WO2021164222A1 - 确定物品分析数据 - Google Patents

确定物品分析数据 Download PDF

Info

Publication number
WO2021164222A1
WO2021164222A1 PCT/CN2020/110036 CN2020110036W WO2021164222A1 WO 2021164222 A1 WO2021164222 A1 WO 2021164222A1 CN 2020110036 W CN2020110036 W CN 2020110036W WO 2021164222 A1 WO2021164222 A1 WO 2021164222A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
remaining
item
data
items
Prior art date
Application number
PCT/CN2020/110036
Other languages
English (en)
French (fr)
Inventor
王建磊
Original Assignee
北京三快在线科技有限公司
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 北京三快在线科技有限公司 filed Critical 北京三快在线科技有限公司
Publication of WO2021164222A1 publication Critical patent/WO2021164222A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0004—Industrial image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10004—Still image; Photographic image

Definitions

  • the embodiments of the present application relate to the field of computer technology, and particularly relate to determining article analysis data.
  • Analyzing the popularity of items can provide data references for improving the supply of items and optimizing items. For example, analyzing the popularity of dishes can provide a basis for catering businesses to improve their dishes or optimize their supply categories to improve user satisfaction.
  • the popularity of the dish is usually analyzed through data such as the sales of the dish, user's evaluation, and scoring, to obtain an index value indicating the popularity of the dish.
  • the accuracy of the evaluation index value of the popularity of the dish determined by the above method in the prior art cannot accurately reflect the popularity of the dish in some cases. For example, some users do not evaluate and score the dishes after the meal, which will cause the evaluation index value of the popularity of the dishes determined according to the evaluation or scoring of the dishes to be inaccurate due to sparse data.
  • an embodiment of the present application provides a method for determining item analysis data, including:
  • the item analysis data indicating the popularity of the item is determined.
  • an article analysis data determination device including:
  • the remaining item image acquisition module is used to acquire the image containing the remaining item
  • An item integrity data determination module configured to determine the remaining items contained in the image and the integrity data of the remaining items by performing image processing on the image
  • the item analysis data determination module is used for determining item analysis data indicating the popularity of the item by performing data analysis processing on the integrity data of each of the remaining items and the collected information of the image containing each of the remaining items.
  • an embodiment of the present application also discloses an electronic device, including a memory, a processor, and a computer program stored on the memory and capable of running on the processor.
  • the processor executes the computer program when the computer program is executed.
  • the method for determining article analysis data described in the embodiments of the present application is described in the embodiments of the present application.
  • an embodiment of the present application provides a computer-readable storage medium on which a computer program is stored.
  • the program is executed by a processor, the steps of the method for determining the article analysis data disclosed in the embodiment of the present application are provided.
  • the method for determining item analysis data disclosed in the embodiments of the present application obtains an image containing remaining items; performs image processing on the image to determine the remaining items contained in the image and the integrity data of the remaining items; The integrity data of each of the remaining items and the collected information of the images containing each of the remaining items are processed for data analysis, and the item analysis data indicating the popularity of the item is determined, which helps to improve the determination of the item analysis data. Accuracy.
  • FIG. 1 is a flowchart of a method for determining article analysis data according to Embodiment 1 of the present application
  • FIG. 3 is an example of shelf images collected in Embodiment 1 of the present application.
  • Fig. 4 is a flowchart of the method for determining article analysis data in the second embodiment of the present application.
  • Fig. 5 is a schematic diagram of an application system of the method for determining article analysis data in the second embodiment of the present application
  • FIG. 6 is one of the schematic structural diagrams of the article analysis data determining device of the third embodiment of the present application.
  • FIG. 7 is the second structural diagram of the article analysis data determining device of the third embodiment of the present application.
  • Fig. 8 schematically shows a block diagram of an electronic device for executing the method according to the present application.
  • Fig. 9 schematically shows a storage unit for holding or carrying program codes for implementing the method according to the present application.
  • An article analysis data determination method disclosed in the embodiment of the present application includes: step 110 to step 130.
  • Step 110 Obtain an image containing the remaining items.
  • the items indicating the popularity of the dishes are determined according to the number of times the dishes are left and the remaining quantity each time. analyze data. For another example, by identifying and analyzing images of products remaining on shelves such as supermarkets or vending machines, analysis data indicating the popularity of the products is determined based on information such as the remaining quantity of the products each day. For another example, identifying and analyzing the images of the ingredients or meals in a buffet store, and determining the analysis data indicating the popularity of the products based on information such as the amount of ingredients or meals remaining each day.
  • the acquiring an image containing the remaining items includes: acquiring an image after a user has a meal in the store through an image acquisition device in the store The image of the table; by performing image processing on the image of the table, the image of each remaining dish contained in the image of the table is determined.
  • the image acquisition equipment in the store can be a video surveillance equipment installed in the store, or a digital camera, a smart phone, a PDA (personal digital assistant), a food ordering terminal with a camera function, and other mobile image acquisition devices manually operated by the store service personnel. equipment.
  • the store service staff collects the image of the table after the meal through the above-mentioned image acquisition device, and uses the above-mentioned image acquisition device to associate each collected table image with the collection information of the image and upload it to the cloud through the network
  • the database is stored.
  • the store service staff collects the image of the table after the meal through the above-mentioned image acquisition device, and sends each collected table image to the store through the above-mentioned image acquisition device through data transmission technology such as Bluetooth or WiFi.
  • the store cashier terminal sets the collection information for each table image separately, and the cashier terminal associates each collected table image and the collection information set for the image and uploads it to the cloud database for storage .
  • the table image includes the image of each dish remaining after the customer dining at the table, as well as the table image and other background images in the store. After the table image is collected, further image processing needs to be performed on the table image, and then the image of each remaining dish is extracted from it.
  • determining the image of each remaining dish included in the table image by performing image processing on the table image includes: sub-step S1 to sub-step S3. The specific implementation of each sub-step when performing digital image processing on each table image is described in detail below.
  • Sub-step S1 performing denoising and binarization processing on the table image to determine the table content image.
  • the cloud data processing server receives the table image uploaded by the cashier terminal or the image acquisition device, it first performs denoising and binarization processing on the received table image, removes the environmental image outside the table area, and extracts the table image.
  • Table outline and table image content (such as extracting images containing tables and tableware on the table).
  • Fig. 2 is an example of a table content image, and the table content image includes at least one tableware.
  • the collection information of the table content image is the collection information of the corresponding table image.
  • Sub-step S2 determining the center coordinates of the tableware in the table content image, and determining at least one tableware image included in the table content image according to the center coordinates of the tableware and the contour of the tableware.
  • the tableware included in the table content image is determined by image recognition technology for recognition, and the center coordinates of each tableware included in the table content image and the contour information of the corresponding tableware are determined.
  • the image area of each tableware included in the table content image is determined according to the center coordinates of each tableware and the corresponding tableware contour information, and the table content image is further cropped based on the image area where each tableware is located, and each table content image is determined.
  • the tableware image of each tableware included in the table content image is determined by image recognition technology for recognition, and the center coordinates of each tableware included in the table content image and the contour information of the corresponding tableware are determined.
  • Sub-step S3 by identifying the empty plate of the tableware image, filtering out the tableware images containing the remaining dishes that meet the preset conditions, and determining the remaining tableware images as the number of the remaining dishes included in the table image image. For each cutlery image intercepted from the table content image, the cutlery image containing the remaining dishes that meets the preset conditions (for example, , Identify tableware images that do not contain leftover dishes or identify tableware images that only contain soup or meal residues). Finally, the identified empty plate images are filtered out, and each remaining tableware image containing the remaining dishes is used as the image of the remaining dishes.
  • the collection information of the image of each remaining dish is the collection information of the table image intercepted by the corresponding tableware image.
  • the acquiring images containing the remaining dishes after a meal includes: acquiring images of the remaining dishes after the user has eaten in the store through an image acquisition device in the store.
  • the image acquisition device in the store can be a mobile image acquisition device such as a digital camera, a smart phone, a PDA (personal digital assistant), and a food ordering terminal with a camera function manually operated by the store service personnel.
  • the store service staff collects images of each remaining dish on the table after the meal through the above-mentioned image acquisition device, and associates the collected images of each of the remaining dishes with the image collection through the above-mentioned image acquisition device
  • the information is uploaded to the cloud database for storage via the network.
  • the store service staff collects the image of each remaining dish on the table after the meal through the above-mentioned image acquisition device, and uses the data transmission technology such as Bluetooth or WiFi to collect each of the images collected by the above-mentioned image acquisition device.
  • the image of the remaining dishes is sent to the cashier terminal of the store.
  • the cashier terminal of the store sets the collection information for each image separately, and the cashier terminal collects the image of each of the remaining dishes and the image set for the image.
  • the collected information is associated and uploaded to the cloud database for storage.
  • the collection information of the images of the remaining dishes includes but is not limited to any one or more of the following: collection time, collection store name, collection geographic location, collection device model, etc.
  • the remaining dishes are usually dishes whose quantity of remaining dishes meets a preset condition, for example, the remaining dishes do not include empty dishes or dishes with only soup remaining.
  • the item is a item
  • the acquiring an image containing the remaining items includes: acquiring an image through an image acquisition device in a store A shelf image of the goods is placed; by performing image processing on the shelf image, the image of each remaining product contained in the shelf image is determined.
  • the image acquisition equipment in the store can be a video surveillance equipment installed in the store, or a digital camera, a smart phone, a PDA (personal digital assistant), a tally robot with a camera function, etc., which are manually operated by the store service personnel. equipment.
  • the store collect images of each row of shelves through the above-mentioned image acquisition device, and use the above-mentioned image acquisition device to associate each acquired shelf image with the collection information of the image and upload it to the cloud through the network
  • the database is stored.
  • the shelf images collected by the above-mentioned image acquisition device, and each shelf image collected by the above-mentioned image acquisition device is sent to the cashier terminal of the store through the data transmission technology such as Bluetooth or WiFi, and then the cashier terminal of the store sends each shelf image to the cashier terminal of the store.
  • the images are respectively set with collection information, and the cashier terminal associates each collected shelf image and the collection information set for the image and uploads it to a cloud database for storage.
  • the shelf includes one or more rows of shelves, each row of shelves includes multiple rows, and each row of shelves is placed with one kind of goods or divided areas to place multiple kinds of goods.
  • the shelf image includes image content of multiple goods placement areas. After the shelf image is collected, the shelf image is identified to determine the image of the placement area of each product, that is, the image that includes the remaining products.
  • the step of performing image processing on the shelf image to determine the image of each remaining product contained in the shelf image includes: segmenting the shelf image according to the placement position of the product to determine Single product image; by identifying the single product image, filtering out the single product image containing the remaining product quantity that meets preset conditions, and determining the remaining single product image as the remaining single product images included in the shelf image The image of the item.
  • the shelf image may be segmented according to the product label to determine a single product image corresponding to each product. For example, by performing product label recognition on the shelf image shown in FIG. 3, the image area 310 corresponding to the label "goods 1" can be determined as a single product image, and the image area 320 corresponding to the label "goods 2" can be determined as An image of a single item.
  • the shelf image may also be segmented according to the similarity of the goods to determine a single product image corresponding to each product. This application does not limit the specific implementation of segmenting the shelf image according to the placement position of the goods to determine a single goods image.
  • the single product image that contains the remaining product volume and meets the preset conditions is filtered out (for example, the single product image that does not contain the product is filtered out) Single product image), the remaining single product image is determined as the image of each remaining product included in the shelf image.
  • Step 120 Determine the remaining items contained in the image and the integrity data of the remaining items by performing image processing on the image.
  • the method further includes: performing image processing on the image, determining the remaining items contained in the image and the integrity data of the remaining items, and storing the remaining items successfully identified and the integrity data in association with each other. Degree data and collection information of the image.
  • the step of performing image processing on the image to determine the remaining items contained in the image and the integrity data of the remaining items includes: performing item recognition on the image to determine The item contained in the image is used as the remaining item; performing geometric measurement based on the image of the remaining item to determine the remaining volume of the remaining item; according to the prior volume of the remaining item and the remaining volume obtained in advance The result of the comparison is to determine the completeness data of the remaining items for indicating the remaining proportion of items.
