的货品陈列不准确或者待分析货架上的货品标签丢失后，为了使得工作人员及时得知货品陈列不准确或者待分析货架上的货品标签丢失，可以向与工作人员关联的终端发送报警信息。其中，与工作人员关联的终端可以是手机、电脑等，报警信息的内容可以是简单的报警声音，还可以是携带有待分析货架的货品陈列不准确或者待分析货架上的货品标签丢失的信息；报警信息的形式可以是播放语音，发短信，发邮件等，本申请实施例对报警信息的内容和形式不做具体限定。
在另一种实施方式中，待确定的货架分析结果的类型包括：促销标签是否准确；目标对象类别包括：促销标签类别和货品标签类别。
相应的，S120，识别目标货架图像中属于目标对象类别的各个目标对象，得到各个目标对象的属性信息的步骤，可以包括：
识别目标货架图像中各个促销标签和各个货品标签，得到各个促销标签的位置信息和促销信息，以及各个货品标签的位置信息和各个货品标签所指示货品的货品标识。其中，可以利用预先训练的、用于识别货品标签的位置信息的神经网络模型，来识别目标货架图像中各个促销标签和各个货品标签，得到各个促销标签的位置信息和促销信息，以及各个货品标签的位置信息和各个货品标签所指示货品的货品标识。关于神经网络模型的类型、结构以及训练过程，在此不做限定。
相应的，S130，基于各个目标对象的属性信息，确定待分析货架对应的货架分析结果的步骤，可以包括如下两个步骤，分别为b1和b2：
b1，针对每一促销标签，从各个货品标签的位置信息中，确定符合第二筛选条件的第二位置信息，将对应第二位置信息的货品标签的货品标识，作为该促销标签所对应的货品标识；其中，第二筛选条件为：所对应的区域与该促销标签的位置信息所对应的区域最近；
b2，针对每一促销标签，判断该促销标签的促销信息，是否与目标指定促销信息匹配，如果匹配，确定该促销标签准确，否则，确定该促销标签不准确；其中，目标指定促销信息为：该促销标签对应的货品标识所关联的指定促销信息。
在该实施方式中，针对每一促销标签，如果一个货品标签的位置信息所对应的区域与该促销标签的位置信息所对应的区域最近，说明该促销标签与该货品标签针对的是同一货品，因此，可以将该货品标签的货品标识作为该促销标签所对应的货品标识。其中，任一种能够计算两个位置信息所对应区域间的距离的实现方式，均可以适用于本申请实施例。
举例而言，如图5所示，目标货架图像的最上层左边有一个促销标签，该促销标签的促销信息为20％，假设所确定的该促销标签所对应的货品标识为A，但与货品标识为A所关联的指定促销信息为30％，可见，20％与30％不同，即该促销标签的促销信息与该促销标签对应的货品标识所关联的制定促销信息不匹配，因此，可以确定该促销标签不准确。
可以理解的是，通过本实施例提供的技术方案，还可以发现促销标签是否漏标。例如，如果与某一货品标识关联的指定促销信息为50％，而在目标货架图像中未识别出来货品标识对应的促销信息，此时，可以判断出促销标签漏标。
同样的，在确定出促销标签不准确或促销标签漏标后，为了使得工作人员及时得知促销标签不准确或促销标签漏标，可以向与工作人员关联的终端发送报警信息，在此不再赘述。
在另一种实施方式中，待确定的货架分析结果的类型包括：货品是否缺货；
目标对象类别包括：货品类别和货品标签类别。
相应的，S120，识别目标货架图像中属于目标对象类别的各个目标对象，得到各个目标对象的属性信息的步骤，可以包括：
识别目标货架图像中各个货品和各个货品标签，得到各个货品的位置信息和各个货品标签的位置信息。
其中，可以利用预先训练的、用于识别货品标签的位置信息的神经网络模型，来识别目标货架图像中各个货品和各个货品标签，得到各个货品的位置信息和各个货品标签的位置信息。关于神经网络模型的类型、结构以及训练过程，在此不做限定。
相应的，S130，基于各个目标对象的属性信息，确定待分析货架对应的货架分析结果的步骤，可以包括如下步骤c1和c2：
c1，基于各个货品标签的位置信息，计算各个货品标签所指示货品的指定存放区域；
可选地，在一种示例性的实现方式中，基于各个货品标签的位置信息，计算各个货品标签所指示货品的指定存放区域的步骤，可以包括：
针对每一货品标签，基于该货品标签的位置信息和参考对象的位置信息，计算该货品标签所指示货品的指定存放区域的左下方的顶点和右下方的顶点，利用左下方的顶点、右下方的顶点以及预定的每一层的高度值，确定该货品标签所指示货品的指定存放区域；
其中，任一货品标签对应的参考对象为：与该货品标签位于同一水平位置、且相邻的货品标签，或者，待分析货架的货架边缘上与该货品标签在同一水平位置的区域。
在该实现方式中，当任一货品标签对应的参考对象为：与该货品标签位于同一水平位置、且相邻的货品标签时，可以通过该货品标签的位置信息与其相邻的货品标签的位置信息，来确定该货品标签所指示货品的指定存放区域的左下方的顶点和右下方的顶点。
例如，任一货品标签位于该货品标签所指示货品的指定存放区域的左下角，可以将该货品标签的左上方的顶点作为指定存放区域的左下方的顶点，将与该货品标签相邻的货品标签的左上方的顶点作为指定存放区域的右下方 的顶点。
同样的，在任一货品标签对应的参考对象为：待分析货架的货架边缘上与该货品标签在同一水平位置的区域时，也可以通过该货品标签的位置信息与参考对象的位置信息，来确定该货品标签所指示货品的指定存放区域的左下方的顶点和右下方的顶点。
需要说明的是，货品标签的位置信息不同，所确定的货品标签所指示货品的指定存放区域的左下方的顶点和右下方的顶点也不同。本申请实施例对此不做具体限定。
c2，针对每一货品标签，从各个货品的位置信息中，确定符合第三筛选条件的第三位置信息，计算各个第三位置信息所对应区域的面积之和，计算面积之和与该货品标签所对应指定存放区域的面积的比值，作为该货品标签所对应货品的存储比值，并判断存储比值是否小于预设存储比值，如果是，确定该货品标签所对应的货品缺货，否则，确定该货品标签所对应的货品未缺货；其中，第三筛选条件为：所对应区域位于该货品标签所指示货品的指定存放区域。
