Trading card discrimination system, trading card discrimination method, and trading card discrimination program
The trading card discrimination system uses a trained model to calculate feature vectors for rapid and accurate identification and authenticity determination, addressing the complexity and counterfeiting issues in existing systems.
Patent Information
- Application Number
- JP2023188215
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Priority Date
- 2022-11-16
- Filing Date
- 2023-11-02
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2043-11-02
AI Technical Summary
Existing trading card identification systems face challenges in efficiently and accurately distinguishing between genuine and counterfeit cards, and require complex processing due to the large number of card variations and increasing counterfeits.
A trading card discrimination system that uses a trained model to calculate feature vectors from image data, comparing them to stored feature vectors to identify cards and determine authenticity, utilizing a combination of feature vectors and hash values for precise matching.
Enables rapid and accurate identification of trading cards, including determining authenticity and type, even for rare or recently released cards, with improved efficiency and reduced processing complexity.
Smart Images

Figure 0007592933000001 
Figure 0007592933000002 
Figure 0007592933000003
Abstract
Description
[Technical field]
[0001] The present invention relates to a trading card discrimination system, a trading card discrimination method, and a trading card discrimination program. [Background technology]
[0002] Trading cards are cards printed with various pictures, photographs, etc., and are used in trading card games, and are also collected, exchanged, and traded by collectors. There is a wide variety of trading cards on the market, and new cards are constantly being released. For this reason, in stores that buy and sell trading cards, it is difficult for inexperienced staff, let alone experienced staff, to easily identify trading cards that are to be bought. In order to solve this problem, a technology has been disclosed in which a predetermined number of sets of comparison image data that can be collectively deployed in a semiconductor memory unit of a store device is extracted from image master data based on purchase needs trend information, and the set of comparison image data is delivered to the store device, and the photographed image data of the object of purchase is compared with the cached multiple sets of comparison image data, and the comparison image data that matches the image data of the object of purchase is identified (see Patent Document 1). [Prior art documents] [Patent documents]
[0003] [Patent Document 1] JP 2019-82870 A Summary of the Invention [Problem to be solved by the invention]
[0004] However, with this technology, even if the comparison image data is narrowed down based on purchase needs, the purchase target image data and the comparison image data must be compared, which makes the process complicated and time-consuming. In addition, since the number of counterfeit trading cards has increased in recent years, it is also necessary to determine whether the trading cards to be purchased are genuine or counterfeit.
[0005] The present invention has been made in view of the above, and has an object to provide a trading card discrimination system, a trading card discrimination method, and a trading card discrimination program that can easily discriminate between trading cards. Another object of the present invention is to provide a trading card discrimination system, a trading card discrimination method, and a trading card discrimination program that can easily discriminate between the authenticity of a trading card. [Means for solving the problem]
[0006] In order to solve the above-mentioned problems, the present invention uses a trained model that takes image data as input and outputs features to calculate a set of features from image data of a trading card to be discriminated, calculates a distance between the first feature vector and a second feature vector based on a first feature vector consisting of a set of features that is stored in a feature memory unit that associates and stores card identification information with the set of features, and a second feature vector consisting of the set of features of the calculated image data, determines whether the calculated distance meets a predetermined condition, and if the calculated distance meets the predetermined condition, obtains card identification information corresponding to the first feature vector that meets the predetermined condition from the feature memory unit, and obtains card information associated with the obtained card identification information from the card information memory unit.
[0007] The present invention is also characterized in that it calculates features of image data of a trading card to be discriminated using a trained model that inputs image data and outputs features, calculates a distance between the first feature vector and the second feature vector based on a first feature vector consisting of a set of calculated features and a second feature vector consisting of a set of features associated with the card identification information of the image data, which is stored in a feature memory unit that associates and stores the card identification information and the set of features, and discriminates the authenticity of the image data based on whether the calculated distance meets a predetermined condition. Effect of the Invention
[0008] According to the present invention configured as described above, it is possible to easily distinguish trading cards. Also, according to the present invention, it is possible to easily distinguish whether a trading card is genuine or counterfeit. [Brief description of the drawings]
[0009] [Figure 1] 1 is a block diagram showing a configuration of a trading card discrimination system 10 including a discrimination server 100 and a shop terminal 200 according to a first embodiment. [Diagram 2] 2 is an explanatory diagram showing an example of the data configuration of a card information storage unit 121 and a feature amount storage unit 122. FIG. [Diagram 3] 13 is a flowchart showing a trading card discrimination process executed by the discrimination server 100 and the shop terminal 200. [Figure 4] FIG. 11 is a block diagram showing a configuration of a trading card discrimination system 20 including a discrimination server 300 and a shop terminal 200 according to a second embodiment. [Diagram 5] 13 is a flowchart showing a trading card authenticity determination process executed by the determination server 300 and the shop terminal 200. [Figure 6] FIG. 13 is an explanatory diagram showing an example of features output by the trained model 323. [Figure 7]FIG. 11 is a block diagram showing a configuration of a trading card discrimination system 30 including a discrimination server 400 and a mobile terminal 500 according to a third embodiment. [Figure 8] 13 is a flowchart showing a trading card discrimination process executed by the discrimination server 400 and the mobile terminal 500. [Figure 9] FIG. 11 is a block diagram showing a configuration of a trading card discrimination system 40 including a discrimination server 600 and a shop terminal 200 according to a fourth embodiment. [Figure 10] 13 is a flowchart showing a procedure of a feature updating process executed by the discrimination server 600 and the shop terminal 200, following the flowchart of FIG. 5 according to the second embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0010] Hereinafter, examples of the trading card discrimination system, trading card discrimination method, and trading card discrimination program according to the present application will be described with reference to the accompanying drawings. Note that the following explanations are examples of the embodiments of the present application, and the trading card discrimination system, trading card discrimination method, and trading card discrimination program according to the present application are not limited to these examples.
