Shelf Cutting Support Device, Shelf Cutting Support Method, and Program

Abstract

To provide a shelf allocation assistance device, a shelf allocation assistance method, and a program that generate recommended shelf allocation indicating a commodity exhibition state including a state in which a specific commodity is exhibited at an effective position.SOLUTION: In a shelf allocation assistance device, a generation unit functions as recognition means for recognizing a commodity from a captured image derived by capturing an image of a commodity shelf, position recognition means for recognizing from the captured image a position of the commodity shelf where no commodity is exhibited, identification means for identifying a commodity to be exhibited at the position where no commodity is exhibited, and generation means for generating a plurality of shelf allocation candidates including specific commodities consisting of a commodity recognized by the recognition means and a commodity identified by the identification means, and a prediction unit functions as prediction means for predicting sales of the plurality of shelf allocation candidates on the basis of relationship information that represents a relationship between positional relations of commodities on commodity shelves and the sales of commodities.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a shelf division support device, a shelf division support method, and a program.

Background Art

[0002] In convenience stores, supermarkets, and other retail stores, since the display position of products greatly affects sales, the display position of products on the product shelves is frequently changed. This change in the display position of products may be carried out using information such as sales prediction based on the display position of products.

[0003] Patent Document 1 describes a method for performing shelf-by-shelf sales prediction based on sales prediction information for each product based on the actual sales results of the products and sales information for each shelf level of each shelf in the store.

[0004] Patent Document 2 describes setting a correspondence condition for displaying products with good sales at positions with good sales using the sales ranking for each product and the sales ranking for each position on the product shelf, and displaying and outputting the state in which the products are displayed according to the set correspondence condition.

[0005] In addition, a method for simulating the product display state using the products and the PI (Purchase Index) values of the products is described in, for example, Patent Document 3.

Prior Art Documents

Patent Documents

[0006]

Patent Document 1

Patent Document 2

Patent Document 3

Summary of the Invention

Problems to be Solved by the Invention

[0007] The sales of a product may vary depending on the position of the shelf level where the product is displayed. Therefore, even if a specific product (for example, a product that the seller desires to sell) is displayed on a shelf level with good sales, the sales of this specific product do not necessarily improve.

[0008] The technologies of Patent Documents 1 to 3 described above do not consider at all which position is effective for displaying this specific product.

[0009] The present invention has been made in view of the above problems, and an object thereof is to provide a technology for generating a recommended shelf layout indicating a product display state including a state in which a specific product is displayed at a more effective position.

Means for Solving the Problems

[0010] A shelf layout support device according to an aspect of the present invention includes: a recognition unit that recognizes products from a captured image of a product shelf; a position recognition unit that recognizes positions on the product shelf where no products are displayed from the captured image; a specification unit that specifies products to be displayed at positions where no products are displayed; a generation unit that generates a plurality of shelf layout candidates including specific products including the products recognized by the recognition unit and the products specified by the specification unit; and a prediction unit that predicts the sales of the plurality of shelf layout candidates based on relationship information representing the relationship between the positional relationship of products on the product shelf and the sales of the products.

[0011] Further, a shelf layout support method according to an aspect of the present invention includes a computer recognizing products from a captured image of a product shelf, a position recognition unit that recognizes positions on the product shelf where no products are displayed from the captured image, specifying products to be displayed at positions where no products are displayed, generating a plurality of shelf layout candidates including specific products including the products recognized by the recognition unit and the products specified by the specification unit, and predicting the sales of the plurality of shelf layout candidates based on relationship information representing the relationship between the positional relationship of products on the product shelf and the sales of the products.

[0012] Also, a program according to an aspect of the present invention causes a computer to execute a process of recognizing a product from a captured image of a product shelf, a position recognition means for recognizing a position on the product shelf where no product is displayed from the captured image, a process of specifying a product to be displayed at a position where no product is displayed, a process of generating a plurality of shelf division candidates including a specific product composed of the product recognized by the recognition means and the product specified by the specifying means, and a process of predicting the sales of the specific product in the plurality of shelf division candidates based on relationship information representing the relationship between the positional relationship of the products on the product shelf and the sales of the products.

[0013] Note that a computer program for realizing the above device, system or method by a computer, and a computer-readable non-transitory recording medium storing the computer program are also included in the scope of the present invention.

Advantages of the Invention

[0014] According to the present invention, it is possible to generate a recommended shelf division indicating the display state of products, including a state in which specific products are displayed at more effective positions.

Brief Description of the Drawings

[0015]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Embodiments for Carrying Out the Invention

[0016] <First Embodiment> The first embodiment of the present invention will be described with reference to the drawings. In this embodiment, the basic configuration for solving the problems of the present invention will be described. FIG. 1 is a functional block diagram showing an example of the functional configuration of the shelf division support device 10 according to this embodiment. As shown in FIG. 1, the shelf division support device 10 according to this embodiment includes a generation unit 11, a prediction unit 12, and a selection unit 13.

[0017] The generation unit 11 generates a plurality of shelf division candidates representing the display states of a plurality of products on the product shelves including a specific product. The specific product is, for example, a product that a seller desires to sell via an input unit (not shown), a product with a large stock quantity, a product with an expiration date such as a best-before date, a consumption date, or a use date approaching, etc. Also, the specific product may be all in-stock products. The generation unit 11 outputs the generated plurality of shelf division candidates to the prediction unit 12.

[0018] The prediction unit 12 receives a plurality of shelf division candidates from the generation unit 11. Then, the generation unit 11 predicts the sales of a specific product among the plurality of shelf division candidates based on the relationship between the positional relationship between the products displayed on the product shelf and the sales of the products. The relationship between the positional relationship between the products and the sales of the products is expressed, for example, by the weight on each shelf level for each product name. In addition, the relationship between the positional relationship between the products and the sales of the products is expressed, among other things, by the weight on each shelf level for each product type, the weight on each shelf level for each adjacent product name, or the weight on each shelf level for each adjacent product type. The prediction unit 12 outputs the prediction result to the selection unit 13.