对于每一货品标签，在确定了该货品标签所指示货品的指定存放区域后，可以计算在该指定存放区域中，货品所占的面积之和，并计算货品所占的面积之和与该指定存放区域的面积的比值，将该比值作为该货品标签所对应货品的存储比值；并判断该存储比值是否小于预设存储比值，如果该存储比值小于预设存储比值，则说明该货品标签所指示货品的指定存放区域中的商品较少，此时，可以确定该货品标签所对应的货品缺货；如果该存储比值不小于预设存储比值，则说明该货品标签所指示货品的指定存放区域中的商品较多，此时，可以确定该货品标签所对应的货品不缺货。
需要说明的是，预设存储比值的大小可以按照实际情况进行设置，如可以是0％，10％等，本申请实施例对预设存储比值的大小不做具体限定。
同样的，在确定出货品缺货后，为了使得工作人员及时得知货品缺货，可以向与工作人员关联的终端发送报警信息，在此不再赘述。
在另一种实施方式中，待确定的货架分析结果的类型包括：货品的热度信息；
目标对象类别包括：货品类别和货品标签类别；
此时，获得目标货架图像的步骤，可以包括：
获得预定时间段内的多张目标货架图像；
相应的，S120，识别目标货架图像中属于目标对象类别的各个目标对象，得到各个目标对象的属性信息的步骤，可以包括：
针对每一目标货架图像，识别该目标货架图像中各个货品和各个货品标签，得到各个货品的位置信息和各个货品标签的位置信息。
在该实施例中，为了得知货品的热度信息，需要获得预定时间段内的多张目标货架图像。该预设时间段可以是分钟、小时、天等，本申请实施例对预设时间段不做具体限定。
相应的，S130，基于各个目标对象的属性信息，确定待分析货架对应的货架分析结果的步骤，可以包括如下步骤d1-d3：
d1，针对每一张目标货架图像，基于该目标货架图像中各个货品的位置信息和各个货品标签的位置信息，确定每一货品标签所对应货品的存储比值；
其中，在上述步骤c2中，已经对如何基于该目标货架图像中各个货品的位置信息和各个货品标签的位置信息，确定每一货品标签所对应货品的存储比值进行了详细介绍，在此不再赘述。
d2，针对相邻的两张目标货架图像，计算同一货品标签所对应货品的存储比值的差值；
d3，针对每一货品标签，将该货品标签对应的各个差值之和，作为该货品标签对应的货品的货品热度。
举例而言，获得一小时内采集的3张目标货架图像，这三张相邻的目标货架图像分别为目标货架图像1、目标货架图像2和目标货架图像3，其中，目标货架图像1中货品标签1所对应货品的存储比值为90％；目标货架图像2中货品标签1所对应货品的存储比值为70％；目标货架图像3中货品标签1所对应货品的的存储比值为60％；目标货架图像1中货品标签1所对应货品的存储比值与目标货架图像2中货品标签1所对应货品的存储比值之差为20％；目标货架图像2中货品标签1所对应货品的存储比值与目标货架图像3中货品标签1所对应货品的存储比值之差为10％；将计算所得到的两个存储 比值的差值求和，得到该货品标签1对应的货品的货品热度为30％。
在另一种实施方式中，待确定的货架分析结果的类型包括：人员的热度信息；
目标对象类别包括：人员类别；
获得目标货架图像的步骤，包括：
获得预定时间段内的多张目标货架图像；
S120，识别目标货架图像中属于目标对象类别的各个目标对象，得到各个目标对象的属性信息的步骤，包括：
针对每一目标货架图像，识别该目标货架图像中各个人员，得到该目标货架图像中所包含人员的人员数量。
在该实施例中，为了得知货品的热度信息，需要获得预定时间段内的多张目标货架图像。该预设时间段可以是分钟、小时、天等，本申请实施例对预设时间段不做具体限定。
在一种实施方式中，S130，基于各个目标对象的属性信息，确定待分析货架对应的货架分析结果的步骤，可以包括如下两个步骤，分别为e1和e2：
e1，计算各个目标货架图像中所包含人员的人员数量之和；
e2，将计算所得到的人员数量之和与多张目标货架图像的图像数量的比值，作为人员的热度信息。举例而言，获得一小时内采集的3张目标货架图像，分别为目标货架图像1、目标货架图像2和目标货架图像3，其中，目标货架图像1中所包含的人员数量为6个；目标货架图像2中所包含的人员数量为9个；目标货架图像3中所包含的人员数量为3个；这三张目标货架图像中所包含人员的人员数量之和为18；计算所得到的人员数量之和与目标货架图像的图像数量的比值为6，那么，人员的热度信息为6。
可以理解的是，在得到货品的热度信息以及人员的热度信息后，还可以计算货品的热度信息与人员的热度信息的比值，即可以得到货品的转化率。
需要强调的是，本申请实施例中待确定的货架分析结果的类型可以只包括一个，还可以同时包括多种。也就是说，可以同时分析货品陈列是否准确、促销标签是否准确、货品是否缺货、货品的热度信息、人员的热度信息中的一种或者多种，本申请实施例对此不作具体限定。并且，可以将货架分析结 果以报表的形式输出，这都是合理的。
第二方面，本申请实施例提供了一种货架分析装置，如图6所示，所述装置包括：
图像获取模块610，用于获得目标货架图像；其中，所述目标货架图像中包含待分析货架对应的图像区域；
对象识别模块620，用于识别所述目标货架图像中属于目标对象类别的各个目标对象，得到所述各个目标对象的属性信息，其中，所述目标对象类别为：与待确定的货架分析结果的类型相对应的对象类别；
货架分析模块630，用于基于所述各个目标对象的属性信息，确定所述待分析货架对应的货架分析结果。
本申请实施例提供的技术方案，获得目标货架图像；其中，目标货架图像中包含待分析货架对应的图像区域；识别目标货架图像中属于目标对象类别的各个目标对象，得到各个目标对象的属性信息，其中，目标对象类别为：与待确定的货架分析结果的类型相对应的对象类别；基于各个目标对象的属性信息，确定待分析货架对应的货架分析结果。可见，本申请实施例所提供的货架分析方法可以自动化地对货架进行分析，而无需通过工作人员人工对货架进行分析，从而提高了货架分析的效率。
可选的，所述装置还可以包括：
类型确定模块，用于在所述对象识别模块识别所述目标货架图像中属于目标对象类别的各个目标对象，得到所述各个目标对象的属性信息之前，获得待确定的货架分析结果的类型；
基于预设的关于货架分析结果的类型与对象类别的映射关系，确定与所述待确定的货架分析结果的类型相对应的目标对象类别。
可选的，所述待确定的货架分析结果的类型包括：货品陈列是否准确；
所述目标对象类别包括：货品标签类别；
所述对象识别模块，具体用于：
识别所述目标货架图像中各个货品标签，得到所述各个货品标签的位置信息。
可选的，所述货架分析模块，包括：