[0011] <Example 1> 1 is a block diagram showing a configuration of a trading card discrimination system 10 including a discrimination server 100 and a store terminal 200 according to a first embodiment. The discrimination server 100 is a computer that discriminates trading cards. The store terminals 200-1 to 200-n (hereinafter referred to as store terminals 200) are computers operated in stores that purchase trading cards and are connected to the discrimination server 100 via a network N. The store terminals 200 transmit image data of trading cards to the discrimination server 100 and display trading card information received from the discrimination server 100 on a display screen.
[0012] First, a description will be given of the discrimination server 100. The discrimination server 100 includes a control unit 110, a storage unit 120, and a communication unit 130, as shown in FIG.
[0013] The memory unit 120 stores programs for controlling each unit, application programs, various control information, intermediate files, etc., as well as a card information memory unit 121, a feature memory unit 122, and a trained model 123.
[0014] FIG. 2 is an explanatory diagram showing an example of the data configuration of the card information storage unit 121 and the feature amount storage unit 122. The card information storage unit 121 stores information about trading cards. As shown in FIG. 2(a), the card information storage unit 121 stores a card type, a card ID, a rarity, image data, a hash value, an evaluation amount, and other information in association with each other. A combination of the card type, the card ID, and the rarity is called card identification information, and uniquely identifies a trading card. Among the data stored in the card information storage unit 121, a part or all of the data other than the card identification information is called card information. Note that the card identification information may be a combination of the card ID and the rarity when the card type is previously determined to be one type, and may be only the rarity when the card type and the card ID are previously determined.
[0015] Card type is information indicating the type of card, such as the game name (e.g., Card Game A, Card Game B, etc.) or sport name (e.g., Professional Baseball, J League, etc.). Card ID is information that identifies individual cards of the same card type. Rarity is information that indicates the rarity of cards when the card ID is the same but there are differences in the design, background color, processing, etc.
[0016] The image data is image data of a trading card. The image data may be image data of only one side of a card, or image data of both sides of a card. The hash value is a hash value calculated from the image data. A perceptual hash algorithm such as dHash (difference hash) is used to calculate the hash value. Unlike conventional hash functions, the perceptual hash value is closer to the hash value when data is similar. The valuation amount is the purchase price of the card. The valuation amount may be obtained at a predetermined timing from a purchase price management server (not shown) and stored. The valuation amount may also be a price or a discount amount or discount rate according to each card condition (good condition, scratches, scuffs, etc.). As other information, a predetermined area (one or more) when the image data input to the trained model 123 is not the entire card, the card name to be displayed on the store terminal 200, etc. are stored.
[0017] The feature storage unit 122 receives image data of a trading card as input, and stores the feature amounts of image data of the trading card output from the trained model 123, which outputs the feature amounts. As shown in Fig. 2(b), the feature storage unit 122 stores the card type, card ID, rarity, feature amounts 1 to m, and other information in association with each other.
[0018] An example of the data configuration of the card information storage unit 121 and the feature amount storage unit 122 is as described above, but if the card type is known in advance by inputting it into the store terminal 200 or by another method, the card information storage unit 121 and the feature amount storage unit 122 may be stored for each card type. Furthermore, if the card type and card ID are known in advance, the card information storage unit 121 and the feature amount storage unit 122 may be stored for each card type and card ID.
[0019] The trained model 123 is a trained model that receives image data of a trading card as input and outputs a feature amount. As an example, the trained model 123 is a trained model used in a convolutional neural network (CNN) or the like. The trained model 123 is trained in advance to read image data of a trading card and output a feature amount capable of identifying a trading card (i.e., a combination of a card type, a card ID, and a rarity). The trained model 123 may be trained in advance to output a feature amount capable of identifying a card ID and a rarity level for each card type, or may be trained in advance to output a feature amount capable of identifying a rarity level for each card type and a card ID. The trained model 123 may use an existing trained model provided for classifying image data.
[0020] The control unit 110 controls the operation of each unit of the discrimination server 100 by expanding and executing various programs and control information stored in the storage unit 120. The control unit 110 includes a feature amount calculation unit 111, a distance calculation unit 112, a card discrimination unit 113, a hash value discrimination unit 114, and an information acquisition unit 115.
[0021] The feature calculation unit 111 calculates the feature amounts of a trading card by inputting image data of the trading card and outputting feature amounts using the trained model 123. The feature calculation unit 111 inputs image data of a trading card to be discriminated (hereinafter referred to as a discrimination target card) to the trained model 123 and outputs feature amounts from the trained model 123 to calculate a set of feature amounts of the discrimination target card.
[0022] The distance calculation unit 112 calculates the distance between the feature vector of the card to be discriminated and the feature vector of each of the trading cards, based on the feature vector consisting of a set of feature amounts of the card to be discriminated and the feature vector of each of the trading cards stored in the feature amount storage unit 122. Note that instead of the distance from the feature vector, the angle between the feature vector of the card to be discriminated and the feature vector of each of the trading cards may be calculated.
[0023] The card discrimination unit 113 discriminates which trading card the discrimination target card is based on whether or not the distance or angle between the characteristic vector of the discrimination target card and the characteristic vector of each trading card meets a predetermined condition.
[0024] The hash value determination unit 114 calculates a hash value from the image data of the target card using a hash function. The hash value determination unit 114 determines whether the hash value of each trading card stored in the card information storage unit 121 and the hash value of the target card match predetermined conditions, and determines which trading card the target card is.
[0025] The information acquisition unit 115 acquires card identification information corresponding to the trading card identified by the card discrimination unit 113 from the feature amount storage unit 122, and acquires card information associated with the card identification information from the card information storage unit 121. The card information may also include image data in order to display the card information on the display screen of the store terminal 200 in comparison with the image data of the card to be identified.