[0019] The selection unit 13 receives the prediction result from the prediction unit 12. Then, the selection unit 13 selects a shelf division candidate based on the received prediction result. The selection unit 13 selects, for example, the shelf division candidate with the largest predicted sales (also referred to as the predicted sales) among the plurality of shelf division candidates. Here, the magnitude of the predicted sales may be the number of predicted sales or the amount of predicted sales.

[0020] For example, assume that the predicted sales of product A in a certain shelf division candidate (referred to as shelf division candidate AA) is 5, and the predicted sales of product B is 4. Also, assume that the predicted sales of product A in another shelf division candidate (referred to as shelf division candidate BB) is 7, and the predicted sales of product B is 1. At this time, the selection unit 13 may calculate the total predicted sales of the products for each shelf division candidate and select shelf division candidate AA with the larger total. Also, the selection unit 13 may select shelf division candidate BB, which is the shelf division candidate for which the largest predicted sales of 7 were predicted.

[0021] As described above, the shelf division support device 10 according to the present embodiment predicts the sales of a specific product based on the relationship between the positional relationship between the products and the sales of the products, and selects a shelf division candidate based on the prediction result.

[0022] Therefore, it can be said that the display position of a specific product and the relationship between the display positions of the specific product and other products included in the selected shelf division candidate have an effect on sales. Therefore, according to the shelf division support device 10 according to the present embodiment, it is possible to generate a recommended shelf division indicating the display state of products, including a state in which a specific product is displayed at a more effective position. In addition, since the seller can perform the shelf division operation based on the recommended shelf division based on the relationship between the product position and sales, the shelf division support device 10 according to the present embodiment can efficiently support the shelf division operation.

[0023] <Second Embodiment> Next, a second embodiment of the present invention, which is based on the above-described first embodiment, will be described with reference to the drawings. FIG. 2 is a diagram showing an example of the overall configuration of the shelf division support system 1 according to the present embodiment. The shelf division support system 1 shown in FIG. 2 includes a shelf division support device 100, an inventory management device 200, and a data analysis device 300. The shelf division support device 100 includes the configuration of the shelf division support device 10 described above. It goes without saying that the shelf division support system 1 shown in FIG. 2 shows a configuration peculiar to the present invention, and the shelf division support system 1 shown in FIG. 2 may have members not shown in FIG. 2.

[0024] The shelf division support device 100, the inventory management device 200, and the data analysis device 300 are communicably connected to each other via a network 400. The communication means between the above devices may be either wired or wireless communication, and may be communication via any of a mobile communication network, a public line network, a LAN (Local Area Network), or a WAN (Wide Area Network). Thus, various communication methods are conceivable for the communication between the above devices, but since it does not relate to the essence of the present embodiment, detailed description is omitted.

[0025] The inventory management device 200 manages the inventory of products in the store. The inventory management device 200 receives sales data indicating the sales of each product name from one or more POS (Point Of Sales) terminals 21, and manages the inventory based on the received sales data and the order data. Note that the order data may be transmitted from an order device (not shown).

[0026] Note that in FIG. 2, a configuration in which the inventory management device 200 is installed in each store is shown. However, the inventory management device 200 may be a server provided at a location separate from the store. In this case, the inventory management device 200 manages the inventory of a plurality of stores for each store. Further, the inventory management device 200 may be integrated with the POS terminal 21. Here, the sales data is assumed to be general POS data such as the sales amount or the number of sales of a certain product, but the present embodiment is not limited to this. Also, the information regarding the inventory managed by the inventory management device 200 is assumed to include the product name, the number, the product type, etc., but the present embodiment is not limited to this. For example, the information regarding the inventory may include the expiration date (use-by date, expiration date, or best-before date) of the product. The inventory management device 200 transmits the information regarding the inventory to be managed to the shelf allocation support device 100.

[0027] The data analysis device 300 is a device that analyzes the relationship between the positional relationship between products and the sales of the products. The method for analyzing the relationship between the positional relationship between products and the sales by the data analysis device 300 is analyzed, for example, based on the photographed image and the sales data. Hereinafter, the method for analyzing the relationship between the positional relationship between products and the sales by the data analysis device 300 will be described. However, the present embodiment may use the results analyzed by a method other than the analysis method described below.

[0028] The data analysis device 300 in this embodiment uses a captured image of a product shelf as learning data, and recognizes products included in the captured image. Then, the data analysis device 300 specifies the arrangement position of the recognized products on the product shelf. Further, the data analysis device 300 receives sales data indicating sales for each product name from, for example, the POS terminal 21. The sales data received by the data analysis device 300 from the POS terminal 21 may be the same as the sales data received by the inventory management device 200, or may be sales data for a different date from the sales data received by the inventory management device 200. The sales data received by the data analysis device 300 may be any data that can be used to analyze the relationship between the positional relationship between products and the sales of products.

[0029] Based on the specified arrangement position of the product and the sales data of the product, the data analysis device 300 analyzes the relationship between the positional relationship between products and the sales of products. Hereinafter, the analysis result analyzed by the data analysis device 300 is also referred to as relationship information.

[0030] For example, the data analysis device 300 uses information indicating the recognized product (for example, product name) and the arrangement position of the recognized product on the product shelf to convert it into a feature vector f (product name, shelf level) with the product name and the shelf level of the product shelf as variables. For example, it is assumed that each component of the feature vector f is represented by the number of arrangements of a certain product on a certain level. That is, "f(product A, 1)=1" indicates that a product with the product name "product A" is arranged one on the first level of the product shelf. Hereinafter, for example, f(product A, 1) is denoted as f A1 as described.

[0031] The data analysis device 300 performs data analysis for each product name using this feature vector f. Hereinafter, the data analysis for the product with the product name "product A" will be described. The feature quantity vector of product A is f A as described. Also, it is assumed that the analysis data used for data analysis is the following (1) and (2). (1) The feature vector f for product A in a certain store A =(fA1 , f A2 , f A3 , f A4 , f A5 ) = (1, 0, 0, 2, 2), and the sales amount (y A ) = 1000 for product A in the same store, including the dataset. (2) The feature vector f for product A in other stores A = (f A1 , f A2 , f A3 , f A4 , f A5 ) = (0, 3, 1, 4, 0), and the sales amount (y A ) = 3000 for product A in the same store, including the dataset.