  • the performing geometric measurement based on the images of the remaining items to determine the remaining volume of the remaining items includes: performing based on at least three images containing the same remaining items collected at the same time and different angles in the same store Geometric measurement to determine the remaining volume of the remaining items.
  • images of the remaining dishes can be collected from multiple angles.
  • the dishes in the tableware can be consumed from multiple angles.
  • the completeness of the dishes contained in the tableware is usually determined by the ratio of the remaining volume of the dishes to the prior volume.
  • the prior volume of the dish is pre-stored in the cloud database as the prior volume of the corresponding dish after measuring and statistically analyzing the volume of the various dishes in the serving stage; the remaining volume of the remaining dish is obtained by real-time measurement of.
  • the prior volume of the dish is determined by the method of measuring the volume of the object based on the image in the prior art. Based on these three images, the volume of the dish contained in the tableware is calculated using aerial triangulation technology as the prior volume of the dish.
  • the dish identifier of the remaining dishes contained in the tableware (such as the name of the dish or the unique identifier of the dish in the dish data analysis system) needs to be identified first, and then the prior volume of the remaining dishes can be further determined.
  • the image of the remaining dishes that is, the image of the tableware containing the dishes
  • the dish identification of each dish in the remaining dish image is recognized.
  • the dish image recognition model can be trained by the following methods: collect the images of each dish in the dish stage as sample data, and use the dish identification corresponding to the corresponding image as the sample label to construct part of the training sample; collect the remaining dishes on the table after the meal The image of the dish is used as the sample data, and the dish identification corresponding to the corresponding image is used as the sample label to construct another part of the training sample; then, based on the constructed training sample, the neural network model is trained to obtain the dish image recognition model.
  • the training method of the dish image recognition model please refer to the prior art, which will not be repeated in the embodiment of this application.
  • the dish identifier of the dish in the image of the remaining dish can also be identified by other methods in the prior art, for example, the image of the remaining dish is combined with the dish image in the pre-built dish image database for image feature Compare, and determine the dish identifier of the dish in the image of the current remaining dish according to the result of the feature comparison.
  • This application does not limit the specific method of identifying the dish identifier in the image of the remaining dish.
  • the prior volume of the remaining dish is obtained through the correspondence between the dish identifier and the prior volume of the dish stored in the cloud database.
  • the remaining volume of the dishes contained in the tableware is determined based on the aerial triangulation technique. For example, at least three images P1, P2, and P3 of a certain remaining dish A are collected by an image acquisition device located in a store, and then, based on the at least three images P1, P2, and P3 of the remaining dish A, an aerial triangle is used The measurement technology calculates the current volume of a certain remaining dish A as the remaining volume of the remaining dish.
  • the ratio of the remaining volume and the prior volume of the remaining dish A can be used as the integrity data of the certain remaining dish A, for example .
  • the completeness data of a certain remaining dish A can be expressed as 80%. It can be seen that the completeness data of the remaining dishes reflects the remaining proportion of the dishes.
  • the performing geometric measurement based on the image of the remaining item to determine the remaining volume of the remaining item includes: performing geometric measurement based on the image of a main viewing angle of the remaining item to determine the remaining item The remaining volume.
  • the image of a certain remaining product contains a blank area and a goods area.
  • the area of the goods area is the remaining volume of the goods, and the image area is the prior volume of the goods.
  • the ratio of the area to the image area determines the integrity data of the item.
  • the integrity data of the remaining items in the image of each remaining item uploaded to the cloud can be determined.
  • the dish identifiers of the remaining items in the image of each remaining item and the corresponding dish integrity data are stored in a cloud database in association, and multiple pieces of each dish identifier and dish integrity data can be obtained.
  • the associated storage also includes image collection information (such as collection time, store, weather, geographic location, etc.) of the remaining dishes that determine the corresponding relationship.
  • a corresponding relationship between the dish identifier and the dish integrity data stored in the cloud database may include: dish name, dish integrity data, and image collection information (such as Collecting stores, collecting time, collecting weather, collecting geographic location).
  • Step 130 Perform data analysis processing on the integrity data of each remaining item and the collected information of the image containing each remaining item to determine item analysis data indicating the popularity of the item.
  • the images of the remaining dishes collected by the collection device located in the store can be uniformly stored in the image data storage space set for the store (for example, a file named after the store’s logo) In the folder).
  • the image data storage space set for the store for example, a file named after the store’s logo
  • one piece of integrity data corresponding to each remaining dish after the customer has eaten in the store can be obtained And the corresponding relationship between the collected information of the image from which the integrity data is obtained, and the obtained multiple corresponding relationships are stored in the analysis data storage space set for the store.
  • Dish analysis data indicating the popularity of the dish in the store.
  • the average of the integrity data of the same dish (such as the dish with the same name) is taken as the popularity of the dish.
  • the degree evaluation index value is used to indicate the popularity of the dish. It can be seen that the greater the remaining amount of the dish, the greater the evaluation index value of the popularity of the obtained dish.
  • the evaluation index value of the popularity of the dish is negatively correlated with the popularity of the dish, that is, the larger the evaluation index value of the popularity of the dish, the less popular the indicating dish; the evaluation index value of the popularity of the dish The smaller the indicator, the more popular it is.
  • item analysis data of different dimensions can be determined.
  • the determining the item analysis data indicating the popularity of the item by performing data analysis processing on the integrity data of each of the remaining items and the collected information of the image containing each of the remaining items includes: determining item analysis data And the value of each of the display dimensions; according to the matching relationship between the collection information of the image including the remaining items and the value of the at least one display dimension, determine the display dimension and the value of the at least one display dimension
  • Data analysis processing is performed on the integrity data of the remaining items matching the value of the at least one display dimension to determine the item analysis data indicating the popularity of the item .
  • the at least one display dimension includes any one or more of the store dimension, time dimension, geographic dimension, and weather dimension.
  • the value of the store dimension is used to indicate the store name or the store identifier of the store described in the item integrity data
  • the value of the geographic dimension is used to indicate the geographic range that the geographic location of the store described in the item integrity data conforms to (such as "Wang Jing Region”)
  • the value of the time dimension is used to indicate the time range (such as "the last three months") corresponding to the collection time of the item integrity data
  • the value of the weather dimension is used to indicate the item integrity data Corresponding weather conditions (such as "sunny day”, “rainy day”, “snowy day”, etc.) corresponding to the collected weather.
  • the value of the store dimension is used, where the value of the store dimension may be a store identifier or a store name.
  • determining the integrity data of the remaining dishes matching the store dimension value may be determining the integrity data of the dishes matching "Xiaowang's shop", for example, determining that the cloud database is the store " ⁇ ".
  • the integrity data of each remaining dish stored in the analysis data storage space of the configuration of "Wang's Shop". After that, the determined completeness data is further analyzed and processed, and the dish analysis data indicating the popularity of each remaining dish in the "Xiao Wang's shop" is determined.
  • the evaluation index for determining the popularity of "potato shreds” is 700
  • the evaluation index for determining the popularity of "braided pork” is 520, and so on.
  • the at least one display dimension of the dish analysis data includes: a geographic dimension, a time dimension, and a weather dimension
  • the geographic location range of the store determines the candidate store that matches the value of the geographic area dimension; then, the correspondence data between the dish integrity data and the collected information of each candidate store is further obtained as a data set to be analyzed.
  • the time dimension determines the time range that the collection time needs to meet in the collection information corresponding to the integrity data, filter the integrity data of the matching dishes that do not meet the time range from the data set to be analyzed, and take Corresponding relationship data between the dish integrity data and the collection information within the time range that the collection time satisfies in the corresponding collection information.
  • the weather dimension determine the weather conditions corresponding to the collected weather from the collected information corresponding to the integrity data of the dishes, and select the corresponding relationship in which the collected weather meets the determined weather conditions from the filtered data set to be analyzed Data, as the final data to be analyzed.
  • the value of the local area dimension is "Wangjing area”
  • the value of the time dimension can be “3rd quarter of 2019”
  • the value of the weather dimension can be "Rainy”
  • the completeness data of the dishes determined by the images uploaded in the third quarter and on rainy days, and the determined completeness data of the dishes are analyzed to obtain an evaluation index of the popularity of each dish.
  • the dish analysis data shown in Table 1 below can be obtained.
  • the determining at least one display dimension of the article analysis data and the value of each display dimension includes: determining the article analysis data according to the data query authority of the output object matched by the article analysis data At least one display dimension of and the value of each display dimension.
  • the dish analysis system configures different query permissions for the dish analysis data demand objects, and associates different display dimensions or display dimension combinations for different query permissions in advance.
  • the dish analysis system configures the highest-level query authority for VIP objects, and sets the display dimensions associated with the highest-level query authority: a combination of geographic, time, and weather dimensions; the dish analysis system configures general-level query authority for ordinary objects, and
  • the display dimension associated with setting the general level query permission is: store dimension.
  • the determining at least one display dimension of the article analysis data and the value of each of the display dimensions includes: determining at least one display dimension and each of the display dimensions of the article analysis data according to the input query condition The value of the display dimension.
  • the dish analysis system can provide a query condition input interface for the object, for the object to input the value of one or more query conditions such as store, geographic area, time range, weather condition, etc., and obtain the input of the object through the data interface.
  • the value of the one or more query conditions such as the store, the geographical range, the time range, and the weather conditions are taken, and the obtained value of the one or more query conditions is determined as the value of each display dimension of the dish analysis data.
  • the evaluation index value of the popularity of each item can be obtained based on the accumulation and sum of all item integrity data that meets the display dimension condition of the item stored in the cloud database. For example, for the dish “Kung Pao Chicken”, stores in Wangjing area uploaded the remaining images of Gong Pao Chicken in the third quarter of 2019 and on rainy days to determine the completeness of the dish. There are 1000 pieces of data, namely: 60%, 75%, 50%... After adding up the completeness data of these 1,000 dishes, the obtained sum can be used as the evaluation index value of the popularity of the dish "Kung Pao Chicken”.
  • the ordering frequency of different dishes is different, and the probability of being left after a meal is also different.
  • the more frequently ordered dishes the greater the probability of being left after the meal, which will lead to the following methods
  • the larger the evaluation index value of the calculated popularity is. From the above calculation method, it can be seen that the larger the evaluation index value of popularity, the less popular the dish is, that is, for dishes with a high frequency of ordering, the calculated popularity evaluation index value may indicate incorrectly Popularity of dishes.
  • a weight value for calculating the popularity evaluation index of the item is set for each item.
  • the corresponding relationship with the evaluation index value indicating the popularity of the item and the value of the at least one display dimension generates article analysis data indicating the popularity of the item.
  • the weight of each item is determined according to business needs. Taking the application scenario where the item is a dish as an example, for ordinary dishes such as "rice” with a higher ordering frequency, a lower weight can be set, while for Signature dishes such as “Roast Duck” can be set with higher weights. By setting different weights for different dishes, the clarity of the calculated evaluation index value on the popularity of the dishes is improved.
  • the weight of the dish “Gong Bao Chicken” is set to 0.8
  • the weight of the dish “Pepper Potato Shredded” is set to 0.3
  • Store X has 50 pieces of completeness data for the remaining dishes of "Kung Pao Chicken” and “Pepper Potato Shredded” uploaded in the third quarter of 2019 and on rainy days, and the completeness data values are both: 50%
  • the evaluation index value indicating the popularity of the dish "Gong Bao Chicken” can be obtained as 20.
  • the evaluation index value indicating the popularity of the dish “spicy pepper potato shreds” can be obtained as 7.5.
  • Table 2 shows the generated dish analysis data generated according to the evaluation index values of the dishes “Kung Pao Chicken” and “Hot Pepper Potato Shreds” indicating the popularity of the dishes.
  • the method for determining item analysis data disclosed in the embodiments of the present application obtains an image containing remaining items; performs image processing on the image to determine the remaining items contained in the image and the integrity data of the remaining items; The integrity data of each of the remaining items and the collected information of the images containing each of the remaining items are processed for data analysis to determine the item analysis data indicating the popularity of the item, which helps to improve the determination of the item analysis data Accuracy.
  • the method for determining item analysis data disclosed in the embodiments of the present application collects images of after-dinner dishes and performs image processing to determine the remaining times of each dish and the amount of remaining dishes each time, and then, according to the remaining times and the amount of each dish
  • the amount of remaining dishes each time determines the evaluation index value indicating the popularity of the dish, and generates dish analysis data based on the evaluation index value of each dish, and calculates it based on the image data of the after-dinner dish collected by the image acquisition device located in the dining store
  • the evaluation index value of the dish has more objective data sources, and it is easier to collect a large amount of data, so that the evaluation index value calculated based on the large amount of data collected objectively reflects the popularity of the dish more objectively and accurately.