[0026] The communication unit 130 is connected to the store terminal 200 and other devices via the network N so that they can communicate with each other, and transmits and receives data between the store terminal 200 and other devices. The network N is any communication network, such as the Internet, an intranet, a LAN (Local Area Network), a VPN (Virtual Private Network), or a mobile communication network, or a combination thereof, and part or all of it may be wired or wireless.
[0027] Next, a description will be given of the configuration of the store terminal 200. The store terminal 200 includes a communication unit 201, a control unit 202, an input / output unit 203, and a scanner 204. The store terminal 200 also works in conjunction with a POS (Point of Sale) system (not shown) to execute a trading card purchase process.
[0028] The communication unit 201 is connected to the discrimination server 100 and other devices via the network N so as to be able to communicate with each other, and transmits and receives data between the discrimination server 100 and other devices. The communication unit 201 transmits image data of trading cards and information related to the trading cards to the discrimination server 100. The communication unit 201 receives the card identification information and card information of each trading card transmitted from the discrimination server 100.
[0029] The control unit 202 controls the operation of each unit of the store terminal 200 by expanding and executing various programs and control information stored in a storage unit (not shown).
[0030] The input / output unit 203 receives operations by an operator from an input device and displays the results of the operations, menu screens, and the like on the display screen. The input / output unit 203 receives input of information about trading cards from the input screen. The input / output unit 203 also displays the card identification information and card information of each trading card transmitted from the identification server 100 on the display screen.
[0031] The scanner 204 reads the trading cards and generates image data for each trading card. The scanner 204 is provided with a feeder that does not bend the trading cards, so that the scanner 204 can continuously read the trading cards without applying unnecessary force to them.
[0032] A description will now be given of the trading card discrimination process executed by the discrimination server 100 and the shop terminal 200 configured as described above.
[0033] In the store terminal 200, the scanner 204 reads a trading card (step S301) and generates image data of the trading card. The scanner 204 may read one or more trading cards. The input / output unit 203 accepts input of information about the read trading card from an input screen displayed on the display device (step S302). The information about the trading card is, for example, the card status (hereinafter referred to as input information). The communication unit 201 transmits the image data of the trading card and the input information to the discrimination server 100 (step S303).
[0034] In the discrimination server 100, the communication unit 130 receives the image data of the trading card and the input information transmitted from the store terminal 200 (step S304). The feature calculation unit 111 inputs the image data of the discrimination target card to the trained model 123 (step S305). The feature calculation unit 111 outputs the feature of the image data of the discrimination target card from the trained model 123 (step S306). More specifically, when the image data of the discrimination target card is input to the trained model 123, a set of feature is output.
[0035] The distance calculation unit 112 calculates the distance between the feature vector of the discrimination target card and the feature vector of the trading card based on the feature vector consisting of a set of feature amounts of the discrimination target card and the feature vector consisting of a set of feature amounts of the trading card (step S307). Here, the feature vector of the trading card is a feature vector consisting of a set of feature amounts of each trading card (i.e., each card identification information) stored in the feature storage unit 122.
[0036] The card discrimination unit 113 judges whether the distance of the feature vector calculated by the distance calculation unit 112 matches a predetermined condition (step S308). The predetermined condition is, for example, whether the distance of the feature vector is within a predetermined threshold. When there are a plurality of trading cards whose distance of the feature vector is within a predetermined threshold, all of the plurality of trading cards may be set as candidates for the discrimination target card. In addition, the trading card whose distance of the feature vector is within a predetermined threshold and whose distance is the shortest may be discriminated as the discrimination target card. Note that the distance calculation unit 112 may calculate the angle of the feature vector instead of the distance of the feature vector. When the angle of the feature vector is calculated, the card discrimination unit 113 discriminates the trading card according to whether the angle of the feature vector is within a predetermined threshold, as in the case of the distance of the feature vector. The same applies to other embodiments.
[0037] If the card discrimination unit 113 determines that the distance of the feature vector matches the predetermined condition (step S308: Yes), the information acquisition unit 115 acquires the card identification information and card information of the trading card that matches the predetermined condition (step S309). More specifically, the information acquisition unit 115 acquires the card identification information of the trading card whose distance of the feature vector matches the predetermined condition from the feature amount information storage unit 122. The information acquisition unit 115 acquires card information such as image data and evaluated amount associated with the card identification information from the card information storage unit 121. The information acquisition unit 115 associates the card identification information and card information with each discrimination target card.
[0038] If the card discrimination unit 113 determines that the distance of the feature vector does not meet the predetermined condition (step S308: No), the card discrimination unit 113 determines that the card cannot be discriminated (step S310). The information acquisition unit 115 sets "no card information" as the card information of the discrimination target card (step S311). The communication unit 130 transmits the card identification information and card information of each discrimination target card to the store terminal 200 (step S312).
[0039] In the store terminal 200, the communication unit 201 receives the card identification information and card information of the cards to be discriminated from the discrimination server 100 (step S313). The input / output unit 203 displays the card identification information and card information of each of the cards to be discriminated on the display screen (step S314). The communication unit 201 transmits the card identification information and card information of each of the cards to be discriminated to a POS system (not shown). The POS system can execute a trading card purchase process by linking the card identification information and card information of each of the cards to be discriminated.
[0040] As described above, the discrimination server 100 can easily discriminate which trading card is the discrimination target card by using the feature amount of the image data of the trading card. Also, by storing the feature amount of the image data of at least one genuine trading card in the feature amount storage unit 122, the discrimination target card can be discriminated. This allows even rare cards that are difficult to obtain in large quantities and recently released cards to be discriminated.