[0032] Note that the analysis data is not limited to two sets and may be multiple sets. Also, in this embodiment, although the explanation is given by taking the example of using two sets of data from different stores, it may also be a dataset generated from the sales data of different dates and times in one store.

[0033] The feature vector f for product A A has components corresponding to the number of shelves of the product shelf. The value of each component indicates the number of product arrangements as described above. From the above (1), it can be seen that on the product shelf arranged in a certain store, there is 1 product A arranged on the first shelf of the product shelf, none on the second and third shelves, and 2 each on the fourth and fifth shelves.

[0034] The data analysis device 300 uses this analysis data to calculate θ that satisfies the following formula (1). A

[0035]

Equation

[0036] In formula (1), i represents the shelf level (i = 1, 2, 3, 4, 5).

[0037] ​The data analysis device 300 performs data analysis using analysis data, and the result of the data analysis (analysis result) is θ A For example, θ A = (f A1 、f A2 、f A3 、f A4 、f A5 ) = (500, 800, 100, 400, 100). The components θ A of θ Ai , which is the analysis result in this embodiment, indicate the weights of the respective shelves of the product shelf for Product A. The component with a larger weight indicates the shelf where the sales of Product A are higher. Therefore, in the example of the above analysis result, it can be seen that Product A is the highest when placed on the second shelf. As described above, the data analysis device 300 specifies the position of the shelf with the highest sales for each product name.

[0038] On the product shelf 20 having a plurality of shelves as shown in FIG. 2, one or more types of one or more products are displayed. Also, on the product shelf 20, products having a certain product name and other products having the same or different product names are often arranged side by side vertically or horizontally. The product shelf 20 thus includes a plurality of products displayed on each shelf. Therefore, this weight takes into account the positional relationship between the products displayed on a certain shelf and the products displayed on other shelves and the relationship with the sales of the products.

[0039] Note that in the data analysis device 300 according to this embodiment, the sales amount is used as the value of y A used for the analysis, but for example, the number of sales may be used. At this time, when the sales amount of a product in the sales data and the unit price of the product are included and the number of sales is not included, the data analysis device 300 may divide the sales amount by the unit price to obtain the number of sales and then use the number of sales as the value of y A .

[0040] In addition, the data analysis device 300 may use, as an analysis method, a regression analysis method such as the least squares method as shown in Equation (1), or a classification method.

[0041] For example, when y A represents specific values such as sales amount or number of sales, it is preferable for the data analysis device 300 to perform analysis using a regression analysis method. As the regression analysis method, in addition to the least squares method described above, for example, linear regression, maximum likelihood method, Bayesian linear regression, neural network, etc. may be used.

[0042] Also, when y A represents, for example, the degree of sales, it is preferable for the data analysis device 300 to perform analysis using a classification method. The case where it represents the degree of sales means, for example, the case where y A is a value shown in 10 levels from 1 to 10 according to sales. As the classification method, for example, generative models such as naive Bayes, logistic regression, support vector machine, neural network, nearest neighbor classification, decision tree, etc. may be used. Thus, the data analysis device 300 can appropriately select an analysis method according to the content of the learning data (for example, the type of the value of y).

[0043] As described above, the analysis result output by the data analysis device 300 shows, for each product name, the weights at each shelf level of the product shelf 20 shown in FIG. 2, for example. Note that the analysis result output by the data analysis device 300 is not limited to each product name, and may be, for example, the weights at each shelf level for each product type, each adjacent product name, or each adjacent product type. Also, the analysis result output by the data analysis device 300 may be the weight for an adjacent product adjacent to the product indicated by the product name for each product name. The analysis result may be a combination of these. Also, an adjacent product is a product that is adjacent to at least one of the left and right or at least one of the top and bottom within a predetermined range.

[0044] The data analysis device 300 transmits the analysis result to the shelf layout support device 100 as relationship information representing the relationship between the positional relationship between products and the sales of products. Note that the data analysis device 300 may be configured to be integrated with the shelf layout support device 100 as an analysis unit.

[0045] (Shelf layout support device 100) FIG. 3 is a functional block diagram showing an example of the functional configuration of the shelf layout support device 100 of the shelf layout support system 1 according to the present embodiment. Note that FIG. 3 shows a configuration specific to the present invention, and it goes without saying that the shelf layout support device 100 shown in FIG. 3 may have members not shown in FIG. 3.

[0046] As shown in FIG. 3, the shelf layout support device 100 includes a generation unit 110, a prediction unit 120, a selection unit 130, an inventory information storage unit 140, and a relationship information storage unit 150. Note that the inventory information storage unit 140 and the relationship information storage unit 150 may be realized by a single storage unit. Also, the inventory information storage unit 140 and the relationship information storage unit 150 may be realized by storage devices separate from the shelf layout support device 100, respectively.

[0047] The inventory information storage unit 140 stores information related to inventory (inventory information) transmitted from the inventory management device 200. The relationship information storage unit 150 stores the relationship information transmitted from the data analysis device 300. Note that the shelf layout support device 100 may not include the inventory information storage unit 140 and the relationship information storage unit 150. In this case, the shelf layout support device 100 may be configured to communicate with the inventory management device 200 and the data analysis device 300 to acquire information necessary for the processing described later.

[0048] The generation unit 110 receives information about a product that a seller desires to sell, for example, via an input unit (not shown), and identifies the product indicated by the received information as a specific product. Further, the generation unit 110 may refer to the inventory information storage unit 140 and identify, as specific products, for example, products with a large inventory or products with an approaching expiration date such as a shelf life, a consumption period, or a use period. Further, the specific products may be all in-stock products.

[0049] The generation unit 110 generates a placement candidate indicating a candidate for a product to be placed at a placeable position indicating the position of the one or more specific products on a product shelf where placement is possible, the placement candidate including at least one specific product. For example, assume that product A is a specific product and the product shelf for displaying products is a two-tier product shelf having two slots on each shelf level. And if the first level is a placeable position for product A, the generation unit 110 generates, as a placement candidate, a candidate for the product to be placed at this placeable position. In the present embodiment, since the specific product is one of product A, the generation unit 110 includes product A in the placement candidate. At this time, the generation unit 110 generates the placement candidate for all of the placeable positions (exhaustively). In this example, as described above, the placeable positions are two places, the first slot of the first level (hereinafter represented as (1,1)) and the second slot of the first level (hereinafter represented as (1,2)), so the generation unit 110 generates a placement candidate for these two places. Also, in the case of this example, since there are two placeable positions, there are two products that can be placed at this placeable position. Therefore, the generation unit 110 generates a placement candidate such that at least one of these two products is product A.