层号列号确定子模块，用于基于各个货品标签的位置信息，确定各个货品标签在待分析货架中的层号和列号；
第一货品标识确定子模块，用于针对每一货品标签，确定该货品标签所指示货品的货品标识；
棚格图生成子模块，用于基于各个货品标签在待分析货架中的层号和列号，以及各个货品标签所指示货品的货品标识，按照预设的生成方式，生成所述目标货架图像的目标棚格图，其中，所述预设的生成方式为：每一货品标签对应所述目标棚格图中的一个棚格，任一货品标签所对应棚格的标识为该货品标签所指示货品的货品标识，且任一货品标识所对应棚格在棚格图中的层号和列号，与该货品标签在待分析货架中的层号和列号相同；
货品陈列确定子模块，用于基于所述目标棚格图和预设的标准棚格图的对比结果，确定所述待分析货架的货品陈列是否准确。
可选的，所述货品陈列确定子模块，具体用于：
针对所述目标棚格图中的每一棚格，判断该棚格的标识与该棚格所对应的标准棚格的标识是否一致，如果是，确定所述待分析货架的货品陈列准确，否则，确定所述待分析货架的货品陈列不准确，该棚格所对应的标准棚格为：所述标准棚格图中与该棚格的层号和列号相同的棚格。
可选的，所述层号列号确定子模块，具体用于：
将各个货品标签位置信息按照纵坐标信息由小到大或由大到小进行垂直投影，得到各个货品标签在待分析货架中的层号；
将每一层号对应的各个货品标签的位置信息按照横坐标信息由小到大或由大到小进行水平投影，得到各个货品标签在待分析货架中的列号。
可选的，所述货品标识确定子模块，具体用于：
针对每一货品标签，识别该货品标签的标签内容，得到该货品标签所指示货品的货品标识。
可选的，所述货品标识确定子模块，具体用于：
识别所述目标货架图像中各个货品，得到所述各个货品的位置信息和货品标识；
针对每一货品标签，从各个货品的位置信息中，确定符合第一筛选条件的第一位置信息，将对应所述第一位置信息的货品的货品标识，作为该货品标签所指示货品的货品标识，其中，所述第一筛选条件为：所对应的区域与该货品标签的位置信息所对应的区域最近。
可选的，所述待确定的货架分析结果的类型包括：促销标签是否准确；
所述目标对象类别包括：促销标签类别和货品标签类别；
所述对象识别模块，具体用于：
识别所述目标货架图像中各个促销标签和各个货品标签，得到所述各个促销标签的位置信息和促销信息，以及各个货品标签的位置信息和各个货品标签所指示货品的货品标识。
可选的，所述货架分析模块，包括：
第二货品标识确定子模块，用于针对每一促销标签，从各个货品标签的位置信息中，确定符合第二筛选条件的第二位置信息，将对应第二位置信息的货品标签的货品标识，作为该促销标签所对应的货品标识；其中，所述第二筛选条件为：所对应的区域与该促销标签的位置信息所对应的区域最近；
促销标签确定子模块，用于针对每一促销标签，判断该促销标签的促销信息，是否与目标指定促销信息匹配，如果匹配，确定该促销标签准确，否则，确定该促销标签不准确；其中，所述目标指定促销信息为：该促销标签对应的货品标识所关联的指定促销信息。
可选的，所述待确定的货架分析结果的类型包括：货品是否缺货；
所述目标对象类别包括：货品类别和货品标签类别；
所述对象识别模块，具体用于：
识别所述目标货架图像中各个货品和各个货品标签，得到所述各个货品的位置信息和所述各个货品标签的位置信息。
可选的，所述货架分析模块，包括：
存放区域确定子模块，用于基于各个货品标签的位置信息，计算各个货品标签所指示货品的指定存放区域；
货品缺货确定子模块，用于针对每一货品标签，从各个货品的位置信息中，确定符合第三筛选条件的第三位置信息，计算各个第三位置信息所对应 区域的面积之和，计算所述面积之和与该货品标签所对应指定存放区域的面积的比值，作为该货品标签所对应货品的存储比值，并判断所述存储比值是否小于预设存储比值，如果是，确定该货品标签所对应的货品缺货，否则，确定该货品标签所对应的货品未缺货；其中，所述第三筛选条件为：所对应区域位于该货品标签所指示货品的指定存放区域。
可选的，所述存放区域确定子模块，具体用于：
针对每一货品标签，基于该货品标签的位置信息和参考对象的位置信息，计算该货品标签所指示货品的指定存放区域的左下方的顶点和右下方的顶点，利用左下方的顶点、右下方的顶点以及预定的每一层的高度值，确定该货品标签所指示货品的指定存放区域；
其中，任一货品标签对应的参考对象为：与该货品标签位于同一水平位置、且相邻的货品标签，或者，所述货架的货架边缘上与该货品标签在同一水平位置的区域。
可选的，所述待确定的货架分析结果的类型包括：货品的热度信息；
所述目标对象类别包括：货品类别和货品标签类别；
所述图像获取模块，具体用于：
获得预定时间段内的多张目标货架图像；
所述对象识别模块，具体用于：
针对每一目标货架图像，识别该目标货架图像中各个货品和各个货品标签，得到所述各个货品的位置信息和所述各个货品标签的位置信息。
可选的，所述货架分析模块，包括：
存储比值确定子模块，用于针对每一张目标货架图像，基于该目标货架图像中各个货品的位置信息和所述各个货品标签的位置信息，确定每一货品标签所对应货品的存储比值；
存储比值差值确定子模块，用于针对相邻的两张目标货架图像，计算同一货品标签所对应货品的存储比值的差值；
货品热度确定子模块，用于针对每一货品标签，将该货品标签对应的各个差值之和，作为该货品标签对应的货品的货品热度。
可选的，所述待确定的货架分析结果的类型包括：人员的热度信息；
所述目标对象类别包括：人员类别；
所述图像获取模块，具体用于：
获得预定时间段内的多张目标货架图像；
所述对象识别模块，具体用于：
针对每一目标货架图像，识别该目标货架图像中各个人员，得到该目标货架图像中所包含人员的人员数量。
可选的，所述货架分析模块，包括：
人员数量计算子模块，用于计算各个目标货架图像中所包含人员的人员数量之和；
人员热度确定子模块，用于将计算所得到的人员数量之和与多张目标货架图像的图像数量的比值，作为人员的热度信息。
第三方面，本申请实施例提供了一种货架分析系统，如图7所示，所述系统包括：
图像采集设备710和服务器720；
其中，所述图像采集设备710，用于采集目标货架图像，并将所采集的目标货架图像发送至服务器；其中，所述目标货架图像中包含待分析货架对应的图像区域；