[0041] Furthermore, when determining the card to be identified, if the card does not meet the specified conditions, it is determined that the card cannot be identified, thereby making it possible to determine whether the card is a trading card not stored in the card information storage unit 121 (for example, a rare card or a card that has just been released) or whether the card is possibly a counterfeit. Even if the card to be identified does not meet the specified conditions, it may be possible to indicate that the card does not meet the specified conditions and present a trading card with the closest feature vector distance. Furthermore, when a trading card is presented as the result of the identification of the card to be identified, the degree of match with the trading card (for example, the distance of the feature vector) may be displayed together with the card type, card ID, and rarity.
[0042] If a card that is determined to be unidentifiable is genuine, it is a rare card or a card that has just been released, and so image data of the card to be identified is input, and the features output from trained model 123 are stored in feature storage unit 122 in association with the card type, card ID, and rarity, and the card type, card ID, rarity, image data, hash value, evaluation amount, etc. may also be stored in card information storage unit 121. This allows new trading cards to be identified.
[0043] In the above process, the distance calculation unit 112 calculates the distance of the feature vector by using all of the feature amounts 1 to m as the feature vector, but it is not necessary to use all of the feature amounts 1 to m, and for example, a feature amount that increases the difference in the distance of the feature vector may be selected for each card type and used as the feature vector. Also, when determining the rarity of the same card ID, a feature amount that increases the difference in the distance of the feature vector may be selected for each card ID and used as the feature vector.
[0044] The image data input to trained model 123 does not have to be the image data of an entire trading card, but may be only a portion of the image data, for example, a predetermined area such as the center area of the card or a corner area of the card, an area surrounded by a frame line within the card, or a different area for each card type or card ID. In this case, the feature amount of the image data of the predetermined area of the card is stored in feature amount storage unit 122, and the distance of the feature vector is calculated from the feature amount of the same area for the image data of the card to be identified, to identify the trading card.
[0045] In the above-described process, the card state is inputted by the store terminal 200, but the card state may be determined by the trained model 123. When the card state is determined using the trained model 123, the card type, card ID, rarity, and card state may be determined by one trained model 123. Also, a trained model 123 that determines only the card state may be generated, and the card type, card ID, and rarity may be determined by a different trained model 123. Note that the card type, card ID, and rarity may each be determined by a separate trained model 123, or any combination of the card type, card ID, and rarity may be determined by the trained model 123.
[0046] The image data of the card to be identified may be the image data read by the scanner 204 as is, or the feature amount of the image data corrected according to the environment set in the scanner 204 may be used to identify the trading card. Furthermore, the trading card may be identified using the feature amount of monochrome or grayscale image data generated from the image data or the feature amount of image data extracted from RBG components, etc. The above-mentioned image data feature amounts may be freely combined and used for card identification.
[0047] Next, the discrimination of trading cards using hash values will be described. The hash value discrimination unit 114 calculates a hash value of the image data of the card to be discriminated received from the store terminal 200. The hash value discrimination unit 114 discriminates which trading card the card to be discriminated is based on whether or not the distance between the hash value of the image data of the card to be discriminated and the hash value of each trading card stored in the card information storage unit 121 is within a threshold. The hash value discrimination unit 114 calculates the distance using the Hamming distance of the hash values of the image data to be compared. The hash value discrimination unit 114 determines that the images match more when the distance between the hash values is closer.
[0048] In the flowchart of Fig. 3, the hash value determination unit 114 performs card determination using a hash value before performing steps S305 to S311 shown in Fig. 3. If the hash value determination unit 114 is able to determine a trading card, the card identification information and card information of the trading card determined by the hash value determination unit 114 may be transmitted to the store terminal 200. Furthermore, even if the hash value determination unit 114 is able to determine a trading card, the processing of steps S305 to S311 may be further performed.
[0049] If the hash value discrimination unit 114 has narrowed down the target card to one of multiple trading cards, the processing of steps S305 to S311 may be further executed to discriminate the trading card from the narrowed down cards. Furthermore, if the card discrimination unit 113 has extracted multiple candidates for the target card, after step S311, the hash value discrimination unit 114 may calculate a hash value for the target card and discriminate the trading card by comparing the hash value with each of the multiple candidate trading cards.
[0050] The preparations to be made before executing the above-mentioned trading card discrimination process will be described. When image data of a genuine trading card is obtained, the image data is input to the trained model 123, and the feature values output from the trained model 123 are stored in the feature storage unit 122. The feature storage unit 122 stores the feature values in association with the card type, card ID, and rarity. When a plurality of image data of the same card type, card ID, and rarity are obtained, the center values of the feature values 1 to m (the average values of the feature values 1 to m of each image data, which may be the center of gravity, median, or mode of the distribution of the feature values 1 to m of each image data) are stored in the feature storage unit 122. The card information storage unit 121 stores information related to the trading card in association with the card type, card ID, and rarity.
[0051] The features and classification results output using the trained model 123 will be described. The trained model 123 includes, as an example, an input layer, multiple convolution layers and max pooling layers, multiple fully connected layers, and an output layer (softmax function). The trained model 123 inputs image data to the input layer, and outputs classification results indicating the probability that the input image data is each trading card from the output layer. The trained model 123 also outputs features from the fully connected layer before outputting the classification results.
[0052] Next, the learning of the trained model 123 itself will be described. The trained model 123 is trained so that it can distinguish between trading cards with different card types, card IDs, and rarity in the identification result by inputting image data of genuine trading cards as training data. By training the trained model 123 itself in this way, the difference in the feature amounts of cards with different card types, card IDs, and rarity becomes large, and the accuracy of card discrimination in the card discrimination unit 113 can be improved. The trained model 123 may generate three trained models that separately discriminate the card type, card ID, and rarity, and discriminate the card type, card ID, and rarity, respectively. In addition, a trained model that discriminates the card type and card ID, a trained model that discriminates the card ID and rarity, and a trained model that discriminates the card type and rarity may be generated, and the card type and card ID, the card ID and rarity, and the card type and rarity may be discriminated.