[0050] FIG. 4 shows an example of a specific placement candidate. The generation unit 110 generates a placement candidate for a product including product A to be placed at a placeable position as shown in FIG. 4. In FIG. 4, x indicates a product other than product A. When there are four types of products other than product A, namely product B, product C, and product D, x can be any of B, C, and D. In this way, the generation unit 110 exhaustively generates a placement candidate for a product including a specific product to be placed at a placeable position.

[0051] Then, the generation unit 110 generates a plurality of shelf division candidates representing the state in which the products are displayed on the product shelf 20. For example, when the in-stock products other than product A are product B, product C, and product D, the generation unit 110 generates shelf division candidates as shown in FIG. 5. Since the position where products can be placed includes product A, which is a specific product as described above, the shelf division candidates generated by the generation unit 110 include product A. Therefore, it can be said that the generation unit 110 generates shelf division candidates representing the state in which a plurality of products including product A are displayed on the product shelf 20.

[0052] Note that the shelf division candidates shown in FIG. 5 show an example of shelf division candidates including the placement candidate (1) in FIG. 4 and an example of shelf division candidates including the placement candidate (3) in FIG. 4. In the present embodiment, the generation unit 110 obtains all (by brute force) combinations of the display positions of the in-stock products for positions other than the positions (placeable positions) where the products are placed as indicated by the placement candidates (the second row in FIG. 4), and generates shelf division candidates based on the obtained combinations. Here, since the in-stock products are products B to D as described above, the generation unit 110 generates combinations of products to be placed in each slot in the second row of the product shelf for all of products B to D. Note that the products placed outside the placeable positions may be of one type or different types. Also, the products placed outside the placeable positions may include a specific product (in this case, product A).

[0053] Note that the placeable positions of the specific product may be all the slots on all the shelves of the product shelf. In this case, the generation unit 110 outputs the generated placement candidates as shelf division candidates.

[0054] The generation unit 110 outputs the generated plurality of shelf division candidates to the prediction unit 120 together with specific product information indicating a specific product (product A in the above example).

[0055] The prediction unit 120 receives, from the generation unit 110, a plurality of shelf division candidates generated by the generation unit 110 together with specific product information. The prediction unit 120 predicts the sales of the product indicated by the received specific product information for each of the received plurality of shelf division candidates based on the relationship information in the relationship information storage unit 150.

[0056] As described above, the relationship information is information as shown in the following (1) to (5). (1) The weight at each shelf level for each product name, (2) The weight at each shelf level for each product type, (3) The weight at each shelf level for each adjacent product name, (4) The weight at each shelf level for each adjacent product type, (5) The weight for the adjacent product adjacent to the product indicated by the product name for each product name.

[0057] Note that the relationship information may be a combination of these (1) to (5).

[0058] For example, in the case of "shelf division candidate (1)-1" shown in FIG. 5, product A is on the first shelf level. And the relationship information θ A indicated by the above (1) A is θ A =(0.9, 0.5). This relationship information θ A indicates the weight at each shelf level of product A, indicating that the weight of the first shelf level is 0.9 and the weight of the second shelf level is 0.5. The prediction unit 120 predicts the sales of product A in this shelf division candidate (1)-1 using the above relationship information θ

[0059] Also, for example, in the case of "Shelf Division Candidate (1)-1" shown in FIG. 5, the product adjacent to the right of Product A is Product B. And assuming that the relational information θ shown by the above (3) is θ = (adjacent product name, first-stage weight, second-stage weight) = (Product B, 0.3, 0.2). This relational information θ indicates that when the product name of the adjacent product is Product B, the first-stage weight is 0.3 and the second-stage weight is 0.2. The prediction unit 120 predicts the sales of Product A in this Shelf Division Candidate (1)-1 using the above relational information θ. In the case of "Shelf Division Candidate (1)-1" shown in FIG. 5, since the adjacent product name is Product B and Product B is in the first stage, the sales of Product A are predicted using the first-stage weight of 0.3.

[0060] Also, for example, the relational information θ shown by the above (5) A is θ A = (weight when the adjacent product is Product A, weight when the adjacent product is Product B, weight when the adjacent product is Product C, weight when the adjacent product is Product D) = (0.5, 0.7, 0.3, 0.8). This relational information θ A indicates that the weight when the adjacent product of Product A is Product A is 0.5, and the weight when the adjacent product of Product A is Product B is 0.7. Similarly, this relational information θ A indicates that the weight when the adjacent product of Product A is Product C is 0.3, and the weight when the adjacent product of Product A is Product D is 0.8. From this, it can be seen that the weight is the highest when the adjacent product of Product A is Product D. The prediction unit 120 predicts the sales of Product A in this Shelf Division Candidate (1)-1 using the above relational information θ A . For example, in the case of "Shelf Division Candidate (1)-1" shown in FIG. 5, since the product adjacent to the right of Product A is Product B, the sales of Product A are predicted using the weight of 0.7 when the adjacent product of Product A is Product B.

[0061] After that, the prediction unit 120 predicts the sales of Product A for all shelf division candidates. The relational information based on which the prediction unit 120 makes predictions may be any one of the above (1) to (5), or may be a plurality of them.

[0062] The prediction unit 120 outputs, as prediction results, the predicted sales for each shelf division candidate to the selection unit 130.

[0063] The selection unit 130 receives the prediction results from the prediction unit 120. Then, based on the received prediction results, the selection unit 130 selects the shelf division candidate with the largest sales among the plurality of shelf division candidates. For example, when the sales of product A in each of "shelf division candidate (1)-1", "shelf division candidate (1)-2", and "shelf division candidate (1)-3" are 50, 100, and 150 respectively, the selection unit 130 selects "shelf division candidate (1)-3" with the largest sales. As a result, it can be understood that for the sales of product A, a more effective product display state is that product A is displayed in the first slot of the first tier of the shelf, product D is displayed on the right side of product A, and product C is displayed below product A. The selection unit 130 can output the selected shelf division candidate as a recommended shelf division indicating the product display state including the state where a specific product is displayed at a more effective position.