服务器720，用于从图像采集设备获得目标货架图像；识别所述目标货架图像中属于目标对象类别的各个目标对象，得到所述各个目标对象的属性信息，其中，所述目标对象类别为：与待确定的货架分析结果的类型相对应的对象类别；基于所述各个目标对象的属性信息，确定所述待分析货架对应的货架分析结果。
本发明实施例提供的技术方案，获得目标货架图像；其中，目标货架图像中包含待分析货架对应的图像区域；识别目标货架图像中属于目标对象类别的各个目标对象，得到各个目标对象的属性信息，其中，目标对象类别为：与待确定的货架分析结果的类型相对应的对象类别；基于各个目标对象的属性信息，确定待分析货架对应的货架分析结果。可见，本发明实施例所提供的货架分析方法可以自动化地对货架进行分析，而无需通过工作人员人工对 货架进行分析，从而提高了货架分析的效率。
可选的，所述服务器，还用于：
在所述对象识别模块识别所述目标货架图像中属于目标对象类别的各个目标对象，得到所述各个目标对象的属性信息之前，获得待确定的货架分析结果的类型；
基于预设的关于货架分析结果的类型与对象类别的映射关系，确定与所述待确定的货架分析结果的类型相对应的目标对象类别。
可选的，在第一种实施方式中，所述待确定的货架分析结果的类型包括：货品陈列是否准确；
所述目标对象类别包括：货品标签类别；
所述服务器识别所述目标货架图像中属于目标对象类别的各个目标对象，得到所述各个目标对象的属性信息，具体为：
识别所述目标货架图像中各个货品标签，得到所述各个货品标签的位置信息。
可选的，所述基于所述各个目标对象的属性信息，确定所述待分析货架对应的货架分析结果，具体为：
基于各个货品标签的位置信息，确定各个货品标签在待分析货架中的层号和列号；
针对每一货品标签，确定该货品标签所指示货品的货品标识；
基于各个货品标签在待分析货架中的层号和列号，以及各个货品标签所指示货品的货品标识，按照预设的生成方式，生成所述目标货架图像的目标棚格图，其中，所述预设的生成方式为：每一货品标签对应所述目标棚格图中的一个棚格，任一货品标签所对应棚格的标识为该货品标签所指示货品的货品标识，且任一货品标识所对应棚格在棚格图中的层号和列号，与该货品标签在待分析货架中的层号和列号相同；
基于所述目标棚格图和预设的标准棚格图的对比结果，确定所述待分析货架的货品陈列是否准确。
可选的，所述服务器基于所述目标棚格图和预设的标准棚格图的对比结果，确定所述待分析货架的货品陈列是否准确，具体为：
针对所述目标棚格图中的每一棚格，判断该棚格的标识与该棚格所对应的标准棚格的标识是否一致，如果是，确定所述待分析货架的货品陈列准确，否则，确定所述待分析货架的货品陈列不准确，该棚格所对应的标准棚格为：所述标准棚格图中与该棚格的层号和列号相同的棚格。
可选的，所述服务器基于各个货品标签的位置信息，确定各个货品标签在待分析货架中的层号和列号，具体为：
将各个货品标签位置信息按照纵坐标信息由小到大或由大到小进行垂直投影，得到各个货品标签在待分析货架中的层号；
将每一层号对应的各个货品标签的位置信息按照横坐标信息由小到大或由大到小进行水平投影，得到各个货品标签在待分析货架中的列号。
可选的，所述服务器针对每一货品标签，确定该货品标签所指示货品的货品标识，具体为：
针对每一货品标签，识别该货品标签的标签内容，得到该货品标签所指示货品的货品标识。
可选的，所述服务器针对每一货品标签，确定该货品标签所指示货品的货品标识，具体为：
识别所述目标货架图像中各个货品，得到所述各个货品的位置信息和货品标识；
针对每一货品标签，从各个货品的位置信息中，确定符合第一筛选条件的第一位置信息，将对应所述第一位置信息的货品的货品标识，作为该货品标签所指示货品的货品标识，其中，所述第一筛选条件为：所对应的区域与该货品标签的位置信息所对应的区域最近。
可选的，所述待确定的货架分析结果的类型包括：促销标签是否准确；
所述目标对象类别包括：促销标签类别和货品标签类别。
所述服务器识别所述目标货架图像中属于目标对象类别的各个目标对象，得到所述各个目标对象的属性信息，具体为：
识别所述目标货架图像中各个促销标签和各个货品标签，得到所述各个促销标签的位置信息和促销信息，以及各个货品标签的位置信息和各个货品标签所指示货品的货品标识。
可选的，所述服务器基于所述各个目标对象的属性信息，确定所述待分析货架对应的货架分析结果，具体为：
针对每一促销标签，从各个货品标签的位置信息中，确定符合第二筛选条件的第二位置信息，将对应第二位置信息的货品标签的货品标识，作为该促销标签所对应的货品标识；其中，所述第二筛选条件为：所对应的区域与该促销标签的位置信息所对应的区域最近；
针对每一促销标签，判断该促销标签的促销信息，是否与目标指定促销信息匹配，如果匹配，确定该促销标签准确，否则，确定该促销标签不准确；其中，所述目标指定促销信息为：该促销标签对应的货品标识所关联的指定促销信息。
可选的，所述待确定的货架分析结果的类型包括：货品是否缺货；
所述目标对象类别包括：货品类别和货品标签类别。
所述识别所述目标货架图像中属于目标对象类别的各个目标对象，得到所述各个目标对象的属性信息，具体为：
识别所述目标货架图像中各个货品和各个货品标签，得到所述各个货品的位置信息和所述各个货品标签的位置信息。
可选的，所述服务器基于所述各个目标对象的属性信息，确定所述待分析货架对应的货架分析结果，具体为：
基于各个货品标签的位置信息，计算各个货品标签所指示货品的指定存放区域；
针对每一货品标签，从各个货品的位置信息中，确定符合第三筛选条件的第三位置信息，计算各个第三位置信息所对应区域的面积之和，计算所述面积之和与该货品标签所对应指定存放区域的面积的比值，作为该货品标签所对应货品的存储比值，并判断所述存储比值是否小于预设存储比值，如果是，确定该货品标签所对应的货品缺货，否则，确定该货品标签所对应的货品未缺货；其中，所述第三筛选条件为：所对应区域位于该货品标签所指示货品的指定存放区域。
可选的，所述服务器基于各个货品标签的位置信息，计算各个货品标签所指示货品的指定存放区域，具体为：
针对每一货品标签，基于该货品标签的位置信息和参考对象的位置信息，计算该货品标签所指示货品的指定存放区域的左下方的顶点和右下方的顶点，利用左下方的顶点、右下方的顶点以及预定的每一层的高度值，确定该货品标签所指示货品的指定存放区域；