  • An article analysis data determination method disclosed in an embodiment of the present application includes: step 410 to step 440.
  • Step 410 Obtain an image containing the remaining items.
  • Step 420 Determine the remaining items contained in the image and the integrity data of the remaining items by performing image processing on the image.
  • Embodiment 1 For a specific implementation manner of determining the remaining items included in the image and the integrity data of the remaining items by performing image processing on the image, refer to Embodiment 1, which will not be repeated in this embodiment.
  • Step 430 Perform data analysis processing on the integrity data of each remaining item and the collected information of the image containing each remaining item to determine item analysis data indicating the popularity of the item.
  • Example 1 By performing data analysis processing on the integrity data of each of the remaining items and the collected information of the images containing each of the remaining items, the specific implementation of determining the item analysis data indicating the popularity of the item is shown in Example 1. This embodiment will not be repeated.
  • the integrity data of the remaining items is used for the remaining proportion of the items.
  • the collection information of the image including each of the remaining items includes information such as the collection store, collection time, collection weather, and collection geographic location of the image.
  • the item analysis data indicating the popularity of the item can be determined .
  • the evaluation index value of the popularity of dishes in a certain store can be determined.
  • a ranking list of the unpopular degrees of the dishes is generated. Wherein, the evaluation index value is calculated based on several pieces of integrity data of the corresponding dish.
  • Step 440 Display the article analysis data according to the determined at least one display dimension.
  • At least one display dimension of the data can be determined according to the display requirements of the item analysis data. Then, when the item data analysis processing is performed, the integrity data of each remaining item can be determined according to the determined display dimension, As well as the collection of stores, collection time, collection of weather, collection of geographic location and other information associated with each completeness data for data analysis and processing. For example, when the determined display dimensions include store and weather, the item integrity data and collected weather can be analyzed and processed based on the values of the determined store and weather dimensions to obtain the information of each item in the designated store under the specified weather conditions. The evaluation index value of the popularity of the item.
  • the dish information analysis system includes: an image acquisition device 510, a cash register terminal 520, a cloud server 530, and a restaurant management system 540.
  • the dish information analysis system includes: an image acquisition device 510, a cash register terminal 520, a cloud server 530, and a restaurant management system 540.
  • the working principles of each device or system are introduced below.
  • the image acquisition device 510 is located in the store, and is used to acquire images of the remaining dishes after the customer has eaten in the store. For example, through the image acquisition device 510 in the store, the image of the table after the user has eaten in the store is acquired; or, Through the image acquisition device 510 in the store, images of the remaining dishes after the user has eaten in the store are acquired.
  • the image acquisition device 510 may be: a video surveillance device, or a digital camera, a smart phone, a PDA (personal digital assistant), a food ordering terminal with a camera function, and other devices manually operated by a store service staff.
  • the cash register terminal 520 is located in the store and is used for meal settlement; the cash register terminal 520 can also be used to receive the image uploaded by the image acquisition device 510 in the store, and de-duplicate and preliminarily process the image. After editing, upload to the cloud server 530.
  • the cash register terminal 520 performs deduplication processing on the images according to the image acquisition device identification and upload time of each image received, and the similarity of the images, and filters out images that are repeatedly uploaded.
  • the cash register terminal 520 adds a store identifier or a store geographic location to the received image to collect geographic location information.
  • the cloud server 530 also includes a cloud database, and the cloud server 530 is configured to receive images containing remaining dishes uploaded by the cashier terminal 520 of each store, and store the received images in the cloud database.
  • the cloud server 530 is also used to check the table Image processing is performed on the image, the image of each remaining dish included in the table image is determined, and the image of each remaining dish is stored in the cloud database.
  • determining the images of the remaining dishes included in the table image by performing image processing on the table image refer to the first embodiment, which will not be repeated in this embodiment.
  • the cloud server 530 is also used to determine the remaining dishes contained in the image and the integrity data of the remaining dishes by performing image processing on the images of the remaining dishes; , And the collected information of the image containing each of the remaining dishes are subjected to data analysis processing, and the dish analysis data indicating the popularity of the dishes is determined.
  • image processing on the image of the remaining dishes the specific implementation of determining the remaining dishes contained in the image and the completeness data of the remaining dishes can be found in the first embodiment by performing image processing on the image of the remaining dishes , The specific implementation manner for determining the remaining items contained in the image and the integrity data of the remaining items will not be repeated in this embodiment.
  • the specific implementation of determining the dish analysis data indicating the popularity of the dishes can be found in Example 1. Details are not repeated in this embodiment.
  • the restaurant management system 540 is configured to obtain the dish analysis data indicating the popularity of the dish through the cloud server 530, and display the dish analysis data according to at least one determined display dimension.
  • the method further includes: displaying the article analysis data according to the determined at least one display dimension. For example, after analyzing and processing the completeness data of the dishes and the collected weather based on the values of the two dimensions of the determined store and the weather, and obtaining the evaluation index value of the popularity of each dish in the specified store under the specified weather conditions, you can follow The store and weather dimensions display the evaluation index values of the popularity of dishes that satisfy the combination of the value of the store dimension and the value of the weather dimension.
  • the method for determining item analysis data disclosed in the embodiments of the present application obtains an image containing remaining items; performs image processing on the image to determine the remaining items contained in the image and the integrity data of the remaining items; Perform data analysis processing on the integrity data of each of the remaining items and the collected information of the image containing each of the remaining items to determine the item analysis data indicating the popularity of the item, and finally, display according to the determined at least one item Dimension, displaying the item analysis data helps to display the data analysis results reflecting the item’s popularity more objectively and accurately, so that each item supply store can understand the item’s popularity in real time and accurately. Provide data support for item improvements and user dining experience.
  • An article analysis data determination device disclosed in an embodiment of the present application, as shown in FIG. 6, includes:
  • the remaining item image acquisition module 610 is used to acquire an image containing the remaining item
  • the item integrity data determining module 620 is configured to determine the remaining items contained in the image and the integrity data of the remaining items by performing image processing on the image;
  • the item analysis data determination module 630 is used for determining the item analysis data indicating the popularity of the item by performing data analysis processing on the integrity data of each remaining item and the collected information including the image of each remaining item .
  • the item analysis that indicates the popularity of the item is determined by performing data analysis processing on the integrity data of each of the remaining items and the collected information of the image containing each of the remaining items Data, including:
  • the performing data analysis processing on the integrity data of the remaining items matching the value of the at least one display dimension to determine the item analysis data indicating the popularity of the item includes:
  • item analysis data indicating the popularity of the item is generated.
  • the at least one display dimension includes any one or more of the store dimension, time dimension, geographic dimension, and weather dimension.
  • the device further includes:
  • the display module 640 is configured to display the article analysis data according to the determined at least one display dimension.
  • the item is a dish
  • the remaining item image acquisition module 610 is further configured to:
  • the image of each remaining dish included in the table image is determined.
  • the step of performing image processing on the table image to determine the image of each remaining dish included in the table image includes:
  • the tableware images containing the remaining dishes that meet the preset condition are filtered out, and the remaining tableware images are determined as the images of the remaining dishes included in the table image.
  • the item is a dish
  • the remaining item image acquisition module 610 is further configured to:
  • the image of each remaining dish after the user has eaten in the store is acquired through the image acquisition device in the store.
  • the item is a product
  • the remaining item image acquisition module 610 is further configured to:
  • the image of each remaining item included in the shelf image is determined.
  • the step of performing image processing on the shelf image to determine the image of each remaining product included in the shelf image includes:
  • the single product image containing the remaining product quantity that meets the preset condition is filtered out, and the remaining single product image is determined as the image of each remaining product included in the shelf image.
  • the item integrity data determining module 620 is further configured to:
  • the integrity data of the remaining item used to indicate the remaining proportion of the item is determined.
  • the article analysis data determination device disclosed in the embodiment of this application is used to implement the article analysis data determination method described in the first embodiment or the second embodiment of this application.
  • the specific implementation of each module of the device will not be repeated, please refer to the method implementation Examples of the specific implementation of the corresponding steps.
  • An article analysis data determining device disclosed in an embodiment of the present application obtains an image containing remaining items; and determines the remaining items contained in the image and the integrity data of the remaining items by performing image processing on the image; By performing data analysis processing on the integrity data of each of the remaining items and the collected information of the images containing each of the remaining items, the item analysis data indicating the popularity of the item is determined, which helps to improve the determined item analysis The accuracy of the data.
  • the article analysis data determination device disclosed in the embodiment of the present application collects images of after-dinner dishes and performs image processing to determine the remaining times of each dish and the remaining amount of each dish, and then, according to the remaining times and the amount of each dish
  • the amount of remaining dishes each time determines the evaluation index value indicating the popularity of the dish, and generates dish analysis data based on the evaluation index value of each dish, and calculates it based on the image data of the after-dinner dish collected by the image acquisition device located in the dining store
  • the evaluation index value of the dish has more objective data sources, and it is easier to collect a large amount of data, so that the evaluation index value calculated based on the large amount of data collected objectively reflects the popularity of the dish more objectively and accurately.
  • the device embodiments described above are merely illustrative.
  • the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in One place, or it can be distributed to multiple network units.
  • Some or all of the modules can be selected according to actual needs to achieve the objectives of the solutions of the embodiments. Those of ordinary skill in the art can understand and implement it without creative work.
  • the various component embodiments of the present application may be implemented by hardware, or by software modules running on one or more processors, or by a combination of them.
  • a microprocessor or a digital signal processor (DSP) may be used in practice to implement some or all of the functions of some or all of the components in the electronic device according to the embodiments of the present application.
  • This application can also be implemented as a device or device program (for example, a computer program and a computer program product) for executing part or all of the methods described herein.
  • Such a program for implementing the present application may be stored on a computer-readable medium, or may have the form of one or more signals.
  • Such a signal can be downloaded from an Internet website, or provided on a carrier signal, or provided in any other form.
  • FIG. 8 shows an electronic device that can implement the method according to the present application.
  • the electronic device may be a PC, a mobile terminal, a personal digital assistant, a tablet computer, etc.
  • the electronic device traditionally includes a processor 820, a memory 810, and a program code 830 that is stored on the memory 810 and can run on the processor 820.
  • the processor 820 executes the program code 830, The method described.
  • the memory 810 may be a computer program product or a computer readable medium.
  • the memory 810 may be an electronic memory such as flash memory, EEPROM (Electrically Erasable Programmable Read Only Memory), EPROM, hard disk, or ROM.
  • the memory 810 has a storage space 8101 of the program code 830 of the computer program for executing any method steps in the above-mentioned method.
  • the storage space 8101 for the program code 830 may include various computer programs respectively used to implement various steps in the above method.
  • the program code 830 is computer readable code. These computer programs can be read from or written into one or more computer program products. These computer program products include program code carriers such as hard disks, compact disks (CDs), memory cards, or floppy disks.
  • the computer program includes computer-readable code, which when run on an electronic device, causes the electronic device to execute the method according to the above-mentioned embodiment.
  • the embodiment of the present application also discloses a computer-readable storage medium on which a computer program is stored.
  • the program is executed by a processor, the steps of the article analysis data determination method described in the first or second embodiment of the present application are realized. .
  • Such a computer program product may be a computer-readable storage medium, and the computer-readable storage medium may have storage segments, storage spaces, etc., arranged similarly to the storage 810 in the electronic device shown in FIG. 8.
  • the program code may be compressed and stored in the computer-readable storage medium in an appropriate form, for example.
  • the computer-readable storage medium is usually a portable or fixed storage unit as described with reference to FIG. 9.
  • the storage unit includes computer readable codes 830', which are codes read by a processor, and when executed by the processor, these codes implement each step in the method described above.
  • any reference signs placed between parentheses should not be constructed as a limitation to the claims.
  • the word “comprising” does not exclude the presence of elements or steps not listed in the claims.
  • the word “a” or “an” preceding an element does not exclude the presence of multiple such elements.
  • the application can be realized by means of hardware including several different elements and by means of a suitably programmed computer. In the unit claims listing several devices, several of these devices may be embodied in the same hardware item.
  • the use of the words first, second, and third, etc. do not indicate any order. These words can be interpreted as names.