[0064] (Flow of processing of the shelf division support device 100) Next, the flow of processing in the shelf division support device 100 will be described. FIG. 6 is a flowchart showing an example of the flow of processing in the shelf division support device 100 according to the present embodiment.

[0065] As shown in FIG. 6, the generation unit 110 generates arrangement candidates, which are candidates for products including a specific product, to be arranged at the available positions on the product shelf (step S61). Then, the generation unit 110 generates a plurality of shelf division candidates including the state where the specific product is arranged at the available position indicated by the generated arrangement candidates (step S62).

[0066] Then, the prediction unit 120 predicts the sales of the specific product for each of the plurality of shelf division candidates generated in step S62 based on the relevant information (step S63).

[0067] Thereafter, the selection unit 130 selects the shelf division candidate with the largest sales among the plurality of shelf division candidates based on the predicted sales (prediction results) of the specific product for each shelf division candidate (step S64).

[0068] Thus, the process of generating the recommended shelf division in the shelf division support device 100 in the present embodiment ends.

[0069] (Effect) As described above, in the shelf division support device 100 according to the present embodiment, the generation unit 110 generates a placement candidate that is a candidate for placing a product at a placement position where one or more specific products can be placed, and indicates a candidate including at least one specific product. Then, the generation unit 110 generates a shelf division candidate representing the display state of a plurality of products including the specific product on the product shelf, including a state in which at least one specific product is placed at the placement position indicated by the generated placement candidate. Then, the prediction unit 120 predicts the sales of the specific product in each of the plurality of generated shelf division candidates based on the relationship information representing the relationship between the positional relationship between the products displayed on the product shelf and the sales of the products. Then, the selection unit 130 selects, based on the prediction result, the shelf division candidate with the largest predicted sales among the plurality of shelf division candidates.

[0070] Since the prediction unit 120 predicts the sales of a specific product based on the relationship information, this prediction result can predict the sales according to the display positions of a plurality of products including the specific product. And it can be said that the shelf division candidate with the largest predicted sales is such that the relationship between the display positions of the specific product and other products has an effect on the sales.

[0071] Therefore, according to the shelf division support device according to the present embodiment, it is possible to generate a recommended shelf division indicating the display state of the products, including a state in which a specific product is displayed at a more effective position. Thus, the shelf division support device 100 according to the present embodiment can efficiently support the shelf division work, similar to the shelf division support device 10 in the first embodiment described above.

[0072] (Modification) In the modification of the present embodiment, a modification of the relationship information will be described.

[0073] Depending on the store, for example, products with an expiration date approaching, such as the best - before date, may be discounted. Therefore, the data analysis device 300 may analyze the product based on information such as products with an approaching expiration date or products on discount. For example, when the expiration date of another product (adjacent product) adjacent to a certain product is approaching, the data analysis device 300 may analyze the relationship between the positional relationship between this certain product and the adjacent product and the sales of the products. Also, the data analysis device 300 may analyze the sales on each shelf level for each product with an approaching expiration date and output it as relationship information.

[0074] And the prediction unit 120 may perform a sales prediction based on the relationship information output by the data analysis device 300, similar to the prediction unit 120 in the second embodiment described above.

[0075] Even with such a configuration, the shelf - division support device 100 according to this modification can achieve the same effects as the shelf - division support device 100 in the second embodiment described above.

[0076] Also, in the second embodiment, the prediction unit 120 predicted the sales of a specific product, but the sales of other products may also be calculated. And, for example, when there are multiple shelf - division candidates with the largest predicted sales, the selection unit 130 may select a shelf - division candidate with a larger total sales of the products included in the shelf - division candidate.

[0077] Even with such a configuration, the shelf - division support device 100 can generate a recommended shelf - division indicating the display state of the products, including a state where a specific product is displayed in a more effective position.

[0078] <Third Embodiment> Next, a third embodiment of the present invention will be described with reference to the drawings. FIG. 7 is a functional block diagram showing the functional configuration of the shelf division support device 101 in the shelf division support system 1 according to the present embodiment. For convenience of explanation, members having the same functions as those included in the drawings described in the above-described second embodiment are denoted by the same reference numerals, and their explanations are omitted. In addition, since the overall configuration of the shelf division support system 1 according to the present embodiment is the same as the configuration of the shelf division support system 1 in the second embodiment shown in FIG. 2, the description thereof is omitted.

[0079] As shown in FIG. 7, the shelf division support device 101 includes a generation unit 111, a prediction unit 120, a selection unit 130, an inventory information storage unit 140, a relationship information storage unit 150, and a template storage unit 160. Further, the shelf division support device 101 shown in FIG. 7 includes an analysis unit 301 corresponding to the data analysis device 300. Since the analysis unit 301 has the same functions as the data analysis device 300, the description thereof is omitted. By providing the shelf division support device 101 with an analysis function, the network load associated with the communication of relationship information can be reduced.

[0080] Note that the inventory information storage unit 140, the relationship information storage unit 150, and the template storage unit 160 may be realized by a single storage unit. Further, the inventory information storage unit 140, the relationship information storage unit 150, and the template storage unit 160 may each be realized by a storage device separate from the shelf division support device 101.

[0081] In the template storage unit 160, information indicating the product display state in each of a plurality of stores is stored as a template. Further, in the template storage unit 160, for example, information indicating the product display state recommended by the head office of a chain store may be stored as a template.

[0082] The generation unit 111 generates a placement candidate including at least one of the specific products to be placed at a placeable position indicating a position where the specific product on the product shelf 20 can be placed, based on the template stored in the template storage unit 160. First, the generation unit 111 identifies a specific product in the same manner as the above-described generation unit 110. Then, the generation unit 111 generates a placement candidate, which is a candidate for a product to be placed at a placeable position where the one or more specific products can be placed, based on the template. For example, assume that product A is a specific product and the product shelf for displaying products is a two-tier product shelf having two slots on each shelf level. And assume that the first level is a placeable position for product A. Here, assume that the template includes information indicating a state where (product A, product A), (product A, product B), and (product C, product A) are placed in each slot of the first level. Note that the state where (product C, product A) is placed means a state where product C is placed in the first slot of the first level and product A is placed in the second slot. In this case, the generation unit 111 generates a placement candidate based on this template. The generation unit 111 may use all of the above-described (product A, product A), (product A, product B), and (product C, product A) as placement candidates based on the template, or may use any of them as a placement candidate.