其中，任一货品标签对应的参考对象为：与该货品标签位于同一水平位置、且相邻的货品标签，或者，所述货架的货架边缘上与该货品标签在同一水平位置的区域。
可选的，所述待确定的货架分析结果的类型包括：货品的热度信息；
所述目标对象类别包括：货品类别和货品标签类别；
所述获得目标货架图像，具体为：
获得预定时间段内的多张目标货架图像；
所述服务器识别所述目标货架图像中属于目标对象类别的各个目标对象，得到所述各个目标对象的属性信息，具体为：
针对每一目标货架图像，识别该目标货架图像中各个货品和各个货品标签，得到所述各个货品的位置信息和所述各个货品标签的位置信息。
可选的，所述服务器基于所述各个目标对象的属性信息，确定所述待分析货架对应的货架分析结果，具体为：
针对每一张目标货架图像，基于该目标货架图像中各个货品的位置信息和所述各个货品标签的位置信息，确定每一货品标签所对应货品的存储比值；
针对相邻的两张目标货架图像，计算同一货品标签所对应货品的存储比值的差值；
针对每一货品标签，将该货品标签对应的各个差值之和，作为该货品标签对应的货品的货品热度。
可选的，所述待确定的货架分析结果的类型包括：人员的热度信息；
所述目标对象类别包括：人员类别；
所述服务器获得目标货架图像，具体为：
获得预定时间段内的多张目标货架图像；
所述服务器识别所述目标货架图像中属于目标对象类别的各个目标对象， 得到所述各个目标对象的属性信息，具体为：
针对每一目标货架图像，识别该目标货架图像中各个人员，得到该目标货架图像中所包含人员的人员数量。
可选的，所述服务器基于所述各个目标对象的属性信息，确定所述待分析货架对应的货架分析结果，具体为：
计算各个目标货架图像中所包含人员的人员数量之和；
将计算所得到的人员数量之和与多张目标货架图像的图像数量的比值，作为人员的热度信息。
第四方面，本申请实施例还提供了一种电子设备，如图8所示，包括处理器801、通信接口802、存储器803和通信总线804，其中，处理器801，通信接口802，存储器803通过通信总线804完成相互间的通信，
存储器803，用于存放计算机程序；
处理器801，用于执行存储器803上所存放的程序时，实现第一方面所述的货架分析方法。
本申请实施例提供的技术方案，获得目标货架图像；其中，目标货架图像中包含待分析货架对应的图像区域；识别目标货架图像中属于目标对象类别的各个目标对象，得到各个目标对象的属性信息，其中，目标对象类别为：与待确定的货架分析结果的类型相对应的对象类别；基于各个目标对象的属性信息，确定货架对应的货架分析结果。可见，本申请实施例所提供的货架分析方法可以自动化地对货架进行分析，而无需通过工作人员人工对货架进行分析，从而提高了货架分析的效率。
上述电子设备提到的通信总线可以是外设部件互连标准(Peripheral Component Interconnect，PCI)总线或扩展工业标准结构(Extended Industry Standard Architecture，EISA)总线等。该通信总线可以分为地址总线、数据总线、控制总线等。为便于表示，图中仅用一条粗线表示，但并不表示仅有一根总线或一种类型的总线。
通信接口用于上述电子设备与其他设备之间的通信。
存储器可以包括随机存取存储器(Random Access Memory，RAM)，也可 以包括非易失性存储器(Non-Volatile Memory，NVM)，例如至少一个磁盘存储器。可选的，存储器还可以是至少一个位于远离前述处理器的存储装置。
上述的处理器可以是通用处理器，包括中央处理器(Central Processing Unit，CPU)、网络处理器(Network Processor，NP)等；还可以是数字信号处理器(Digital Signal Processing，DSP)、专用集成电路(Application Specific Integrated Circuit，ASIC)、现场可编程门阵列(Field-Programmable Gate Array，FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件。
第五方面，本申请实施例提供了一种计算机可读存储介质，所述计算机可读存储介质内存储有计算机程序，所述计算机程序被处理器执行时实现实现第一方面所述的货架分析方法。
第六方面，本申请实施例提供了一种可执行程序代码，所述可执行程序代码用于被运行以执行第一方面所述的货架分析方法。
本申请实施例提供的技术方案，获得目标货架图像；其中，目标货架图像中包含待分析货架对应的图像区域；识别目标货架图像中属于目标对象类别的各个目标对象，得到各个目标对象的属性信息，其中，目标对象类别为：与待确定的货架分析结果的类型相对应的对象类别；基于各个目标对象的属性信息，确定待分析货架对应的货架分析结果。可见，本申请实施例所提供的货架分析方法可以自动化地对货架进行分析，而无需通过工作人员人工对货架进行分析，从而提高了货架分析的效率。
需要说明的是，在本文中，诸如第一和第二等之类的关系术语仅仅用来将一个实体或者操作与另一个实体或操作区分开来，而不一定要求或者暗示这些实体或操作之间存在任何这种实际的关系或者顺序。而且，术语“包括”、“包含”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、方法、物品或者设备不仅包括那些要素，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、方法、物品或者设备所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括所述要素的过程、方法、物品或者设备中还存在另外的相同要素。
本说明书中的各个实施例均采用相关的方式描述，各个实施例之间相同 相似的部分互相参见即可，每个实施例重点说明的都是与其他实施例的不同之处。尤其，对于装置实施例、系统实施例、电子设备实施例、存储介质实施例及可执行程序代码实施例而言，由于其基本相似于方法实施例，所以描述的比较简单，相关之处参见方法实施例的部分说明即可。
以上所述仅为本申请的较佳实施例而已，并非用于限定本申请的保护范围。凡在本申请的精神和原则之内所作的任何修改、等同替换、改进等，均包含在本申请的保护范围内。