Landscapes

  • Engineering & Computer Science (AREA)
  • Quality & Reliability (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Image Analysis (AREA)

Abstract

本申请公开了一种物品分析数据确定方法，属于计算机技术领域。本申请实施例公开的物品分析数据确定方法包括：获取包含剩余物品的图像；通过对所述图像进行图像处理，确定所述图像中包含的剩余物品以及所述剩余物品的完整度数据；通过对各所述剩余物品的完整度数据，以及包含各所述剩余物品的所述图像的采集信息进行数据分析处理，确定指示物品受欢迎程度的物品分析数据。

Description

确定物品分析数据
本申请要求在2020年02月20日提交中国专利局、申请号为202010105304.X、发明名称为“物品分析数据确定方法、装置、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
本申请实施例涉及计算机技术领域，特别是涉及确定物品分析数据。
背景技术
分析物品的受欢迎程度，可以为改善物品的供给量以及物品优化等提供数据参考。例如，分析菜品的受欢迎程度，可以为餐饮商家改进菜品或优化菜品供应品类，以提升用户满意度提供依据。现有技术中，通常通过菜品的销量、用户的评价、打分等数据分析菜品的受欢迎程度，得到指示菜品受欢迎程度的指标值。然而，现有技术中的上述方法确定的菜品受欢迎程度的评价指标值的准确率在某些情况下并不能准确地反映菜品的受欢迎程度。例如，某些用户在用餐后不对菜品进行评价和打分，会导致根据菜品的评价或打分确定的菜品受欢迎程度的评价指标值由于数据稀疏而不准确。
发明内容
第一方面，本申请实施例提供了一种物品分析数据确定方法，包括：
获取包含剩余物品的图像；
通过对所述图像进行图像处理，确定所述图像中包含的剩余物品以及所述剩余物品的完整度数据；
通过对各所述剩余物品的完整度数据，以及包含各所述剩余物品的所述图像的采集信息进行数据分析处理，确定指示物品受欢迎程度的物品分析数据。
第二方面，本申请实施例提供了一种物品分析数据确定装置，包括：
剩余物品图像获取模块，用于获取包含剩余物品的图像；
物品完整度数据确定模块，用于通过对所述图像进行图像处理，确定所述图像中包含的剩余物品以及所述剩余物品的完整度数据；
物品分析数据确定模块，用于通过对各所述剩余物品的完整度数据，以及包含各所述剩余物品的所述图像的采集信息进行数据分析处理，确定指示 物品受欢迎程度的物品分析数据。
第三方面，本申请实施例还公开了一种电子设备，包括存储器、处理器及存储在所述存储器上并可在处理器上运行的计算机程序，所述处理器执行所述计算机程序时实现本申请实施例所述的物品分析数据确定方法。
第四方面，本申请实施例提供了一种计算机可读存储介质，其上存储有计算机程序，该程序被处理器执行时本申请实施例公开的物品分析数据确定方法的步骤。
本申请实施例公开的物品分析数据确定方法，通过获取包含剩余物品的图像；通过对所述图像进行图像处理，确定所述图像中包含的剩余物品以及所述剩余物品的完整度数据；通过对各所述剩余物品的完整度数据，以及包含各所述剩余物品的所述图像的采集信息进行数据分析处理，确定指示物品受欢迎程度的物品分析数据，有助于提升确定的物品分析数据的准确度。
上述说明仅是本申请技术方案的概述，为了能够更清楚了解本申请的技术手段，而可依照说明书的内容予以实施，并且为了让本申请的上述和其它目的、特征和优点能够更明显易懂，以下特举本申请的具体实施方式。
附图说明
为使本申请实施例的目的、技术方案和优点更加清楚，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作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。
图1是本申请实施例一的物品分析数据确定方法流程图；
图2是本申请实施例一中采集的桌台内容图像的一个示例；
图3是本申请实施例一中采集的货架图像的一个示例；
图4本申请实施例二的物品分析数据确定方法流程图；
图5本申请实施例二的物品分析数据确定方法应用系统示意图；
图6是本申请实施例三的物品分析数据确定装置结构示意图之一；
图7是本申请实施例三的物品分析数据确定装置结构示意图之二；
图8示意性地示出了用于执行根据本申请的方法的电子设备的框图；以及
图9示意性地示出了用于保持或者携带实现根据本申请的方法的程序 代码的存储单元。
具体实施例
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作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。
实施例一
本申请实施例公开的一种物品分析数据确定方法，如图1所示，所述方法包括：步骤110至步骤130。
步骤110，获取包含剩余物品的图像。
本申请的一些实施例中，通过对若干顾客在门店用餐后餐桌上剩余的菜品的图像进行识别分析，根据菜品被剩余的次数和每次剩余数量的多少等信息确定指示菜品受欢迎程度的物品分析数据。再例如，通过对超市或自动售货机等货架上剩余的货品的图像进行识别分析，根据货品每天剩余数量的多少等信息确定指示货品受欢迎程度的分析数据。又例如，通过自助餐门店的食材或餐食的图像进行识别分析，根据食材或餐食每天剩余数量的多少等信息确定指示货品受欢迎程度的分析数据。
在确定物品分析数据之前，首先需要获取包含剩余物品的若干图像。下面详细描述不同应用场景下剩余物品图像的技术方案。
本申请的一些实施例中，以所述物品为餐厅为顾客提供的菜品为例，所述获取包含剩余物品的图像，包括：通过门店内的图像采集设备，获取用户在所述门店内用餐后的桌台图像；通过对所述桌台图像进行图像处理，确定所述桌台图像中包含的各剩余菜品的图像。其中，门店内的图像采集设备可以为设置于门店内的视频监控设备，或门店服务人员手动操作的数码相机、智能手机、PDA(个人数字助理)、具有拍照功能的点餐终端等移动图像采集设备。顾客在门店结束用餐后，门店服务人员通过上述图像采集设备采集用餐后的桌台图像，并通过上述图像采集设备将采集的每幅桌台图像关联该图像的采集信息一并通过网络上传至云端数据库进行存储。或者，顾客在门店结束用餐后，门店服务人员通过上述图像采集设备采集用餐后的桌台图像，并通过蓝牙或WiFi等数据传输技术由上述图像采集设备将采集的每幅桌台图像发送至门店的收银终端，然后，由门店收银终端为每幅桌台图像分别设 置采集信息，并由所述收银终端将采集的每幅桌台图像以及为该图像设置的采集信息关联上传至云端数据库进行存储。
通常，所述桌台图像包括就餐于该桌的顾客用餐后剩余的每个菜品的图像，以及，桌台图像和门店内的其他背景的图像。在采集到桌台图像后，还需要对桌台图像进一步执行图像处理，之后从中提取每个剩余菜品的图像。本申请的一些实施例中，所述通过对所述桌台图像进行图像处理，确定所述桌台图像中包含的各剩余菜品的图像，包括：子步骤S1至子步骤S3。下面详细介绍对每幅桌台图像进行数字图像处理时各子步骤的具体实施方式。
子步骤S1，对所述桌台图像进行去噪和二值化处理，确定桌台内容图像。例如，云端数据处理服务器在接收到收银终端或图像采集设备上传的桌台图像之后，首先对接收到的桌台图像进行去噪和二值化处理，去掉桌台区域外的环境图像，提取桌台轮廓和桌台图像内容(如提取包含桌台及桌台上餐具的图像)。图2为桌台内容图像的一个示例，所述桌台内容图像中至少包括一个餐具。所述桌台内容图像的采集信息为相应桌台图像的采集信息。
子步骤S2，确定所述桌台内容图像中的餐具中心坐标，并根据所述餐具中心坐标和餐具轮廓确定所述桌台内容图像中包括的至少一个餐具图像。例如，通过图像识别技术确定所述桌台内容图像中包括的餐具进行识别，确定所述桌台内容图像中包括的各餐具的中心坐标和相应餐具的轮廓信息。之后，根据每个餐具中心坐标和相应的餐具轮廓信息确定所述桌台内容图像中包括的各餐具所在图像区域，进一步基于各餐具所在图像区域对所述桌台内容图像进行裁剪，确定各所述桌台内容图像中包括的各餐具的餐具图像。
子步骤S3，通过对所述餐具图像进行空盘识别，过滤掉包含剩余菜品量符合预设条件的餐具图像，将剩余的所述餐具图像确定为所述桌台图像中包含的各剩余菜品的图像。对于从所述桌台内容图像中截取的每个餐具图像，通过现有技术中的空盘识别技术或者通过预先训练的空盘识别模型识别出包含剩余菜品量符合预设条件的餐具图像(例如，识别出不包含剩余菜品的餐具图像或者识别出仅包含汤汁、或饭菜残渣的餐具图像)。最后，过滤掉识别出的空盘图像，将剩余的包含剩余菜品的每个餐具图像分别作为剩余菜品的图像。每个剩余菜品的图像的采集信息为对应的餐具图像所截取至的桌台图像的采集信息。
本申请的另一些实施例中，所述获取包含餐后剩余菜品的图像，包括：通过门店内的图像采集设备，获取用户在所述门店内用餐后各剩余菜品的图 像。其中，门店内的图像采集设备可以为门店服务人员手动操作的数码相机、智能手机、PDA(个人数字助理)、具有拍照功能的点餐终端等移动图像采集设备。顾客在门店结束用餐后，门店服务人员通过上述图像采集设备采集用餐后桌台上每个剩余菜品的图像，并通过上述图像采集设备将采集的每个所述剩余菜品的图像关联该图像的采集信息一并通过网络上传至云端数据库进行存储。或者，顾客在门店结束用餐后，门店服务人员通过上述图像采集设备采集用餐后桌台上每个剩余菜品的图像，并通过蓝牙或WiFi等数据传输技术由上述图像采集设备将采集的每个所述剩余菜品的图像发送至门店的收银终端，然后，由门店收银终端为每幅图像分别设置采集信息，并由所述收银终端将采集的每个所述剩余菜品的图像以及为该图像设置的采集信息关联上传至云端数据库进行存储。其中，剩余菜品的图像的采集信息包括但不限于以下任意一项或多项：采集时间、采集门店名称、采集地理位置、采集设备型号等。所述剩余菜品通常为剩余菜量满足预设条件的菜品，例如，所述剩余菜品不包括空盘菜品或仅剩余汤汁的菜品。
本申请的另一些实施例中，以所述物品为超市货架上摆放的货品为例，所述物品为货品，所述获取包含剩余物品的图像，包括：通过门店内的图像采集设备，获取摆放货品的货架图像；通过对所述货架图像进行图像处理，确定所述货架图像中包含的各剩余货品的图像。其中，门店内的图像采集设备可以为设置于门店内的视频监控设备，或门店服务人员手动操作的数码相机、智能手机、PDA(个人数字助理)、具有拍照功能的理货机器人等移动图像采集设备。每天的预设时间，如在门店打烊之后，通过上述图像采集设备采集各列货架的图像，并通过上述图像采集设备将采集的每幅货架图像关联该图像的采集信息一并通过网络上传至云端数据库进行存储。或者，通过上述图像采集设备采集的货架图像，并通过蓝牙或WiFi等数据传输技术由上述图像采集设备将采集的每幅货架图像发送至门店的收银终端，然后，由门店收银终端为每幅货架图像分别设置采集信息，并由所述收银终端将采集的每幅货架图像以及为该图像设置的采集信息关联上传至云端数据库进行存储。