[0083] Then, the generation unit 111 generates a plurality of shelf division candidates representing a state where the products are placed on the product shelf 20. As described above, this shelf division candidate includes a state where product A is placed in at least one of the placeable positions. Note that when there are a plurality of specific products, this shelf division candidate includes a state where at least one specific product is placed in at least one of the placeable positions. The generation unit 111 determines the combination of the display positions of the in-stock products other than product A based on the template stored in the template storage unit 160. Note that the present embodiment is not limited to this, and the generation unit 111 may determine the combination of the display positions of the in-stock products other than product A by brute force in the same manner as in the above-described second embodiment. Then, the generation unit 111 generates a shelf division candidate based on the determined combination.

[0084] Note that, similar to the second embodiment, the generation unit 111 may obtain arrangement candidates for all positions where a specific product can be arranged (by brute force), and determine combinations of display positions of in-stock products other than the specific product based on the templates stored in the template storage unit 160.

[0085] Note that the positions where a specific product can be arranged may be all slots of all shelves of the product shelf. In this case, the generation unit 111 outputs the arrangement candidates generated based on the templates as shelf division candidates.

[0086] Thereafter, similar to the second embodiment, the prediction unit 120 predicts the sales of a specific product for each of the plurality of shelf division candidates, and the selection unit 130 selects a shelf division candidate based on the prediction results.

[0087] As described above, the shelf division support device 101 according to the present embodiment can obtain the same effects as the shelf division support device 100 according to the above-described second embodiment. Further, as described above, the shelf division support device 101 according to the present embodiment generates a plurality of shelf division candidates using templates prepared in advance. Thereby, compared with the shelf division support device 100 according to the above-described second embodiment, the processing amount of the shelf division candidate generation process, the processing amount of the sales prediction process, etc. can be reduced, and the load on the shelf division support device 101 can be reduced.

[0088] <Fourth Embodiment> Next, a fourth embodiment of the present invention will be described with reference to the drawings. FIG. 8 is a diagram showing an example of the overall configuration of the shelf division support system 2 according to the present embodiment. For the sake of convenience of explanation, members having the same functions as those included in the drawings described in the above-described embodiments are denoted by the same reference numerals, and the description thereof is omitted.

[0089] The shelf division support system 2 shown in FIG. 8 includes a shelf division support device 102, an inventory management device 200, a data analysis device 300, and an imaging device 500. Further, FIG. 9 is a diagram for explaining a usage scene of the shelf division support system 2 according to the present embodiment. FIG. 9 is a functional block diagram showing an example of the functional configuration of the shelf division support system 2 according to the present embodiment.

[0090] In the shelf division support system 2 according to the present embodiment, the imaging device 500 photographs the products displayed on the product shelves 20 in the store and transmits the photographed images to the shelf division support device 102.

[0091] As shown in FIG. 9, the imaging device 500 may be, for example, a terminal equipped with an imaging function such as a mobile phone terminal, a smartphone, a digital camera, a tablet, etc., or may be a surveillance camera installed in the store. If there is a location on the product shelf 20 photographed by the imaging device 500 where no product is displayed (referred to as an empty slot), the shelf division support device 102 outputs a recommended shelf division for the product shelf 20. Thereby, the operator who arranges the products can confirm this recommended shelf division on a display device (not shown) and arrange the products effective for sales in the empty slots. In this way, the shelf division support system 2 efficiently supports the shelf division task of selecting the products to be arranged in the empty slots.

[0092] Next, with reference to FIG. 10, the functional configuration of the shelf division support device 102 in the shelf division support system 2 according to the present embodiment will be described. FIG. 10 is a functional block diagram showing an example of the functional configuration of the shelf division support device 102 according to the present embodiment. As shown in FIG. 10, the shelf division support device 102 in the present embodiment includes a generation unit 112, a prediction unit 120, a selection unit 130, an inventory information storage unit 140, a relationship information storage unit 150, a recognition unit 170, and a product information storage unit 180. The inventory information storage unit 140, the relationship information storage unit 150, and the product information storage unit 180 may be realized by a single storage unit. Further, the inventory information storage unit 140, the relationship information storage unit 150, and the product information storage unit 180 may be realized by storage devices separate from the shelf division support device 102, respectively.

[0093] The product information storage unit 180 stores information for recognizing products included in the captured images captured by the imaging device 500. Specifically, the product information storage unit 180 stores an image of a product (also referred to as a master image) and / or feature amounts included in the image of the product, which are associated with information for identifying the product (for example, a product identifier for identifying the product, a product name, etc.) and stored.

[0094] The recognition unit 170 receives a captured image of the product shelf 20 captured by the imaging device 500 from the imaging device 500. Then, the recognition unit 170 refers to the information for recognizing products stored in the product information storage unit 180 and recognizes the products included in the captured image from the captured image. The method by which the recognition unit 170 recognizes products may be, for example, one using local feature amounts, templates, luminance, edges, outer shapes, shapes, color information, depth, etc., or one using other information. The method by which the recognition unit 170 recognizes products is not particularly limited and may be a general recognition method, so detailed description is omitted in this specification. Then, as a recognition result, the recognition unit 170 outputs information for identifying the recognized products (for example, a product identifier, a product name, etc.) and information indicating the position of the products on the captured image of the products (for example, coordinate values in the captured image) to the generation unit 112. Here, an example of the captured image is shown in FIG. 11. The captured image is, for example, an image as shown in FIG. 11. The product shelf 20 included in the captured image has four shelf levels, and is a product shelf in which the number of products that can be arranged on each shelf level (the number of slots) is four. A plurality of products are displayed in the product shelf 20 of FIG. 11. The alphabet in each product shown in FIG. 11 indicates the last character of the product name. In FIG. 11, for example, the product with the product name "Product A" is indicated as "A".