Claims (22)

  1. 一种货架分析方法，其特征在于，所述方法包括：
    获得目标货架图像；其中，所述目标货架图像中包含待分析货架对应的图像区域；
    识别所述目标货架图像中属于目标对象类别的各个目标对象，得到所述各个目标对象的属性信息，其中，所述目标对象类别为：与待确定的货架分析结果的类型相对应的对象类别；
    基于所述各个目标对象的属性信息，确定所述待分析货架对应的货架分析结果。
  2. 根据权利要求1所述的方法，其特征在于，在所述识别所述目标货架图像中属于目标对象类别的各个目标对象，得到所述各个目标对象的属性信息的步骤之前，所述方法还包括：
    获得待确定的货架分析结果的类型；
    基于预设的关于货架分析结果的类型与对象类别的映射关系，确定与所述待确定的货架分析结果的类型相对应的目标对象类别。
  3. 根据权利要求1或2所述的方法，其特征在于，所述待确定的货架分析结果的类型包括：货品陈列是否准确；
    所述目标对象类别包括：货品标签类别；
    所述识别所述目标货架图像中属于目标对象类别的各个目标对象，得到所述各个目标对象的属性信息的步骤，包括：
    识别所述目标货架图像中各个货品标签，得到所述各个货品标签的位置信息。
  4. 根据权利要求3所述的方法，其特征在于，所述基于所述各个目标对象的属性信息，确定所述待分析货架对应的货架分析结果的步骤，包括：
    基于各个货品标签的位置信息，确定各个货品标签在所述待分析货架中的层号和列号；
    针对每一货品标签，确定该货品标签所指示货品的货品标识；
    基于各个货品标签在所述待分析货架中的层号和列号，以及各个货品标签所指示货品的货品标识，按照预设的生成方式，生成所述目标货架图像的目标棚格图，其中，所述预设的生成方式为：每一货品标签对应所述目标棚格图中的一个棚格，任一货品标签所对应棚格的标识为该货品标签所指示货品的货品标识，且任一货品标识所对应棚格在棚格图中的层号和列号，与该货品标签在所述待分析货架中的层号和列号相同；
    基于所述目标棚格图和预设的标准棚格图的对比结果，确定所述待分析货架的货品陈列是否准确。
  5. 根据权利要求4所述的方法，其特征在于，所述基于所述目标棚格图和预设的标准棚格图的对比结果，确定所述待分析货架的货品陈列是否准确的步骤，包括：
    针对所述目标棚格图中的每一棚格，判断该棚格的标识与该棚格所对应的标准棚格的标识是否一致，如果是，确定所述待分析货架的货品陈列准确，否则，确定所述待分析货架的货品陈列不准确，该棚格所对应的标准棚格为：所述标准棚格图中与该棚格的层号和列号相同的棚格。
  6. 根据权利要求4所述的方法，其特征在于，所述基于各个货品标签的位置信息，确定各个货品标签在所述待分析货架中的层号和列号的步骤，包括：
    将各个货品标签位置信息按照纵坐标信息由小到大或由大到小进行垂直投影，得到各个货品标签在所述待分析货架中的层号；
    将每一层号对应的各个货品标签的位置信息按照横坐标信息由小到大或由大到小进行水平投影，得到各个货品标签在所述待分析货架中的列号。
  7. 根据权利要求4所述的方法，其特征在于，针对每一货品标签，确定该货品标签所指示货品的货品标识的步骤，包括：
    针对每一货品标签，识别该货品标签的标签内容，得到该货品标签所指示货品的货品标识。
  8. 根据权利要求4所述的方法，其特征在于，所述针对每一货品标签，确定该货品标签所指示货品的货品标识的步骤，包括：
    识别所述目标货架图像中各个货品，得到所述各个货品的位置信息和货品标识；
    针对每一货品标签，从各个货品的位置信息中，确定符合第一筛选条件的第一位置信息，将对应所述第一位置信息的货品的货品标识，作为该货品标签所指示货品的货品标识，其中，所述第一筛选条件为：所对应的区域与该货品标签的位置信息所对应的区域最近。
  9. 根据权利要求1或2所述的方法，其特征在于，所述待确定的货架分析结果的类型包括：促销标签是否准确；
    所述目标对象类别包括：促销标签类别和货品标签类别；
    所述识别所述目标货架图像中属于目标对象类别的各个目标对象，得到所述各个目标对象的属性信息的步骤，包括：
    识别所述目标货架图像中各个促销标签和各个货品标签，得到所述各个促销标签的位置信息和促销信息，以及各个货品标签的位置信息和各个货品标签所指示货品的货品标识。
  10. 根据权利要求9所述的方法，其特征在于，基于所述各个目标对象的属性信息，确定所述待分析货架对应的货架分析结果，包括：
    针对每一促销标签，从各个货品标签的位置信息中，确定符合第二筛选条件的第二位置信息，将对应第二位置信息的货品标签的货品标识，作为该促销标签所对应的货品标识；其中，所述第二筛选条件为：所对应的区域与该促销标签的位置信息所对应的区域最近；
    针对每一促销标签，判断该促销标签的促销信息，是否与目标指定促销信息匹配，如果匹配，确定该促销标签准确，否则，确定该促销标签不准确；其中，所述目标指定促销信息为：该促销标签对应的货品标识所关联的指定促销信息。
  11. 根据权利要求1或2所述的方法，其特征在于，所述待确定的货架分 析结果的类型包括：货品是否缺货；
    所述目标对象类别包括：货品类别和货品标签类别；
    所述识别所述目标货架图像中属于目标对象类别的各个目标对象，得到所述各个目标对象的属性信息的步骤，包括：
    识别所述目标货架图像中各个货品和各个货品标签，得到所述各个货品的位置信息和所述各个货品标签的位置信息。