通常，所述货架包括一列或多列货架，每列货架包括多行，每行货架摆放一种货品或分区域摆放多种货品。如图3所示，所述货架图像中包括多个货品摆放区域的图像内容。在采集到货架图像后，通过对货架图像进行摆放区域识别，确定每种货品的摆放区域的图像，即包括剩余货品的图像。本申 请的一些实施例中，所述通过对所述货架图像进行图像处理，确定所述货架图像中包含的各剩余货品的图像，包括：根据货品摆放位置对所述货架图像进行分割，确定单一货品图像；通过对所述单一货品图像进行识别，过滤掉包含剩余货品量符合预设条件的所述单一货品图像，将剩余的所述单一货品图像确定为所述货架图像中包含的各剩余货品的图像。
本申请的一些实施例中，可以根据货品标签对所述货架图像进行分割，确定每种货品对应的单一货品图像。例如，通过对图3中所示的货架图像进行货品标签识别，可以将标签“货品1”对应的图像区域310确定为一幅单一货品图像，将标签“货品2”对应的图像区域320确定为一幅单一货品图像。本申请的另一些实施例中，还可以根据货品的相似度对所述货架图像进行分割，确定每种货品对应的单一货品图像。本申请对根据货品摆放位置对所述货架图像进行分割，确定单一货品图像的具体实施方式不做限定。在确定了货架图像中包括的多个单一货品图像之后，通过对所述单一货品图像进行识别，过滤掉包含剩余货品量符合预设条件的所述单一货品图像(例如，过滤掉不包含货品的单一货品图像)，将剩余的所述单一货品图像确定为所述货架图像中包含的各剩余货品的图像。
步骤120，通过对所述图像进行图像处理，确定所述图像中包含的剩余物品以及所述剩余物品的完整度数据。
在确定了包含剩余物品的图像之后，还包括：对所述图像进行图像处理，确定所述图像中包含的剩余物品以及所述剩余物品的完整度数据，并关联存储识别成功的剩余物品及完整度数据以及所述图像的采集信息。
本申请的一些实施例中，所述通过对所述图像进行图像处理，确定所述图像中包含的剩余物品以及所述剩余物品的完整度数据，包括：通过对所述图像进行物品识别，确定所述图像中包含的物品作为剩余物品；基于所述剩余物品的所述图像进行几何测量，确定所述剩余物品的剩余体积；根据预先获取的所述剩余物品的先验体积和所述剩余体积的比较结果，确定所述剩余物品的用于指示物品剩余比例的完整度数据。
对于不同应用场景，不同物品的盛放方式或摆放方式不同，图像的采集方式不同，因此，根据剩余物品的图像确定物品的剩余体积的方式也会有所不同。所述基于所述剩余物品的所述图像进行几何测量，确定所述剩余物品的剩余体积，包括：基于在同一门店同一时间不同角度采集的包含同一所述剩余物品的至少三幅所述图像进行几何测量，确定所述剩余物品的剩余体 积。
对于摆放位置比较分散，摆放在比较大的空间中的物品，如餐厅中餐桌上的菜品举例，可以从多个角度采集剩余菜品的图像。并且，餐具中菜品可以从多个角度消耗，为了提升菜品剩余体积测量的准确度，需要从多个角度采集图像并进行体积计算。餐具中所盛放的菜品的完整度通常通过菜品的剩余体积和先验体积的比例确定的。其中，菜品的先验体积通过对出菜环节各种菜品的体积进行测量和统计分析后，作为相应菜品的先验体积预先存储在云端数据库的；所述剩余菜品的剩余体积则是实时测量得到的。
本申请的一些实施例中，菜品的先验体积通过现有技术中基于图像测量物体体积的方法确定，例如，在出菜阶段，通过采集出菜口某一菜品的至少三个不同角度的三张图像，基于这三张图像，采用空中三角测量技术计算餐具中盛放的该菜品的体积，作为该菜品的先验体积。
本申请的一些实施例中，首先需要识别出餐具中盛放的剩余菜品的菜品标识(如菜品名称或菜品数据分析系统中菜品的唯一标识)，然后才能进一步确定该剩余菜品的先验体积。例如，通过预先训练的菜品图像识别模型对剩余菜品的图像(即盛有菜品的餐具图像)进行图像识别，识别出每个剩余菜品图像中菜品的菜品标识。其中，菜品图像识别模型可以通过以下方法训练：采集出菜阶段的各菜品的图像作为样本数据，以相应图像对应的菜品标识作为样本标签，构建部分训练样本；采集用餐后桌台上剩余的各菜品的图像作为样本数据，以相应图像对应的菜品标识作为样本标签，构建另一部分训练样本；之后，基于构建的训练样本训练神经网络模型，得到菜品图像识别模型。菜品图像识别模型的训练方法参见现有技术，本申请实施例中不再赘述。
本申请的一些实施例中，还可以通过现有技术中的其他方式识别剩余菜品的图像中菜品的菜品标识，例如，将剩余菜品的图像与预先构建的菜品图像数据库中的菜品图像进行图像特征比对，根据特征比对结果，确定当前剩余菜品的图像中菜品的菜品标识。本申请对识别剩余菜品的图像中的菜品标识的具体方法不做限定。
在识别出剩余菜品的图像中的剩余菜品的菜品标识之后，通过云端数据库中存储的菜品标识与菜品的先验体积的对应关系，获取剩余菜品的先验体积。
对于物品为货品、自助餐食等形式时，可以采用与菜品相似的技术手段， 或采用现有技术中的技术手段对所述图像进行物品识别，确定所述图像中包含的物品作为剩余物品，本申请实施例中不再赘述。
本申请的一些实施例中，基于空中三角测量技术，确定餐具中盛放的菜品的剩余体积。例如，通过位于门店内的图像采集设备采集某一剩余菜品A的至少三幅图像P1、P2和P3，然后，基于该剩余菜品A的所述至少三幅图像P1、P2和P3，采用空中三角测量技术计算所述某一剩余菜品A的当前体积，作为该剩余菜品的剩余体积。
在确定了某一剩余菜品A的剩余体积和先验体积之后，可以将所述剩余菜品A的所述剩余体积和先验体积的比值，作为所述某一剩余菜品A的完整度数据，例如，某一剩余菜品A的完整度数据可以表示为80％。可见，剩余菜品的完整度数据反映了菜品剩余比例。
对于摆放位置比较紧凑，只能从正面采集图像的物品，如超市中货架上摆放的商品、自动售货机中摆放的商品，可以从主视角度(即货品摆放位置的正面)采集剩余货品的一幅图像，然后，基于所述剩余货品的一幅主视角度的图像进行几何测量，确定所述剩余物品的剩余体积。所述基于所述剩余物品的所述图像进行几何测量，确定所述剩余物品的剩余体积，包括：基于所述剩余物品的一幅主视角度的所述图像进行几何测量，确定所述剩余物品的剩余体积。
例如，对于货架上摆放的货品，某个剩余货品的图像中包含空白区域和货品区域，其中，货品区域的面积为货品的剩余体积，图像面积为货品的先验体积，之后，根据货品区域面积占图像面积的比例确定该货品的完整度数据。本申请的其他实施例中，还可以通过其他方式基于所述剩余物品的一幅主视角度的所述图像进行几何测量，确定所述剩余物品的剩余体积，本申请实施例中不再一一例举。
按照上述方法，可以确定上传至云端的每个剩余物品的图像中剩余物品的完整度数据。本申请的一些实施例中，将每幅剩余物品的图像中剩余物品的菜品标识，以及对应的菜品完整度数据关联存储在云端数据库中，可以得到每个菜品标识与菜品完整度数据的多条对应关系。进一步的，对于每条菜品标识与菜品完整度数据的一条对应关系，关联存储的还包括确定该条对应关系的剩余菜品的图像的采集信息(如采集时间、门店、天气、地理位置等)。本申请的一些实施例中，云端数据库存储的菜品标识与菜品的完整度数据的一条对应关系可以包括：菜品名称、菜品的完整度数据、得到该菜品的完整 度数据的图像的采集信息(如采集门店、采集时间、采集天气、采集地理位置)。
步骤130，通过对各所述剩余物品的完整度数据，以及包含各所述剩余物品的所述图像的采集信息进行数据分析处理，确定指示物品受欢迎程度的物品分析数据。
仍以物品为菜品为例，对于每个门店，可以将位于该门店的采集设备采集的剩余菜品的图像统一存储在为该门店设置的图像数据存储空间中(例如，以该门店标识命名的文件夹中)。之后，对于每个所述门店，通过对存储在为该门店设置的图像数据存储空间中的剩余菜品的图像进行分析处理，可以得到顾客在该门店用餐后每个剩余菜品对应的一条完整度数据和得到该完整度数据的图像的采集信息之间的对应关系，并将得到的上述多条对应关系存储在为该门店设置的分析数据存储空间中。进一步的，对于某个门店，通过对为该门店设置的分析数据存储空间中存储的菜品的完整度数据与得到该完整度数据的图像的采集信息之间的对应关系进行分析处理，可以得到该门店的指示菜品受欢迎程度的菜品分析数据。例如，对于门店X，通过对与门店X对应的分析数据存储空间中的完整度数据进行分析，将同一个菜品(如菜品名称相同的菜品)的完整度数据的平均值作为该菜品的受欢迎程度评价指标值，用于指示该菜品受欢迎程度。可见，菜品的剩余量越大，得到的菜品受欢迎程度的评价指标值越大。本申请实施例中，菜品受欢迎程度的评价指标值与菜品的受欢迎程度负相关，即菜品受欢迎程度的评价指标值越大，指示菜品越不受欢迎；菜品受欢迎程度的评价指标值越小，指示菜品越受欢迎。
本申请的一些实施例中，根据对物品分析数据的不同展示维度的需求，可以确定不同维度的物品分析数据。所述通过对各所述剩余物品的完整度数据，以及包含各所述剩余物品的所述图像的采集信息进行数据分析处理，确定指示物品受欢迎程度的物品分析数据，包括：确定物品分析数据的至少一个展示维度及各所述展示维度的取值；根据包含所述剩余物品的所述图像的采集信息与所述至少一个展示维度的取值的匹配关系，确定与所述至少一个展示维度的取值匹配的所述剩余物品的完整度数据；对与所述至少一个展示维度的取值匹配的所述剩余物品的完整度数据进行数据分析处理，确定指示物品受欢迎程度的物品分析数据。其中，所述至少一个展示维度包括：门店维度、时间维度、地域维度、天气维度中的任意一个或多个维度。其中，门 店维度的取值用于指示物品完整度数据所述门店的门店名称或门店标识，地域维度的取值用于指示物品完整度数据所述门店的地理位置符合的地理范围(如“望京地区”)；所述时间维度的取值用于指示物品完整度数据对应的采集时间符合的时间范围(如“最近三个月”)；所述天气维度的取值用于指示物品完整度数据对应的采集天气符合的天气情况(如“晴天”、“雨天”、“雪天”等)。
例如，当所述菜品分析数据的至少一个展示维度为门店维度时，根据门店维度的取值，其中，门店维度的取值可以为门店标识或门店名称。具体到本实施例而言，确定与所述门店维度取值匹配的剩余菜品的完整度数据，可以为确定匹配“小王的店”的菜品完整度数据，例如，确定云端数据库为门店“小王的店”的配置的分析数据存储空间中存储的各剩余菜品的完整度数据。之后，对确定的完整度数据进行进一步分析处理，确定“小王的店”的每个剩余菜品的指示菜品受欢迎程度的菜品分析数据。如，确定“土豆丝”的受欢迎程度的评价指标为700，确定“红烧肉”的受欢迎程度的评价指标为520等。通过将“小王的店”中各菜品的受欢迎程度的评价指标进行输出展示，便于“小王的店”进行菜品优化，提升顾客的用餐体验。
再例如，当所述菜品分析数据的至少一个展示维度包括：地域维度、时间维度和天气维度时，首先根据地域维度的取值确定菜品的完整度数据对应的门店地理位置范围，并根据确定的门店地理位置范围确定匹配所述地理区域维度取值的候选门店；之后，进一步获取各候选门店的菜品完整度数据和采集信息的对应关系数据，作为待分析数据集合。然后，根据时间维度的取值确定完整度数据对应的采集信息中采集时间需要符合的时间范围，从所述待分析数据集合中过滤掉不符合所述时间范围的匹配菜品的完整度数据，取对应的采集信息中采集时间满足的时间范围的菜品完整度数据和采集信息的对应关系数据。接下来，根据天气维度的取值确定菜品的完整度数据对应的采集信息中采集天气符合的天气情况，从过滤后的待分析数据集合中，选择采集天气符合确定的天气情况的所述对应关系数据，作为最终待分析数据。例如，当地域维度的取值为“望京地区”、时间维度的取值可以为“2019年第三季度”、天气维度的取值可以为“雨天”时，确定望京地区的门店在2019年第三季度内且在雨天上传的图像确定的菜品完整度数据，并对确定的所述菜品完整度数据进行分析，得到每个菜品的受欢迎程度的评价指标。例如可以得到如下表1所示的菜品分析数据。