[0095] The recognition unit 170 recognizes products from this captured image. Then, the recognition unit 170 outputs the recognition result together with the captured image to the generation unit 112.

[0096] Note that the recognition unit 170 may be implemented as a device separate from the shelf division support device 102. In this case, the shelf division support device 102 receives the recognition result from the separate device. Thereby, the shelf division support device 102 can reduce the processing load imposed on the shelf division support device 102. Also, since the shelf division support device 102 includes the recognition unit 170, the network load related to the transmission and reception of the recognition result can be reduced.

[0097] The generation unit 112 receives the recognition result of the product from the recognition unit 170 together with the captured image. Then, the generation unit 112 determines the positions where the products are not displayed from the captured image. That is, the generation unit 112 identifies the empty slots from the captured image. Note that the identification of the empty slots may be performed by the recognition unit 170. In this case, the recognition unit 170 may transmit information representing the positions of the empty slots together with the recognition result of the product. In the case of the captured image shown in FIG. 11, the generation unit 112 identifies the third slot in the second row and the fourth slot in the second row as the empty slots.

[0098] This empty slot is a position where a product can be placed. Therefore, the generation unit 112 generates candidates for products to be placed in this empty slot, the candidates including at least one of one or more specific products. This will be described with reference to FIG. 12. FIG. 12 is a diagram for explaining the generation process of the placement candidates by the generation unit 112. As shown in FIG. 12, in the present embodiment, it is assumed that the specific product is a product that the seller wants to sell, and the products are product L, product M, and product N. Note that the specific products may be all the in-stock products managed by the inventory management device 200.

[0099] The generation unit 112 generates placement candidates that are candidates when the specific product is placed in this empty slot. For the sake of convenience of explanation, the placement candidates shown in FIG. 12 are shown corresponding to the portions of two empty slots inside the product shelf.

[0100] The generation unit 112 generates, for each of the two slots that are positions where items can be placed, a combination of specific items to be placed as placement candidates. Fig. 12 shows nine placement candidates. In Fig. 12, the empty slot on the left shows the third slot in the second row in Fig. 11, and the empty slot on the right shows the fourth slot in the second row in Fig. 11. For example, among the placement candidates shown in Fig. 12, the upper left placement candidate indicates that the combination of items to be placed in the third slot and the fourth slot in the second row of the merchandise shelf 20 is (Item L, Item L). That is, the upper left placement candidate indicates that the candidates for the items to be placed in the third slot and the fourth slot in the second row of the merchandise shelf 20 are both Item L.

[0101] Note that the placement candidates shown in Fig. 12 are combinations of specific items to be placed in each of the two slots that are positions where items can be placed, but the present embodiment is not limited to this. As long as a specific item is included in the item to be placed in any one of the two slots that are positions where items can be placed, the generation unit 112 is acceptable. For example, the generation unit 112 may set the combination of items to be placed in the third slot and the fourth slot in the second row as (Item L, Item A). This Item A is an inventory item that is not a specific item. Also, when there is only one slot where items can be placed, the generation unit 112 sets any one of the specific items as a placement candidate.

[0102] Note that the generation unit 112 may generate the placement candidates for all positions where items can be placed (by brute force), similar to the second embodiment. Also, the generation unit 112 may generate based on a template, similar to the third embodiment. In this case, the shelf division support device 102 may have a configuration including a template storage unit 160, similar to the shelf division support device 101 in the third embodiment.

[0103] Then, based on the generated placement candidates, the generation unit 112 places at least one of the specific products at the placement possible positions, and generates a shelf division candidate representing a state in which the recognized products are placed at positions corresponding to the positions in the captured images of the recognized products. In the first row of the product shelf 20 included in the captured image shown in FIG. 11, products A, A, B, and B are arranged in order from the left. The recognition unit 170 recognizes these products and their positions from the captured image. Therefore, the generation unit 112 sets the state of the first row included in the shelf division candidate to a state in which products A, A, B, and B are arranged in order from the left. Similarly, the generation unit 112 determines the products to be placed on the other shelves based on the recognition result.

[0104] Then, the generation unit 112 outputs the generated shelf division candidate to the prediction unit 120.

[0105] Thereafter, similar to the second embodiment, the prediction unit 120 predicts the sales of the specific product for each of the plurality of shelf division candidates, and the selection unit 130 selects a shelf division candidate based on the prediction result.

[0106] (Flow of processing of the shelf division support device 102) Next, the flow of processing in the shelf division support device 102 will be described. FIG. 13 is a flowchart showing an example of the flow of processing in the shelf division support device 102 according to the present embodiment.

[0107] As shown in FIG. 13, first, the recognition unit 170 receives a captured image of the product shelf captured by the imaging device 500 (step S131). Then, the recognition unit 170 recognizes products from the received captured image (step S132).

[0108] After that, the generation unit 112 generates a placement candidate, which is a candidate for a product including a specific product, to be placed at a position (placeable position) determined by the recognition unit 170 as a position where no product is displayed (step S133). Then, the generation unit 112 generates a shelf division candidate representing a state in which the specific product is placed at the placeable position and the recognized product is placed at a position corresponding to the position of the recognized product in the captured image of the product recognized from the captured image (step S134).

[0109] Then, the prediction unit 120 predicts the sales of the specific product for each of the plurality of shelf division candidates generated in step S134 based on the related information (step S135).

[0110] After that, the selection unit 130 selects the shelf division candidate with the largest sales among the plurality of shelf division candidates based on the predicted sales (prediction results) of the specific product for each shelf division candidate (step S136).

[0111] Thus, the generation process of the recommended shelf division in the shelf division support device 102 in the present embodiment ends.

[0112] As described above, according to the shelf division support system 2 according to the present embodiment, similar to each of the above-described embodiments, it is possible to generate a recommended shelf division indicating the display state of products, including a state in which a specific product is displayed at a more effective position. Therefore, the shelf division support system 2 according to the present embodiment can efficiently support the shelf division work, similar to each of the above-described embodiments.