  12. 根据权利要求11所述的方法，其特征在于，所述基于所述各个目标对象的属性信息，确定所述待分析货架对应的货架分析结果的步骤，包括：
    基于各个货品标签的位置信息，计算各个货品标签所指示货品的指定存放区域；
    针对每一货品标签，从各个货品的位置信息中，确定符合第三筛选条件的第三位置信息，计算各个第三位置信息所对应区域的面积之和，计算所述面积之和与该货品标签所对应指定存放区域的面积的比值，作为该货品标签所对应货品的存储比值，并判断所述存储比值是否小于预设存储比值，如果是，确定该货品标签所对应的货品缺货，否则，确定该货品标签所对应的货品未缺货；其中，所述第三筛选条件为：所对应区域位于该货品标签所指示货品的指定存放区域。
  13. 根据权利要求12所述的方法，其特征在于，所述基于各个货品标签的位置信息，计算各个货品标签所指示货品的指定存放区域的步骤，包括：
    针对每一货品标签，基于该货品标签的位置信息和参考对象的位置信息，计算该货品标签所指示货品的指定存放区域的左下方的顶点和右下方的顶点，利用左下方的顶点、右下方的顶点以及预定的每一层的高度值，确定该货品标签所指示货品的指定存放区域；
    其中，任一货品标签对应的参考对象为：与该货品标签位于同一水平位置、且相邻的货品标签，或者，所述货架的货架边缘上与该货品标签在同一水平位置的区域。
  14. 根据权利要求1或2所述的方法，其特征在于，所述待确定的货架分 析结果的类型包括：货品的热度信息；
    所述目标对象类别包括：货品类别和货品标签类别；
    所述获得目标货架图像的步骤，包括：
    获得预定时间段内的多张目标货架图像；
    所述识别所述目标货架图像中属于目标对象类别的各个目标对象，得到所述各个目标对象的属性信息的步骤，包括：
    针对每一目标货架图像，识别该目标货架图像中各个货品和各个货品标签，得到所述各个货品的位置信息和所述各个货品标签的位置信息。
  15. 根据权利要求14所述的方法，其特征在于，所述基于所述各个目标对象的属性信息，确定所述待分析货架对应的货架分析结果的步骤，包括：
    针对每一张目标货架图像，基于该目标货架图像中各个货品的位置信息和所述各个货品标签的位置信息，确定每一货品标签所对应货品的存储比值；
    针对相邻的两张目标货架图像，计算同一货品标签所对应货品的存储比值的差值；
    针对每一货品标签，将该货品标签对应的各个差值之和，作为该货品标签对应的货品的货品热度。
  16. 根据权利要求1或2所述的方法，其特征在于，所述待确定的货架分析结果的类型包括：人员的热度信息；
    所述目标对象类别包括：人员类别；
    所述获得目标货架图像的步骤，包括：
    获得预定时间段内的多张目标货架图像；
    所述识别所述目标货架图像中属于目标对象类别的各个目标对象，得到所述各个目标对象的属性信息的步骤，包括：
    针对每一目标货架图像，识别该目标货架图像中各个人员，得到该目标货架图像中所包含人员的人员数量。
  17. 根据权利要求16所述的方法，其特征在于，所述基于所述各个目标对象的属性信息，确定所述待分析货架对应的货架分析结果的步骤，包括：
    计算各个目标货架图像中所包含人员的人员数量之和；
    将计算所得到的人员数量之和与多张目标货架图像的图像数量的比值，作为人员的热度信息。
  18. 一种货架分析装置，其特征在于，所述装置包括：
    图像获取模块，用于获得目标货架图像；其中，所述目标货架图像中包含待分析货架对应的图像区域；
    对象识别模块，用于识别所述目标货架图像中属于目标对象类别的各个目标对象，得到所述各个目标对象的属性信息，其中，所述目标对象类别为：与待确定的货架分析结果的类型相对应的对象类别；
    货架分析模块，用于基于所述各个目标对象的属性信息，确定所述待分析货架对应的货架分析结果。
  19. 一种货架分析系统，其特征在于，所述系统包括：
    图像采集设备和服务器；
    其中，所述图像采集设备，用于采集目标货架图像，并将所采集的目标货架图像发送至服务器；其中，所述目标货架图像中包含待分析货架对应的图像区域；
    服务器，用于从图像采集设备获得目标货架图像；识别所述目标货架图像中属于目标对象类别的各个目标对象，得到所述各个目标对象的属性信息，其中，所述目标对象类别为：与待确定的货架分析结果的类型相对应的对象类别；基于所述各个目标对象的属性信息，确定所述待分析货架对应的货架分析结果。
  20. 一种电子设备，其特征在于，包括处理器、通信接口、存储器和通信总线，其中，处理器，通信接口，存储器通过通信总线完成相互间的通信；
    存储器，用于存放计算机程序；
    处理器，用于执行存储器上所存放的程序时，实现权利要求1-17任一所述的方法步骤。
  21. 一种计算机可读存储介质，其特征在于，所述计算机可读存储介质内存储有计算机程序，所述计算机程序被处理器执行时实现权利要求1-17任一所述的方法步骤。
  22. 一种可执行程序代码，其特征在于，所述可执行程序代码用于被运行以执行权利要求1-17任一所述的方法步骤。
PCT/CN2019/103291 2018-09-07 2019-08-29 一种货架分析方法、装置、系统及电子设备 WO2020048376A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201811045439.0A CN110889419B (zh) 2018-09-07 2018-09-07 一种货架分析方法、装置、系统及电子设备
CN201811045439.0 2018-09-07