天气 菜品 受欢迎程度的评价指标
雨天 宫保鸡丁 897
雨天 辣子鸡 786
表1
本申请的一些实施例中，所述确定物品分析数据的至少一个展示维度及各所述展示维度的取值，包括：根据所述物品分析数据匹配的输出对象的数据查询权限，确定物品分析数据的至少一个展示维度及各所述展示维度的取值。例如，菜品分析系统为菜品分析数据需求对象配置不同的查询权限，并预先为不同的查询权限关联不同的展示维度或展示维度组合。例如，菜品分析系统为VIP对象配置最高级别查询权限，并设置最高级别查询权限关联的展示维度为：地域维度、时间维度和天气维度的组合；菜品分析系统为普通对象配置通用级别查询权限，并设置通用级别查询权限关联的展示维度为：门店维度。
本申请的一些实施例中，所述确定物品分析数据的至少一个展示维度及各所述展示维度的取值，包括：根据输入的查询条件，确定物品分析数据的至少一个展示维度及各所述展示维度的取值。例如，菜品分析系统可以为对象提供查询条件输入界面，供对象输入门店、地域范围、时间范围、天气情况等一种或多种查询条件取值，并通过所述数据界面获取所述对象输入的所述门店、地域范围、时间范围、天气情况等一种或多种查询条件取值，将获取的所述一种或多种查询条件取值确定为菜品分析数据的各展示维度的取值。
其中，每个物品的受欢迎程度的评价指标值，可以根据云端数据库中存储的该物品的满足展示维度条件的所有物品完整度数据的累加和得到。例如，对于菜品“宫保鸡丁”,望京地区的门店在2019年第三季度内且在雨天上传的剩余宫保鸡丁的图像确定的菜品完整度数据有1000条，分别为：60％、75％、50％.......将这1000条菜品完整度数据累加求和之后，可以将求得的和作为菜品“宫保鸡丁”的受欢迎程度的评价指标值。
具体到菜品数据分析应用场景中，不同菜品的点餐频率不同，餐后被剩余的概率也不同，点餐频率越高的菜品，餐后被剩余的概率也会越大，会导致按照上述方法计算得到的受欢迎程度的评价指标值越大。而由上述计算方法可以看出，受欢迎程度的评价指标值越大表示该菜品越不受欢迎，即对于点餐频率很高的菜品，计算得到的受欢迎程度的评价指标值可能错误地指示 菜品受欢迎程度。
为了使受欢迎程度的评价指标值准确地指示物品的受欢迎程度，本申请的一些实施例中，为每个物品设置了用于计算物品受欢迎程度评价指标的权值，相应的，所述对与所述至少一个展示维度的取值匹配的所述剩余物品的完整度数据进行数据分析处理，确定指示物品受欢迎程度的物品分析数据，包括：对于每个所述剩余物品的所述完整度数据，以所述剩余物品匹配的预设权值对所述物品的各完整度数据进行加权运算，确定所述剩余物品的指示物品受欢迎程度的评价指标值；根据每个所述剩余物品与指示物品受欢迎程度的所述评价指标值、所述至少一个展示维度的取值的对应关系，生成指示物品受欢迎程度的物品分析数据。其中，每个物品的所述权值根据业务需求确定，以物品为菜品的应用场景为例，对于“米饭”这类点餐频率较大的普通菜品，可以设置较低的权值，而对于“烤鸭”这类门店招牌菜品可以设置较高的权值。通过为不同菜品设置不同的权值，提升计算得到的评价指标值对菜品的受欢迎程度的指示明确度。
以将菜品“宫保鸡丁”的权值设置为0.8，而将菜品“尖椒土豆丝”的权值设置为0.3为例，假设对于菜品“宫保鸡丁”和“尖椒土豆丝”,门店X在2019年第三季度内且在雨天上传的剩余“宫保鸡丁”和“尖椒土豆丝”的菜品完整度数据有50条，完整度数据取值均为：50％，那么，在对上述门店X的“宫保鸡丁”的50条完整度数据进行加权运算之后，可以得到指示菜品“宫保鸡丁”的受欢迎程度的评价指标值为20，而在对上述门店X的“尖椒土豆丝”的50条完整度数据进行加权运算之后，可以得到指示菜品“尖椒土豆丝”的受欢迎程度的评价指标值为7.5。根据菜品“宫保鸡丁”和“尖椒土豆丝”的指示菜品受欢迎程度的评价指标值生成的生成菜品分析数据如表2所示。
门店 时间 天气 菜品 受欢迎程度的评价指标
门店X 2019年第三季度 雨天 宫保鸡丁 20
门店X 2019年第三季度 雨天 尖椒土豆丝 7.5
表2
由上述表2可以看出，对于不同菜品，虽然被剩余次数相同，每次剩余的菜量相同，但是由于不同的菜品设置了不同的权限，因此得到了不同的评价指标值。在浏览如表2中所示的菜品分析数据时，可以明确得出“宫保鸡丁”相较于“尖椒土豆丝”的不受欢迎程度更强。
本申请实施例公开的物品分析数据确定方法，通过获取包含剩余物品的图像；通过对所述图像进行图像处理，确定所述图像中包含的剩余物品以及所述剩余物品的完整度数据；通过对各所述剩余物品的完整度数据，以及包含各所述剩余物品的所述图像的采集信息进行数据分析处理，确定指示物品受欢迎程度的物品分析数据，有助于提升确定的物品分析数据的准确度。例如，本申请实施例公开的物品分析数据确定方法，通过采集餐后菜品的图像，并进行图像处理，确定各个菜品的剩余次数和每次的剩余菜量，然后，根据各个菜品的剩余次数和每次的剩余菜量确定指示菜品受欢迎程度的评价指标值，并基于各菜品的所述评价指标值生成菜品分析数据，根据位于就餐门店内的图像采集设备采集的餐后菜品图像数据计算出菜品的评价指标值，数据来源更客观，更易于采集大量数据，从而使得基于客观采集的大量数据计算得到的评价指标值更客观、更准确地反映菜品的受欢迎程度。再例如，在自助餐厅、智慧售卖等场景中，不同餐食的取餐量和不同商品的售卖量会有所不同，通过对剩余餐食或剩余商品进行图像采集和分析，可以实现餐食、商品等物品的受欢迎程度数据自助分析。
进一步的，通过基于多个展示维度进行菜品数据分析，有助于提升确定的物品分析数据的实用性。
实施例二
本申请实施例公开的一种物品分析数据确定方法，如图4所示，所述方法包括：步骤410至步骤440。
步骤410，获取包含剩余物品的图像。
获取包含剩余物品的图像的具体实施方式参见实施例一，本实施例不再赘述。
步骤420，通过对所述图像进行图像处理，确定所述图像中包含的剩余物品以及所述剩余物品的完整度数据。
通过对所述图像进行图像处理，确定所述图像中包含的剩余物品以及所述剩余物品的完整度数据的具体实施方式参见实施例一，本实施例不再赘述。
步骤430，通过对各所述剩余物品的完整度数据，以及包含各所述剩余物品的所述图像的采集信息进行数据分析处理，确定指示物品受欢迎程度的物品分析数据。
通过对各所述剩余物品的完整度数据，以及包含各所述剩余物品的所述 图像的采集信息进行数据分析处理，确定指示物品受欢迎程度的物品分析数据的具体实施方式参见实施例一，本实施例不再赘述。
本申请的实施例中，所述剩余物品的完整度数据用于物品剩余比例。包含各所述剩余物品的所述图像的采集信息包括所述图像的采集门店、采集时间、采集天气、采集地理位置等信息。通过对各所述剩余物品的完整度数据，以及各完整度数据关联的采集门店、采集时间、采集天气、采集地理位置等信息进行数据分析处理，则可以确定指示物品受欢迎程度的物品分析数据。例如，可以确定某个门店的菜品受欢迎程度的评价指标值。再例如，根据菜品受欢迎程度的评价指标值，生成菜品的不受欢迎程度排行榜。其中，所述评价指标值是根据相应菜品的若干条完整度数据计算得到的。
步骤440，按照确定的所述至少一个展示维度，展示所述物品分析数据。
本申请的一些实施例中，可以根据物品分析数据的展示需求确定数据的至少一个展示维度，则在进行物品数据分析处理时，可以按照确定的展示维度对各所述剩余物品的完整度数据，以及各完整度数据关联的采集门店、采集时间、采集天气、采集地理位置等信息进行数据分析处理。例如，当确定的展示维度包括门店和天气时，则可以基于确定的门店和天气两个维度的取值对物品完整度数据和采集天气进行分析处理，得到指定门店在指定天气情况下各物品的物品受欢迎程度的评价指标值。
本申请实施例所述的物品分析数据确定方法可以应用于如图5所示的菜品信息分析系统。如图5所示，所述菜品信息分析系统包括：图像采集设备510、收银终端520、云端服务器530和餐厅管理系统540。下面分别介绍各设备或系统的工作原理。
所述图像采集设备510位于门店内，用于采集门店顾客用餐后的剩余菜品的图像，例如，通过门店内的图像采集设备510，获取用户在所述门店内用餐后的桌台图像；或者，通过门店内的图像采集设备510，获取用户在所述门店内用餐后各剩余菜品的图像。所述图像采集设备510可以为：视频监控设备，或门店服务人员手动操作的数码相机、智能手机、PDA(个人数字助理)、具有拍照功能的点餐终端等设备。
所述收银终端520位于门店内，用于进行用餐结算；所述收银终端520还可以用于接收所述门店内的所述图像采集设备510上传的图像，并将所述图像经过去重、初步编辑之后，上传至所述云端服务器530。例如，所述收银终端520根据接收到的各图像的图像采集设备标识和上传时间，以及图像 相似度对所述图像进行去重处理，过滤掉重复上传的图像。再例如，所述收银终端520对接收到的所述图像添加门店标识或门店地理位置等采集地理位置信息。
所述云端服务器530还包括云端数据库，所述云端服务器530用于接收各个门店的所述收银终端520上传的包含剩余菜品的图像，并将接收到的图像存储在所述云端数据库中。
本申请的一些实施例中，当所述收银终端520上传的包含剩余菜品的图像为顾客在所述门店内用餐后的桌台图像时，所述云端服务器530还用于通过对所述桌台图像进行图像处理，确定所述桌台图像中包含的各剩余菜品的图像，并将各剩余菜品的图像存储在所述云端数据库中。通过对所述桌台图像进行图像处理，确定所述桌台图像中包含的各剩余菜品的图像的具体实施方式参见实施例一，本实施例中不再赘述。
所述云端服务器530还用于通过对所述剩余菜品的图像进行图像处理，确定所述图像中包含的剩余菜品以及所述剩余菜品的完整度数据；通过对各所述剩余菜品的完整度数据，以及包含各所述剩余菜品的所述图像的采集信息进行数据分析处理，确定指示菜品受欢迎程度的菜品分析数据。通过对所述剩余菜品的图像进行图像处理，确定所述图像中包含的剩余菜品以及所述剩余菜品的完整度数据的具体实施方式参见实施例一中通过对所述剩余物品的图像进行图像处理，确定所述图像中包含的剩余物品以及所述剩余物品的完整度数据的具体实施方式，本实施例中不再赘述。通过对各所述剩余菜品的完整度数据，以及包含各所述剩余菜品的所述图像的采集信息进行数据分析处理，确定指示菜品受欢迎程度的菜品分析数据的具体实施方式参见实施例一，本实施例中不再赘述。
所述餐厅管理系统540用于通过所述云端服务器530获取指示菜品受欢迎程度的菜品分析数据，并按照确定的至少一个展示维度，展示所述菜品分析数据。