[0053] <Example 2> Fig. 4 is a block diagram showing the configuration of a trading card discrimination system 20 including a discrimination server 300 and a store terminal 200 according to the second embodiment. The discrimination server 300 is a computer that discriminates the authenticity of cards to be discriminated. As shown in Fig. 4, the discrimination server 300 includes a control unit 310, a storage unit 320, and a communication unit 130. In each device and each unit, when the reference numerals are the same as those in the first embodiment, they have almost the same functions and configurations, so the description here is omitted and the above description is referred to.
[0054] The authenticity discrimination unit 313 discriminates whether the target card is genuine or counterfeit based on the image data of the target card. More specifically, the authenticity discrimination unit 313 inputs the image data of the target card to the trained model 323, calculates the distance between a feature vector consisting of a set of feature amounts of the output target card and a feature vector consisting of a set of feature amounts of a genuine trading card, and discriminates whether the target card is genuine. The authenticity discrimination unit 313 also discriminates whether the target card is genuine based on the difference between the distance between the feature vector of the target card and the feature vector of the genuine trading card and the distance between the feature vector of the target card and the feature vector of the counterfeit trading card.
[0055] The feature storage unit 322 stores authenticity information (whether the product is genuine or counterfeit) in association with the card type, card ID, and rarity, in addition to the card type, card ID, rarity, features 1 to m, and other information that constitute the data configuration of the feature storage unit 122. The feature storage unit 322 stores at least features 1 to m of the genuine product for each trading card (each piece of card identification information). If a counterfeit product is obtained, the feature storage unit 322 stores features 1 to m of the counterfeit product.
[0056] Trained model 323 is a trained model trained to increase the difference in feature quantities between genuine and counterfeit products for each piece of card identification information (for each card type, card ID, and rarity). In addition, when counterfeit products are not available, trained model 323 is trained to decrease the difference in feature quantities between genuine and counterfeit products. Trained model 323 may be a trained model trained to increase the difference in feature quantities between genuine and counterfeit products for each piece of card identification information (for each card type, or for each card type and card ID).
[0057] A description will now be given of the authenticity discrimination process of a trading card executed by the discrimination server 300 and the shop terminal 200 configured as described above.
[0058] In the store terminal 200, the scanner 204 reads a trading card (step S501) and generates image data of the trading card. The scanner 204 may read one or more trading cards. The input / output unit 203 accepts input of information related to the read trading card from an input screen displayed on the display device (step S502). The input information related to the trading card includes card identification information. The communication unit 201 transmits the image data of the trading card and the input information to the identification server 100 (step S503).
[0059] In the determination server 300, the communication unit 130 receives the image data of the trading card and the input information from the store terminal 200 (step S504).
[0060] The feature calculation unit 111 inputs image data of the discrimination target card to the trained model 323 (step S505). The feature calculation unit 111 outputs the feature of the image data of the discrimination target card from the trained model 323 (step S506). The distance calculation unit 112 acquires the feature associated with the card type, card ID, rarity, and authenticity, which are stored in the feature storage unit 322. The distance calculation unit 112 calculates the distance between the feature vector, which is a set of the feature of the discrimination target card, and the feature vector, which is a set of the feature acquired from the feature storage unit 322 (step S507).
[0061] The authenticity discrimination unit 313 judges whether the distance of the feature vector meets a predetermined condition (step S508). Specifically, the authenticity discrimination unit 313 judges whether the distance of the feature vector is within a predetermined threshold. If the distance between the feature vector of the discrimination target card and the feature vector of the genuine product is within the predetermined threshold, the card is judged to be genuine. In addition, if the feature amount of a counterfeit product is stored in the feature amount storage unit 322, the distance between the feature vector of the genuine product and the feature vector of the counterfeit product may be calculated, and the card may be judged to be the one having the closer distance between the feature vectors of the genuine product and the counterfeit product.
[0062] If the authenticity discrimination unit 313 determines that the distance of the feature vector meets a predetermined condition (step S508: Yes), it determines that the target card is genuine (step S509). If the authenticity discrimination unit 313 determines that the distance of the feature vector does not meet a predetermined condition (step S508: No), it determines that the card is a counterfeit (step S510). The communication unit 130 transmits authenticity information corresponding to each target card, indicating whether it is genuine or counterfeit, to the store terminal 200 (step S511).
[0063] In the store terminal 200, the communication unit 201 receives the authenticity information associated with each of the cards to be discriminated from the discrimination server 300 (step S512). The input / output unit 203 displays the authenticity information associated with each of the cards to be discriminated on the display screen (step S513).
[0064] In this way, the discrimination server 300 can easily discriminate whether the card to be discriminated is genuine or counterfeit by using the feature amounts of the image data of the trading card. Even if it is difficult to obtain a counterfeit card to serve as training data, the genuineness of the trading card can be discriminated by accumulating the feature amounts of the image data of the genuine card and using the median value of the accumulated feature amounts. Note that in this embodiment, similarly to the first embodiment, instead of calculating the distance using all the feature amounts output from the trained model 323 as a feature vector, the distance may be calculated using selected feature amounts as a feature vector. This embodiment may be executed before or after the first embodiment.
[0065] FIG. 6 is an explanatory diagram showing an example of feature amounts output by the trained model 323. FIG. 6(a) is a diagram showing two feature amounts of image data of a genuine product output by the trained model 323. In FIG. 6, the feature amount is typically shown as two axes, feature amount x and feature amount y, but the feature amount is not limited to two, and may be m-dimensional indicating all feature amounts, or may be one or more feature amounts determined in advance. Although there is no counterfeit product data in FIG. 6(a), if the median value of the feature amount of a genuine product and the feature amount of the card to be discriminated match a predetermined condition (for example, within a predetermined distance), the product can be determined to be genuine.