Publications (1)

Publication Number Publication Date
WO2020048376A1 true WO2020048376A1 (zh) 2020-03-12

Family

ID=69722749

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/103291 WO2020048376A1 (zh) 2018-09-07 2019-08-29 一种货架分析方法、装置、系统及电子设备

Country Status (2)

Country Link
CN (1) CN110889419B (zh)
WO (1) WO2020048376A1 (zh)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111604922A (zh) * 2020-06-11 2020-09-01 合肥立木科技有限责任公司 一种基于机器视觉目标识别技术的超市上架率巡检机器人
CN113298597A (zh) * 2020-08-06 2021-08-24 阿里巴巴集团控股有限公司 对象热度分析系统、方法及装置
CN113344478A (zh) * 2021-08-06 2021-09-03 捷尔杰（天津）设备有限公司 离散生产线连续配料方法、装置、电子设备及存储介质
CN113643083A (zh) * 2020-04-27 2021-11-12 京东方科技集团股份有限公司 一种商品标记及展示方法、服务器、终端及商品展示系统

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111507253B (zh) * 2020-04-16 2023-06-30 腾讯科技（深圳）有限公司 基于人工智能的陈列物品审核方法、装置
CN113627418B (zh) * 2020-05-07 2023-08-25 杭州海康威视数字技术股份有限公司 数据管理方法、计算机设备及计算机可读存储介质
CN113627415A (zh) * 2021-08-26 2021-11-09 浙江大华技术股份有限公司 确定目标对象摆放信息的方法及装置
CN115205665B (zh) * 2022-09-15 2022-12-02 创新奇智(合肥)科技有限公司 一种识别产品位置的方法、装置、电子设备及介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060032915A1 (en) * 2004-08-12 2006-02-16 International Business Machines Retail store method and system
CN102184405A (zh) * 2011-04-19 2011-09-14 清华大学 图像采集分析方法
CN102930264A (zh) * 2012-09-29 2013-02-13 李炳华 基于图像识别技术的商品陈列信息采集分析系统及方法
CN107292248A (zh) * 2017-06-05 2017-10-24 广州诚予国际市场信息研究有限公司 一种基于图像识别技术的商品管理方法及系统

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8189855B2 (en) * 2007-08-31 2012-05-29 Accenture Global Services Limited Planogram extraction based on image processing
CN102789601A (zh) * 2012-04-12 2012-11-21 北京京东世纪贸易有限公司 货场结构控制方法和装置
US10002344B2 (en) * 2016-10-17 2018-06-19 Conduent Business Services, Llc System and method for retail store promotional price tag detection
CN107045641B (zh) * 2017-04-26 2020-07-28 广州图匠数据科技有限公司 一种基于图像识别技术的货架识别方法
CN108491825A (zh) * 2018-03-30 2018-09-04 百度在线网络技术（北京）有限公司 信息生成方法和装置

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060032915A1 (en) * 2004-08-12 2006-02-16 International Business Machines Retail store method and system
CN102184405A (zh) * 2011-04-19 2011-09-14 清华大学 图像采集分析方法
CN102930264A (zh) * 2012-09-29 2013-02-13 李炳华 基于图像识别技术的商品陈列信息采集分析系统及方法
CN107292248A (zh) * 2017-06-05 2017-10-24 广州诚予国际市场信息研究有限公司 一种基于图像识别技术的商品管理方法及系统

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113643083A (zh) * 2020-04-27 2021-11-12 京东方科技集团股份有限公司 一种商品标记及展示方法、服务器、终端及商品展示系统
CN111604922A (zh) * 2020-06-11 2020-09-01 合肥立木科技有限责任公司 一种基于机器视觉目标识别技术的超市上架率巡检机器人
CN113298597A (zh) * 2020-08-06 2021-08-24 阿里巴巴集团控股有限公司 对象热度分析系统、方法及装置
CN113344478A (zh) * 2021-08-06 2021-09-03 捷尔杰（天津）设备有限公司 离散生产线连续配料方法、装置、电子设备及存储介质
CN113344478B (zh) * 2021-08-06 2022-01-04 捷尔杰（天津）设备有限公司 离散生产线连续配料方法、装置、电子设备及存储介质

Also Published As

Publication number Publication date
CN110889419B (zh) 2023-04-07
CN110889419A (zh) 2020-03-17

Similar Documents

Publication Publication Date Title
WO2020048376A1 (zh) 一种货架分析方法、装置、系统及电子设备
CN108416403B (zh) 商品与标签的自动关联方法、系统、设备及存储介质
US8941645B2 (en) Comparing virtual and real images in a shopping experience
CN113205037B (zh) 事件检测的方法、装置、电子设备以及可读存储介质
EP3944221A1 (en) Price tag information display method and device, and shelf system
CN111104879B (zh) 房屋功能间的识别方法和装置、可读存储介质、电子设备
CN111832447A (zh) 建筑图纸构件识别方法、电子设备及相关产品
CN108875531B (zh) 人脸检测方法、装置、系统及计算机存储介质
CN112508109B (zh) 一种图像识别模型的训练方法及装置
CN110516628A (zh) 货架空缺位置商品信息获取方法、系统、设备及存储介质
CN112200631B (zh) 一种行业分类模型的训练方法及装置
CN112153320B (zh) 一种物品尺寸的测量方法、装置、电子设备和存储介质
CN108389230A (zh) 冰箱容量自动检测方法、系统、设备及存储介质
WO2021179138A1 (zh) 商超货架上商品的分析方法和系统
CN112183460A (zh) 一种智能识别环境卫生的方法及装置
CN111428743B (zh) 商品识别方法、商品处理方法、装置及电子设备
CN111753614A (zh) 一种商品货架的监控方法和装置
CN110765825A (zh) 物品摆放状态获取方法和系统
CN113674049A (zh) 基于图片搜索的商品在架位置识别方法、系统和存储介质
CN111047731A (zh) 一种基于ar技术的电信机房巡检方法及系统
CN113743382B (zh) 一种货架陈列检测方法、装置及系统
CN115223003A (zh) 基于YOLOv5的入侵检测模型的训练方法和装置
CN112950329A (zh) 商品动态信息生成方法、装置、设备及计算机可读介质
CN112989957A (zh) 适用于设备柜的安全监测方法及系统
CN110738260A (zh) 一种零售门店太空箱摆放检测方法、装置和设备

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: 19858419

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: 19858419

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 19858419

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 15/10/2021)

122 Ep: pct application non-entry in european phase

Ref document number: 19858419

Country of ref document: EP

Kind code of ref document: A1