本申请的一些实施例中，在通过对各所述剩余物品的完整度数据，以及包含各所述剩余菜品的所述图像的采集信息进行数据分析处理，确定指示物品受欢迎程度的物品分析数据之后，还包括：按照确定的所述至少一个展示维度，展示所述物品分析数据。例如，在基于确定的门店和天气两个维度的取值对菜品完整度数据和采集天气进行分析处理，得到指定门店在指定天气情况下各菜品的菜品受欢迎程度的评价指标值之后，可以按照门店和天气维 度，对满足所述门店维度的取值和所述天气维度的取值的组合的菜品受欢迎程度的评价指标值进行展示。
本申请实施例公开的物品分析数据确定方法，通过获取包含剩余物品的图像；通过对所述图像进行图像处理，确定所述图像中包含的剩余物品以及所述剩余物品的完整度数据；通过对各所述剩余物品的完整度数据，以及包含各所述剩余物品的所述图像的采集信息进行数据分析处理，确定指示物品受欢迎程度的物品分析数据，最后，按照确定的所述至少一个展示维度，展示所述物品分析数据，有助于更客观、更准确地展示反映物品的受欢迎程度的数据分析结果，使得各物品供应门店可以实时、准确地了解物品的受欢迎程度，进一步可以为进行物品改进和提升用户就餐体验等提供数据支撑。
实施例三
本申请实施例公开的一种物品分析数据确定装置，如图6所示，所述装置包括：
剩余物品图像获取模块610，用于获取包含剩余物品的图像；
物品完整度数据确定模块620，用于通过对所述图像进行图像处理，确定所述图像中包含的剩余物品以及所述剩余物品的完整度数据；
物品分析数据确定模块630，用于通过对各所述剩余物品的完整度数据，以及包含各所述剩余物品的所述图像的采集信息进行数据分析处理，确定指示物品受欢迎程度的物品分析数据。本申请的一些实施例中，所述通过对各所述剩余物品的完整度数据，以及包含各所述剩余物品的所述图像的采集信息进行数据分析处理，确定指示物品受欢迎程度的物品分析数据，包括：
确定物品分析数据的至少一个展示维度及各所述展示维度的取值；
根据包含所述剩余物品的所述图像的采集信息与所述至少一个展示维度的取值的匹配关系，确定与所述至少一个展示维度的取值匹配的所述剩余物品的完整度数据；
对与所述至少一个展示维度的取值匹配的所述剩余物品的完整度数据进行数据分析处理，确定指示物品受欢迎程度的物品分析数据。
本申请的一些实施例中，所述对与所述至少一个展示维度的取值匹配的所述剩余物品的完整度数据进行数据分析处理，确定指示物品受欢迎程度的物品分析数据，包括：
对于每个所述剩余物品的所述完整度数据，以所述剩余物品匹配的预设权值对所述物品的各完整度数据进行加权运算，确定所述剩余物品的指示物 品受欢迎程度的评价指标值；
根据每个所述剩余物品与指示物品受欢迎程度的所述评价指标值、所述至少一个展示维度的取值的对应关系，生成指示物品受欢迎程度的物品分析数据。
本申请的一些实施例中，所述至少一个展示维度包括：门店维度、时间维度、地域维度、天气维度中的任意一个或多个维度。
本申请的一些实施例中，如图7所示，所述装置还包括：
展示模块640，用于按照确定的所述至少一个展示维度，展示所述物品分析数据。
本申请的一些实施例中，所述物品为菜品，所述剩余物品图像获取模块610进一步用于：
通过门店内的图像采集设备，获取用户在所述门店内用餐后的桌台图像；
通过对所述桌台图像进行图像处理，确定所述桌台图像中包含的各剩余菜品的图像。
本申请的一些实施例中，所述通过对所述桌台图像进行图像处理，确定所述桌台图像中包含的各剩余菜品的图像，包括：
对所述桌台图像进行去噪和二值化处理，确定桌台内容图像；
确定所述桌台内容图像中的餐具中心坐标，并根据所述餐具中心坐标和餐具轮廓确定所述桌台内容图像中包括的至少一个餐具图像；
通过对所述餐具图像进行空盘识别，过滤掉包含剩余菜品量符合预设条件的餐具图像，将剩余的所述餐具图像确定为所述桌台图像中包含的各剩余菜品的图像。
本申请的一些实施例中，所述物品为菜品，所述剩余物品图像获取模块610进一步用于：
通过门店内的图像采集设备，获取用户在所述门店内用餐后各剩余菜品的图像。
本申请的一些实施例中，所述物品为货品，所述剩余物品图像获取模块610进一步用于：
通过门店内的图像采集设备，获取摆放货品的货架图像；
通过对所述货架图像进行图像处理，确定所述货架图像中包含的各剩余货品的图像。
本申请的一些实施例中，所述通过对所述货架图像进行图像处理，确定所述货架图像中包含的各剩余货品的图像，包括：
根据货品摆放位置对所述货架图像进行分割，确定单一货品图像；
通过对所述单一货品图像进行识别，过滤掉包含剩余货品量符合预设条件的所述单一货品图像，将剩余的所述单一货品图像确定为所述货架图像中包含的各剩余货品的图像。
本申请的一些实施例中，所述物品完整度数据确定模块620进一步用于：
通过对所述图像进行物品识别，确定所述图像中包含的物品作为剩余物品；
基于所述剩余物品的所述图像进行几何测量，确定所述剩余物品的剩余体积；
根据预先获取的所述剩余物品的先验体积和所述剩余体积的比较结果，确定所述剩余物品的用于指示物品剩余比例的完整度数据。本申请实施例公开的物品分析数据确定装置，用于实现本申请实施例一或实施例二中所述的物品分析数据确定方法，装置的各模块的具体实施方式不再赘述，可参见方法实施例相应步骤的具体实施方式。
本申请实施例公开的一种物品分析数据确定装置，通过获取包含剩余物品的图像；通过对所述图像进行图像处理，确定所述图像中包含的剩余物品以及所述剩余物品的完整度数据；通过对各所述剩余物品的完整度数据，以及包含各所述剩余物品的所述图像的采集信息进行数据分析处理，确定指示物品受欢迎程度的物品分析数据，有助于提升确定的物品分析数据的准确度。例如，本申请实施例公开的物品分析数据确定装置，通过采集餐后菜品的图像，并进行图像处理，确定各个菜品的剩余次数和每次的剩余菜量，然后，根据各个菜品的剩余次数和每次的剩余菜量确定指示菜品受欢迎程度的评价指标值，并基于各菜品的所述评价指标值生成菜品分析数据，根据位于就餐门店内的图像采集设备采集的餐后菜品图像数据计算出菜品的评价指标值，数据来源更客观，更易于采集大量数据，从而使得基于客观采集的大量数据计算得到的评价指标值更客观、更准确地反映菜品的受欢迎程度。再例如，在自助餐厅、智慧售卖等场景中，不同餐食的取餐量和不同商品的售卖量会有所不同，通过对剩余餐食或剩余商品进行图像采集和分析，可以实现餐食、商品等物品的受欢迎程度数据自助分析。
进一步的，通过基于多个展示维度进行菜品数据分析，有助于提升确定 的物品分析数据的实用性。
本说明书中的各个实施例均采用递进的方式描述，每个实施例重点说明的都是与其他实施例的不同之处，各个实施例之间相同相似的部分互相参见即可。对于装置实施例而言，由于其与方法实施例基本相似，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。
以上对本申请提供的一种物品分析数据确定方法及装置进行了详细介绍，本文中应用了具体个例对本申请的原理及实施方式进行了阐述，以上实施例的说明只是用于帮助理解本申请的方法及其一种核心思想；同时，对于本领域的一般技术人员，依据本申请的思想，在具体实施方式及应用范围上均会有改变之处，综上所述，本说明书内容不应理解为对本申请的限制。
以上所描述的装置实施例仅仅是示意性的，其中所述作为分离部件说明的单元可以是或者也可以不是物理上分开的，作为单元显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部模块来实现本实施例方案的目的。本领域普通技术人员在不付出创造性的劳动的情况下，即可以理解并实施。
本申请的各个部件实施例可以以硬件实现，或者以在一个或者多个处理器上运行的软件模块实现，或者以它们的组合实现。本领域的技术人员应当理解，可以在实践中使用微处理器或者数字信号处理器(DSP)来实现根据本申请实施例的电子设备中的一些或者全部部件的一些或者全部功能。本申请还可以实现为用于执行这里所描述的方法的一部分或者全部的设备或者装置程序(例如，计算机程序和计算机程序产品)。这样的实现本申请的程序可以存储在计算机可读介质上，或者可以具有一个或者多个信号的形式。这样的信号可以从因特网网站上下载得到，或者在载体信号上提供，或者以任何其他形式提供。
例如，图8示出了可以实现根据本申请的方法的电子设备。所述电子设备可以为PC机、移动终端、个人数字助理、平板电脑等。该电子设备传统上包括处理器820和存储器810及存储在所述存储器810上并可在处理器820上运行的程序代码830，所述处理器820执行所述程序代码830时实现上述实施例中所述的方法。所述存储器810可以为计算机程序产品或者计算机可读介质。存储器810可以是诸如闪存、EEPROM(电可擦除可编程只读存储器)、EPROM、硬盘或者ROM之类的电子存储器。存储器810具有用于执行 上述方法中的任何方法步骤的计算机程序的程序代码830的存储空间8101。例如，用于程序代码830的存储空间8101可以包括分别用于实现上面的方法中的各种步骤的各个计算机程序。所述程序代码830为计算机可读代码。这些计算机程序可以从一个或者多个计算机程序产品中读出或者写入到这一个或者多个计算机程序产品中。这些计算机程序产品包括诸如硬盘，紧致盘(CD)、存储卡或者软盘之类的程序代码载体。所述计算机程序包括计算机可读代码，当所述计算机可读代码在电子设备上运行时，导致所述电子设备执行根据上述实施例的方法。
本申请实施例还公开了一种计算机可读存储介质，其上存储有计算机程序，该程序被处理器执行时实现如本申请实施例一或实施例二所述的物品分析数据确定方法的步骤。
这样的计算机程序产品可以为计算机可读存储介质，该计算机可读存储介质可以具有与图8所示的电子设备中的存储器810类似布置的存储段、存储空间等。程序代码可以例如以适当形式进行压缩存储在所述计算机可读存储介质中。所述计算机可读存储介质通常为如参考图9所述的便携式或者固定存储单元。通常，存储单元包括计算机可读代码830’，所述计算机可读代码830’为由处理器读取的代码，这些代码被处理器执行时，实现上面所描述的方法中的各个步骤。
本文中所称的“一个实施例”、“实施例”或者“一个或者多个实施例”意味着，结合实施例描述的特定特征、结构或者特性包括在本申请的至少一个实施例中。此外，请注意，这里“在一个实施例中”的词语例子不一定全指同一个实施例。
在此处所提供的说明书中，说明了大量具体细节。然而，能够理解，本申请的实施例可以在没有这些具体细节的情况下被实践。在一些实例中，并未详细示出公知的方法、结构和技术，以便不模糊对本说明书的理解。
在权利要求中，不应将位于括号之间的任何参考符号构造成对权利要求的限制。单词“包含”不排除存在未列在权利要求中的元件或步骤。位于元件之前的单词“一”或“一个”不排除存在多个这样的元件。本申请可以借助于包括有若干不同元件的硬件以及借助于适当编程的计算机来实现。在列举了若干装置的单元权利要求中，这些装置中的若干个可以是通过同一个硬件项来具体体现。单词第一、第二、以及第三等的使用不表示任何顺序。可将这些单词解释为名称。
最后应说明的是：以上实施例仅用以说明本申请的技术方案，而非对其限制；尽管参照前述实施例对本申请进行了详细的说明，本领域的普通技术人员应当理解：其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案的本质脱离本申请各实施例技术方案的精神和范围。