[0066] Fig. 6(b) is an explanatory diagram showing the feature amounts of genuine products and counterfeit products output by the trained model 323. Fig. 6(b) shows the state before genuine products and counterfeit products are learned, but even at this stage, it is possible to distinguish between genuine and counterfeit products by comparing the feature amount of the trading card to be discriminated with the median value of the feature amount of the genuine product and the distance from the median value of the feature amount of the counterfeit product.
[0067] Fig. 6(c) is an explanatory diagram showing the feature amounts of genuine and counterfeit products output after learning to be able to distinguish between genuine and counterfeit products. By learning to be able to distinguish between genuine and counterfeit products, it is possible to distinguish the genuineness of the target card with higher accuracy than Fig. 6(b).
[0068] FIG. 6(d) is an explanatory diagram showing the feature amounts of genuine and counterfeit products that have been further trained to distinguish between genuine and counterfeit products. The trained model 323 has been trained so that the distance 1 between the feature amounts of genuine and counterfeit products is greater, and the distance 2 between the feature amounts of other genuine products and counterfeit products is closer. This makes it possible to accurately distinguish the genuineness of the card to be distinguished, even when there is little image data of the counterfeit product. The trained model 323 may also be trained to change the combination or number of feature amounts and select feature amounts that make the distance 1 between the feature amounts of genuine and counterfeit products greater, and the distance 2 between the feature amounts of other genuine products and counterfeit products closer.
[0069] The image data of the trading card for calculating the feature amount may be image data of the entire trading card or image data of a predetermined area that is a part of the trading card, as in the first embodiment. The predetermined area may be different for each card type, card ID, and rarity, and may be a plurality of areas. In the above embodiment, the authenticity of each trading card specified by the card type, card ID, and rarity is determined, but the trained model 323 may be trained so that the authenticity can be determined for each card type or for each card type and card ID, and the store terminal 200 may receive an input of the card type or receive an input of the card type and card ID to determine the authenticity of the card type or the authenticity of the card type and card ID. In this case, the feature amount storage unit 322 stores the feature amount for each card type, card type, and card ID when the authenticity can be determined for each card type, card type, and card ID.
[0070] <Example 3> 7 is a block diagram showing a configuration of a trading card discrimination system 30 including a discrimination server 400 and a mobile terminal 500 according to a third embodiment. The discrimination server 400 is a computer that discriminates cards to be discriminated. The discrimination server 400 has almost the same function and configuration as the discrimination server 100 described in the first embodiment. The mobile terminals 500-1 to 500-k (hereinafter referred to as mobile terminals 500) are computers operated by users who make requests for the purchase or the like of trading cards, and are connected to the discrimination server 400 via a network N.
[0071] The discrimination server 400 includes an image adjustment unit 416 in addition to the configuration of the discrimination server 100. The image adjustment unit 416 corrects and adjusts the image so that the image data of one or more sheets becomes similar to the image data read by the scanner 204, based on information related to the image capture when the trading card is captured by the mobile terminal 500 (hereinafter, referred to as imaging information).
[0072] The mobile terminal 500 is a smartphone, a mobile phone, a tablet terminal, a personal computer, or the like, and transmits and receives data to and from the discrimination server 400. The mobile terminal 500 includes a communication unit 501, a control unit 502, an input / output unit 503, and a camera 504. The camera 504 can capture an image of a trading card by being built into the mobile terminal 500 or attached externally.
[0073] A description will now be given of the trading card discrimination process executed by the discrimination server 400 and the mobile terminal 500 configured as described above.
[0074] In the mobile terminal 500, the camera 504 captures an image of a trading card in response to an operation by an operator (step S801). More specifically, the camera 504 captures an image of the trading card to generate image data including the trading card, and acquires image information at the time of capturing the image. The input / output unit 503 displays an image captured by the camera 504 on the display screen, and also displays marks such as frames and symbols that allow the entire trading card or a region of the trading card required for card discrimination to be captured. The input / output unit 503 may determine whether the entire trading card or a predetermined region can be captured according to the marks, and may display instructions for adjusting the position and direction of the camera so that the entire trading card or a predetermined region can be captured. The instructions for adjusting the position and direction of the camera 504 may be voice. The control unit 502 may determine whether the image data of the trading card captured by the camera 504 matches a predetermined condition required for card discrimination, and may instruct the camera to capture the image again if it determines that the condition is not met. Image data of a trading card is not limited to one card, but may be captured of multiple cards with different magnifications, card sizes, inclinations, image brightness, etc. Also, instead of an image, a video may be captured.
[0075] The input / output unit 503 receives input of information about the captured trading card from the input screen displayed on the display device (step S802). The communication unit 501 transmits the image data of the trading card, the captured image information, and the input information to the discrimination server 400 (step S803).
[0076] In the discrimination server 400, the communication unit 130 receives the image data, imaging information, and input information of the trading card transmitted from the mobile terminal 500 (step S804). The image adjustment unit 416 adjusts the image data of the trading card based on the image data and imaging information of the trading card (step S805). When multiple pieces of image data are received, adjustments are made based on the multiple pieces of image data and imaging information so that the image data becomes the image data of one discrimination target card. The processing of steps S806 to S812 is similar to steps S305 to S311 in FIG. 3, so please refer to the explanation of steps S305 to S311 in FIG. 3 and the explanation will be omitted here.
[0077] The communication unit 130 transmits the card identification information and the card information associated with the discrimination target card to the mobile terminal 500 (step S813).
[0078] In the mobile terminal 500, the communication unit 501 receives the card identification information and card information of the card to be discriminated transmitted from the discrimination server 400 (step S814). The input / output unit 503 displays the card identification information and card information of the card to be discriminated on the display screen (step S815).
[0079] As described above, the mobile terminal 500 can identify a trading card by transmitting captured image data of the trading card to the identification server 400, and the card identification information and card information of the trading card can be confirmed on the mobile terminal 500. Furthermore, the identification server 400 has the functions and configuration of the identification server 300, so that the mobile terminal 500 can confirm the authenticity of the trading card instead of or in addition to identifying the trading card.