[0113] <Example of Hardware Configuration> Here, an example of the hardware configuration capable of realizing the shelf division support devices (10, 100 to 102) according to each of the above-described embodiments will be described. The above-described shelf division support devices (10, 100 to 102) may be realized as dedicated devices, or may be realized using a computer (information processing device).

[0114] FIG. 14 is a diagram illustrating the hardware configuration of a computer (information processing apparatus) capable of realizing each embodiment of the present invention.

[0115] The hardware of the information processing apparatus (computer) 90 shown in FIG. 14 includes a CPU (Central Processing Unit) 91, a communication interface (I / F) 92, an input / output user interface 93, a ROM (Read Only Memory) 94, a RAM (Random Access Memory) 95, a storage device 97, and a drive device 98 for a computer-readable storage medium 99, and these are configured to be connected via a bus 96. The input / output user interface 93 is a man-machine interface such as a keyboard which is an example of an input device and a display as an output device. The communication interface 92 is a general communication means for the devices (FIGS. 1, 3, 7, and 10) according to each of the above-described embodiments to communicate with an external device via a communication network 80. In such a hardware configuration, the CPU 91 controls the overall operation of the information processing apparatus 90 that realizes the shelf division support apparatuses (10, 100 to 102) according to each of the embodiments.

[0116] The present invention described by taking each of the above-described embodiments as an example can be achieved, for example, by supplying a program (computer program) capable of realizing the processing described in each of the above-described embodiments to the information processing apparatus 90 shown in FIG. 14 and then reading out and executing the program by the CPU 91. Note that such a program may be, for example, a program capable of realizing various processes described in the flowcharts (FIGS. 6 and 13) referred to in the description of each of the above-described embodiments, or a program capable of realizing each part (each block) shown in the device in the block diagrams shown in FIGS. 1, 3, 7, and 10.

[0117] Also, the program supplied into the information processing apparatus 90 may be stored in a readable and writable temporary storage memory (95) or a non-volatile storage device (97) such as a hard disk drive. That is, in the storage device 97, the program group 97A is a program capable of realizing the functions of each part shown in the shelf division support apparatuses (10, 100 to 102) in each of the above-described embodiments. Also, the various storage information 97B is, for example, the shelf division candidates, related information, inventory information, photographed images, recognition results, templates, recommended shelf divisions, etc. in each of the above-described embodiments. However, when implementing the program in the information processing apparatus 90, the constituent units of the individual programs and modules are not limited to the divisions of the respective blocks shown in the block diagram, and those skilled in the art may appropriately select them during implementation.

[0118] Also, in the above case, the method of supplying the program into the apparatus may adopt a general procedure at present, such as a method of installing it into the apparatus via various computer-readable recording media (99) such as a CD (Compact Disk)-ROM and a flash memory, or a method of downloading it from the outside via a communication line (80) such as the Internet. And in such a case, the present invention can be regarded as being constituted by the code (program group 97A) constituting such a computer program or the storage medium (99) in which such code is stored.

[0119] In each of the above-described embodiments, the case where the functions shown in each block shown in the block diagram are realized by a software program as an example executed by the CPU 95 shown in FIG. 14 has been described. However, the functions shown in each block shown in the block diagram may be realized in part or in whole as a hardware circuit.

[0120] Note that each of the above-described embodiments is a preferred embodiment of the present invention, and the scope of the present invention is not limited only to the above-described embodiments. Those skilled in the art can make modifications and substitutions to the above-described embodiments and construct various modified forms without departing from the gist of the present invention.

[0121] This application claims priority based on Japanese Patent Application No. 2015-116477 filed on June 9, 2015, and incorporates all of its disclosures herein.

Description of Reference Numerals

[0122] 1 Shelf Division Support System 2 Shelf Division Support System 10 Shelf Division Support Device 11 Generation Unit 12 Prediction Unit 13 Selection Unit 100 Shelf Division Support Device 101 Shelf Division Support Device 102 Shelf Division Support Device 110 Generation Unit 111 Generation Unit 112 Generation Unit 120 Prediction Unit 130 Selection Unit 140 Inventory Information Storage Unit 150 Relationship Information Storage Unit 160 Template Storage Unit 170 Recognition Unit 180 Product Information Storage Unit 200 Inventory Management Device 300 Data Analysis Device 301 Analysis Unit 400 Network 500 Imaging Device 20 Product Shelf 21 POS Terminal

Claims

1. recognition means for recognizing a product from a captured image of a product shelf; position recognition means for recognizing a position on the product shelf where no product is displayed from the captured image; specification means for specifying a product to be displayed at a position where no product is displayed; generation means for generating a plurality of shelf division candidates including a specific product composed of the product recognized by the recognition means and the product specified by the specification means; prediction means for predicting the sales of the plurality of shelf division candidates based on relationship information representing the relationship between the positional relationship of products on the product shelf and the sales of the products; A shelf division support device comprising:

2. The generation means generates a placement candidate which is a candidate for a product to be placed at a placement possible position on the product shelf where one or more of the specific products can be placed, and which indicates a candidate including at least one of the specific products, and generates the shelf division candidate including a state in which at least one of the specific products is placed at the placement possible position indicated by the generated placement candidate. The shelf division support device according to claim 1.

3. The generation means generates the placement candidate for all of the placement possible positions. The shelf division support device according to claim 2.

4. The generation means generates the placement candidate based on a template prepared in advance. The shelf division support device according to claim 2.

5. A computer recognizes a product from a captured image of a product shelf, recognizes a position on the product shelf where no product is displayed from the captured image, specifies a product to be displayed at a position where no product is displayed, generates a plurality of shelf division candidates including a specific product composed of the recognized product and the specified product, based on relationship information representing the relationship between the positional relationship of products on the product shelf and the sales of the products, predicts the sales of the plurality of shelf division candidates. A shelf division support method.

6. A process of recognizing a product from a captured image of a product shelf, a process of recognizing a position on the product shelf where no product is displayed from the captured image, a process of specifying a product to be displayed at a position where no product is displayed, a process of generating a plurality of shelf division candidates including a specific product composed of the recognized product and the specified product, a process of predicting the sales of the specific product in the plurality of shelf division candidates based on relationship information representing the relationship between the positional relationship of products on the product shelf and the sales of the products; A program for causing a computer to execute.

Images