Claims (15)

  1. 一种物品分析数据确定方法，包括：
    获取包含剩余物品的图像；
    通过对所述图像进行图像处理，确定所述图像中包含的剩余物品以及所述剩余物品的完整度数据；
    通过对各所述剩余物品的完整度数据，以及包含各所述剩余物品的所述图像的采集信息进行数据分析处理，确定指示物品受欢迎程度的物品分析数据。
  2. 根据权利要求1所述的方法，所述通过对各所述剩余物品的完整度数据，以及包含各所述剩余物品的所述图像的采集信息进行数据分析处理，确定指示物品受欢迎程度的物品分析数据的步骤，包括：
    确定物品分析数据的至少一个展示维度及各所述展示维度的取值；
    根据包含所述剩余物品的所述图像的采集信息与所述至少一个展示维度的取值的匹配关系，确定与所述至少一个展示维度的取值匹配的所述剩余物品的完整度数据；
    对与所述至少一个展示维度的取值匹配的所述剩余物品的完整度数据进行数据分析处理，确定指示物品受欢迎程度的物品分析数据。
  3. 根据权利要求2所述的方法，所述对与所述至少一个展示维度的取值匹配的所述剩余物品的完整度数据进行数据分析处理，确定指示物品受欢迎程度的物品分析数据的步骤，包括：
    对于每个所述剩余物品的所述完整度数据，以所述剩余物品匹配的预设权值对所述物品的各完整度数据进行加权运算，确定所述剩余物品的指示物品受欢迎程度的评价指标值；
    根据每个所述剩余物品与指示物品受欢迎程度的所述评价指标值、所述至少一个展示维度的取值的对应关系，生成指示物品受欢迎程度的物品分析数据。
  4. 根据权利要求2所述的方法，所述至少一个展示维度包括：门店维度、时间维度、地域维度、天气维度中的任意一个或多个维度。
  5. 根据权利要求1至4任一项所述的方法，所述通过对各所述剩余物品的完整度数据，以及包含各所述剩余物品的所述图像的采集信息进行数据分析处理，确定指示物品受欢迎程度的物品分析数据的步骤之后，还包括：
    按照确定的所述至少一个展示维度，展示所述物品分析数据。
  6. 根据权利要求1所述的方法，所述物品为菜品，所述获取包含剩余物品的图像的步骤，包括：
    通过门店内的图像采集设备，获取用户在所述门店内用餐后的桌台图像；
    通过对所述桌台图像进行图像处理，确定所述桌台图像中包含的各剩余菜品的图像。
  7. 根据权利要求6所述的方法，所述通过对所述桌台图像进行图像处理，确定所述桌台图像中包含的各剩余菜品的图像的步骤，包括：
    对所述桌台图像进行去噪和二值化处理，确定桌台内容图像；
    确定所述桌台内容图像中的餐具中心坐标，并根据所述餐具中心坐标和餐具轮廓确定所述桌台内容图像中包括的至少一个餐具图像；
    通过对所述餐具图像进行空盘识别，过滤掉包含剩余菜品量符合预设条件的餐具图像，将剩余的所述餐具图像确定为所述桌台图像中包含的各剩余菜品的图像。
  8. 根据权利要求1所述的方法，所述物品为菜品，所述获取包含剩余物品的图像的步骤，包括：
    通过门店内的图像采集设备，获取用户在所述门店内用餐后各剩余菜品的图像。
  9. 根据权利要求1所述的方法，所述物品为货品，所述获取包含剩余物品的图像的步骤，包括：
    通过门店内的图像采集设备，获取摆放货品的货架图像；
    通过对所述货架图像进行图像处理，确定所述货架图像中包含的各剩余货品的图像。
  10. 根据权利要求9所述的方法，所述通过对所述货架图像进行图像处理，确定所述货架图像中包含的各剩余货品的图像的步骤，包括：
    对根据货品摆放位置对所述货架图像进行分割，确定单一货品图像；
    通过对所述单一货品图像进行识别，过滤掉包含剩余货品量符合预设条件的所述单一货品图像，将剩余的所述单一货品图像确定为所述货架图像中包含的各剩余货品的图像。
  11. 根据权利要求1至4任一项所述的方法，所述通过对所述图像进行图像处理，确定所述图像中包含的剩余物品以及所述剩余物品的完整度数据 的步骤，包括：
    通过对所述图像进行物品识别，确定所述图像中包含的物品作为剩余物品；
    基于所述剩余物品的所述图像进行几何测量，确定所述剩余物品的剩余体积；
    根据预先获取的所述剩余物品的先验体积和所述剩余体积的比较结果，确定所述剩余物品的用于指示物品剩余比例的完整度数据。
  12. 一种物品分析数据确定装置，包括：
    剩余物品图像获取模块，用于获取包含剩余物品的图像；
    物品完整度数据确定模块，用于通过对所述图像进行图像处理，确定所述图像中包含的剩余物品以及所述剩余物品的完整度数据；
    物品分析数据确定模块，用于通过对各所述剩余物品的完整度数据，以及包含各所述剩余物品的所述图像的采集信息进行数据分析处理，确定指示物品受欢迎程度的物品分析数据。
  13. 一种电子设备，包括存储器、处理器及存储在所述存储器上并可在处理器上运行的程序代码，所述处理器执行所述程序代码时实现权利要求1至11任意一项所述的物品分析数据确定方法。
  14. 一种计算机可读存储介质，其上存储有程序代码，该程序代码被处理器执行时实现权利要求1至11任意一项所述的物品分析数据确定方法的步骤。
  15. 一种计算机程序，包括计算机可读代码，当所述计算机可读代码在电子设备上运行时，导致所述电子设备执行根据权利要求1至11中的任意一项所述的物品分析数据确定方法。
PCT/CN2020/110036 2020-02-20 2020-08-19 确定物品分析数据 WO2021164222A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010105304.XA CN111415328B (zh) 2020-02-20 2020-02-20 物品分析数据确定方法、装置、电子设备
CN202010105304.X 2020-02-20

Publications (1)

Publication Number Publication Date
WO2021164222A1 true WO2021164222A1 (zh) 2021-08-26

Family

ID=71492693

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/110036 WO2021164222A1 (zh) 2020-02-20 2020-08-19 确定物品分析数据

Country Status (2)

Country Link
CN (1) CN111415328B (zh)
WO (1) WO2021164222A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111415328B (zh) * 2020-02-20 2023-07-11 北京三快在线科技有限公司 物品分析数据确定方法、装置、电子设备

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9818128B2 (en) * 2008-05-30 2017-11-14 Red Hat, Inc. Encouraging content distribution without restrictions
CN109508401A (zh) * 2018-11-23 2019-03-22 维沃移动通信有限公司 一种提示方法及移动终端
CN109840810A (zh) * 2019-03-19 2019-06-04 深圳创维－Rgb电子有限公司 数据分析推送方法、装置、后台服务器及可读存储介质
CN110555750A (zh) * 2019-07-29 2019-12-10 北京三快在线科技有限公司 用餐人数预测方法、装置、电子设备及可读存储介质
CN111415328A (zh) * 2020-02-20 2020-07-14 北京三快在线科技有限公司 物品分析数据确定方法、装置、电子设备

Family Cites Families (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CA2219172C (en) * 1997-10-24 2008-03-18 Trevor Kern Method and apparatus for order verification
JP2003212316A (ja) * 2002-01-28 2003-07-30 Matsushita Electric Works Ltd 物品棚割り付け支援方法、物品棚割り付け支援装置及び記録媒体
US20070143126A1 (en) * 2005-12-15 2007-06-21 Anjan Ghose Method and system to determine a desirable diet
US8401911B1 (en) * 2011-03-22 2013-03-19 Google Inc. Display of popular, in-stock products of a merchant
CN103024825B (zh) * 2011-09-26 2015-07-01 中国移动通信集团公司 终端多应用间网络资源分配方法及其设备
US20150100509A1 (en) * 2013-10-09 2015-04-09 Salesforce.Com, Inc. Systems and methods for enhancing user engagement
CN104732413B (zh) * 2013-12-20 2017-11-21 中国科学院声学研究所 一种智能个性化视频广告推送方法及系统
CN105512331B (zh) * 2015-12-28 2019-03-26 海信集团有限公司 一种视频推荐方法及装置
CN105868267B (zh) * 2016-03-04 2019-04-12 江苏工程职业技术学院 一种移动社交网络用户兴趣的建模方法
CN107657476A (zh) * 2017-08-29 2018-02-02 口碑(上海)信息技术有限公司 店铺的评价方法及装置、店铺推荐方法及装置
CN107992583B (zh) * 2017-12-07 2020-09-04 合肥美的智能科技有限公司 信息推送方法及信息推送装置、设备和存储介质
CN108280124B (zh) * 2017-12-11 2020-12-04 北京三快在线科技有限公司 产品分类方法及装置，排行榜生成方法及装置，电子设备
CN109934427B (zh) * 2017-12-15 2022-03-04 北京京东尚科信息技术有限公司 生成物品分配方案的方法和装置
CN110211000A (zh) * 2018-02-28 2019-09-06 阿里巴巴集团控股有限公司 桌位状态信息处理方法、装置及系统
CN110008829A (zh) * 2019-02-21 2019-07-12 秒针信息技术有限公司 食品的调整方法及装置
CN110245274A (zh) * 2019-04-23 2019-09-17 五八有限公司 一种标签热度计算方法、装置、电子设备和存储介质
CN110602532A (zh) * 2019-09-24 2019-12-20 腾讯科技（深圳）有限公司 实体物品推荐方法、装置、服务器及存储介质
CN110727754B (zh) * 2019-10-11 2023-06-02 北京旷视机器人技术有限公司 货架热力图的生成方法、装置及电子设备

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9818128B2 (en) * 2008-05-30 2017-11-14 Red Hat, Inc. Encouraging content distribution without restrictions
CN109508401A (zh) * 2018-11-23 2019-03-22 维沃移动通信有限公司 一种提示方法及移动终端
CN109840810A (zh) * 2019-03-19 2019-06-04 深圳创维－Rgb电子有限公司 数据分析推送方法、装置、后台服务器及可读存储介质
CN110555750A (zh) * 2019-07-29 2019-12-10 北京三快在线科技有限公司 用餐人数预测方法、装置、电子设备及可读存储介质
CN111415328A (zh) * 2020-02-20 2020-07-14 北京三快在线科技有限公司 物品分析数据确定方法、装置、电子设备

Also Published As

Publication number Publication date
CN111415328A (zh) 2020-07-14
CN111415328B (zh) 2023-07-11

Similar Documents

Publication Publication Date Title
CN107332910B (zh) 信息推送方法和装置
WO2019056661A1 (zh) 一种搜索词推送方法、装置及终端
CN111984837B (zh) 商品数据的处理方法、装置及设备
CN104050187B (zh) 搜索结果展示方法及系统
CN109635198B (zh) 在商品展示平台上呈现用户搜索结果的方法、装置、介质及电子设备
CN103858142A (zh) 店铺信息提供系统
CN109214955B (zh) 餐品套餐的生成方法及装置
KR101353996B1 (ko) 온라인 쇼핑을 위한 상품관리 시스템 및 상품관리 방법
TW201501059A (zh) 訊息推薦方法及裝置
WO2017092602A1 (zh) 一种信息投放用户的筛选方法和服务器
CN105468630A (zh) 提供用户信息的方法及装置
CN107067293A (zh) 商家分类方法、装置及电子设备
JP2019508756A (ja) 電子配信プラットフォーム上で提示対象を選択および推奨するための方法および装置
WO2020215952A1 (zh) 物品识别方法和系统
CN109409964B (zh) 优质品牌的识别方法及装置
US20140052585A1 (en) Information processing system, information processing method, program, and information recording medium
US20160086200A1 (en) Product code analysis system and product code analysis program
CN110750697B (zh) 商户分类方法、装置、设备及存储介质
WO2015124024A1 (zh) 一种提升信息的曝光率的方法和装置、确定搜索词的价值的方法和装置
CN107220334A (zh) 商户名称的相似度计算方法、装置及设备
CN110942293A (zh) 物品信息的处理方法、设备、存储介质及系统
WO2021164222A1 (zh) 确定物品分析数据
CN105469291A (zh) 用户信息提供方法及装置
CN111159163A (zh) 商品信息数据库生成方法、商品搜索方法及相关装置
CN112132391A (zh) 一种商品评价采集方法、装置、服务器及存储介质

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: 20920022

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: 20920022

Country of ref document: EP

Kind code of ref document: A1