[0080] <Example 4> In this embodiment, after the trading card authenticity determination process of the second embodiment, when the store terminal 200 decides to purchase a trading card, the process of updating the features stored in the feature memory unit 322 by adding the features of the trading card that has been decided to be purchased will be described.
[0081] 9 is a block diagram showing a configuration of a trading card discrimination system 40 including a discrimination server 600 and a store terminal 200 according to a fourth embodiment. The discrimination server 600 includes a feature updating unit 616 in a control unit 610 and an individual feature storage unit 624 in a storage unit 620 in addition to the configuration of the second embodiment shown in FIG.
[0082] The individual feature storage unit 624 stores the card identification information (card type, card ID, rarity) of each genuine trading card, which has been acquired by a scanner or the like, in association with the features and other information. More specifically, the individual feature storage unit 624 stores the features of the purchased trading cards, which are the features output from the trained model 323 in step S506 in Fig. 5, in association with the card identification information.
[0083] The feature amount updating unit 616 stores the card identification information and feature amounts of the trading card for which the store terminal 200 has accepted an input for purchase in the individual feature amount storage unit 624 in association with each other, calculates the feature amounts of the trading card for each piece of card identification information from the feature amounts stored in the individual feature amount storage unit 624, and stores them in the feature amount storage unit 322. More specifically, the feature amount updating unit 616 determines whether or not a purchase has been input in the store terminal 200 for each trading card for which authenticity information has been sent to the store terminal 200. The feature amount updating unit 616 stores the card identification information and feature amounts of the trading card for which the store terminal 200 has accepted an input for purchase in the individual feature amount storage unit 624, calculates the feature amounts of the trading card for each piece of card identification information from the feature amounts in the individual feature amount storage unit 624 that match the card identification information, and stores them in the feature amount storage unit 322, thereby updating the feature amounts.
[0084] Fig. 10 is a flowchart showing the feature update process executed by the discrimination server 600 and the store terminal 200, following the flowchart of Fig. 5 in the second embodiment. In this embodiment, in the process in the second embodiment (step S506 in Fig. 5), the feature calculation unit 111 associates a set of features of a trading card with card identification information and stores them in a temporary storage unit (not shown). In addition, when the hash value calculation unit 114 calculates a hash value of image data, the hash value is also stored in a temporary storage unit (not shown) in association with the card identification information.
[0085] Returning to Fig. 10, the input / output unit 203 of the store terminal 200 accepts input as to whether or not to buy each trading card displayed on the display screen (step S514). The display screen displays at least information identifying each trading card read by the scanner, authenticity information for each trading card, and a form for inputting whether or not to buy. The communication unit 201 transmits buy information indicating whether or not to buy each trading card to the discrimination server 300 (step S515).
[0086] In the determination server 300, the communication unit 130 receives the trade-in information for each trading card from the store terminal 200 (step S516).
[0087] The feature amount update unit 616 judges whether or not a purchase input has been received for each trading card (step S517). If it is judged that a purchase input has been received for the trading card (step S517: Yes), the feature amount update unit 616 acquires the feature amount associated with the trading card from a temporary storage unit (not shown) (step S518). The feature amount update unit 616 stores the card identification information and the feature amount corresponding to the trading card acquired from the temporary storage unit (not shown) in the individual feature amount storage unit 624 (step S519). The feature amount update unit 616 calculates a new feature amount for each card identification information from the feature amount stored in the individual feature amount storage unit 624 (step S520). The feature amount calculation method is the same as the above-mentioned case where a plurality of image data of genuine trading cards is acquired, in which the central value of the feature amount (average value of each of the plurality of feature amounts, center of gravity of the distribution, median value, mode, etc.) is calculated. The feature amount updating unit 616 stores the calculated feature amount for each card identification information in the feature amount storage unit 322 (step S521). The communication unit 130 transmits information about the trading cards to be purchased to the POS system (step S522). The POS system executes a trading card purchase process. Note that if it is determined that multiple trading cards should be purchased, new feature amounts for each card identification information may be calculated in step S520 after storing the card identification information and feature amounts of all trading cards for which purchase input has been accepted in the individual feature amount storage unit 624.
[0088] In this way, by accepting input for purchase at the store terminal 200, the trading card is judged to be genuine, and new feature values for each card identification information are calculated using feature values added with feature values of the trading card judged to be genuine, and the feature values stored in the feature value storage unit 322 can be updated. This allows feature values based on a larger number of genuine products to be used in determining whether a card is genuine, improving the accuracy of determining whether it is genuine. Also, during the flow of the trading card purchase process at the store, it is possible to obtain feature values of the trading card to be purchased, i.e., the trading card judged to be genuine. This allows the feature values of the genuine trading card to be easily reflected in the feature value storage unit 322, improving the accuracy of determining whether it is genuine.
[0089] In addition, for trading cards for which input has been received at the store terminal 200 indicating that they will not be purchased, the information may be added that the card is a counterfeit, the information may be stored in the individual feature memory unit 624, and the counterfeit feature information may be calculated for each card identification information.
[0090] In the above-mentioned embodiment, the process has been described as continuing from the flowchart in FIG. 5, but similarly, the process may be continuing from the flowchart in FIG. 3 of the first embodiment and the flowchart in FIG. 8 of the third embodiment. In this case, the discrimination server 100 of the first embodiment or the discrimination server 400 of the third embodiment includes a feature amount update unit 616 and an individual feature amount storage unit 624 in addition to the respective configurations. Also, in step S306 of FIG. 3 or step S807 of FIG. 8, the feature amount calculation unit 111 stores the feature amount of the trading card in a temporary storage unit (not shown) in association with the card identification information. Also, when the hash value calculation unit 114 calculates a hash value of the image data, the hash value is also stored in association with the card identification information.
[0091] In this case too, following steps S314 and S312 in FIG. 3, or after step S815 in FIG. 8, a purchase request can be sent to the discrimination server 400 from the mobile terminal 500 operated by the user, and the discrimination server 400 can execute the processing of the flowchart in FIG. 10, thereby updating the feature memory unit 122 with features that include the features of a trading card for which the store terminal 200 or discrimination server 400 has accepted the purchase input, i.e., which has been determined to be a genuine item.
[0092] Furthermore, trained model 123 and trained model 323 may be trained using image data of trading cards and information about trading cards for which input for purchase has been received. This allows training using a larger number of genuine trading cards, and enables trained model 123 and trained model 323 to output features that can more accurately determine the card type, card ID, and rarity of trading cards, as well as the authenticity of trading cards.
[0093] The hardware configuration of the determination server 100, 300, 400, 600, the store terminal 200, and the mobile terminal 500 according to the above-mentioned first to fourth embodiments is a normal computer including one or more processors such as a CPU (Central Processing Unit), an MPU (Micro-Processing Unit), or a GPU (Graphics Processing Unit), and equipped with an external storage device such as a ROM (Read Only Memory), a RAM (Random Access Memory), an HDD (Hard Disk Drive), a flash memory, or an SSD (Solid State Drive), and a communication control device, etc., and the above-mentioned configuration and function are realized by the CPU or the like reading and running the programs stored in the ROM, RAM, HDD, etc. The control unit may be an electronic circuit such as an ASIC (Application Specific Integrated Circuit) or a PLD (Programmable Logic Device).
[0094] The programs that run on the determination servers 100, 300, 400, 600, the store terminal 200, and the mobile terminal 500 may be stored on a computer connected to a network such as the Internet and provided by downloading the programs via the network, or may be recorded in an installable or executable format on a computer-readable recording medium such as a CD-ROM, DVD, USB memory, or SD card and provided. The programs that realize the above-mentioned functions and processes may be provided in the form of an API (Application Programming Interface), SaaS (Software as a Service), or cloud computing.
[0095] In the above-described embodiments, the discrimination servers 100, 300, 400, 600, the store terminal 200, and the mobile terminal 500 have been described as separate devices, but all or part of the functions and configurations of the discrimination servers 100, 300, 400, 600, the store terminal 200, and the mobile terminal 500 may be configured as a single device, or some or all of the functions and configurations of the discrimination servers 100, 300, 400, 600, the store terminal 200, and the mobile terminal 500 may be combined.
[0096] The present invention is not limited to the above-described embodiment, and does not necessarily have to be physically configured as illustrated. The present invention can be configured by functionally or physically dividing, integrating, replacing, modifying, or deleting all or part of the components described in the embodiment in any unit according to various loads, usage conditions, etc. [Explanation of symbols]
[0097] 100... discrimination server, 110... control unit, 111... feature amount calculation unit, 112... distance calculation unit, 113... card discrimination unit, 114... hash value discrimination unit, 115... information acquisition unit, 120... storage unit, 121... card information storage unit, 122... feature amount storage unit, 123... trained model, 130... communication unit, 200... store terminal, 201... communication unit, 202... control unit, 203... input / output unit, 204... scanner, 300... discrimination server, 310... control unit, 313... authenticity discrimination unit, 320... storage unit, 322... feature amount storage unit, 323... trained model, 400... discrimination server, 410... control unit, 416... image adjustment unit, 500... mobile terminal, 501... communication unit, 502... control unit, 503... input / output unit, 504... camera, 600... discrimination server, 616... feature amount update unit, 624... individual feature amount storage unit
Claims
1. A trading card discrimination system, comprising: The discrimination server is a feature storage means for storing card identification information and a set of feature values in association with each other; a card information storage means for storing the card identification information and card information in association with each other; A receiving means for receiving image data of a trading card transmitted from a store terminal connected via a communication network; A feature calculation means for calculating a set of features from the image data received by the receiving means using a trained model that receives image data as input and outputs features; a distance calculation means for calculating a distance between each of the first feature vectors and the second feature vectors based on each of the first feature vectors formed by the set of feature amounts stored in the feature storage means and a second feature vector formed by the set of feature amounts calculated by the feature calculation means; a card discrimination means for discriminating a card based on whether or not each of the distances calculated by the distance calculation means matches a predetermined condition; an information acquiring means for acquiring, when the distance calculated by the distance calculating means matches the predetermined condition, the card identifying information corresponding to the first feature vector that matches the predetermined condition from the feature amount storing means, and acquiring, from the card information storing means, card information associated with the acquired card identifying information; a transmission means for transmitting the card identification information and the card information acquired by the information acquisition means to the store terminal, The store terminal includes: a display means for displaying the card identification information and the card information transmitted from the discrimination server; A trading card discrimination system comprising:
2. 1. A computer-implemented trading card discrimination method, comprising: a feature calculation step of calculating a set of features from image data of a trading card to be discriminated using a trained model that receives image data as input and outputs features; a distance calculation step of calculating a distance between each of the first feature vectors and the second feature vectors based on a first feature vector consisting of a set of feature amounts stored in a feature amount storage unit that stores a set of card identification information and a set of feature amounts in association with each other, and a second feature vector consisting of a set of feature amounts of the image data calculated in the feature amount calculation step; a card discrimination step of discriminating a card based on whether or not each of the distances calculated in the distance calculation step matches a predetermined condition; an information acquisition step of acquiring, when the distance calculated by the distance calculation step meets a predetermined condition, the card identification information corresponding to the first feature vector that meets the predetermined condition from the feature amount storage unit, and acquiring, from a card information storage unit, card information associated with the acquired card identification information; A method for identifying trading cards.
3. 1. A computer-implemented trading card discrimination method, comprising: a feature calculation step of calculating a set of features from image data of a trading card to be discriminated using a trained model that receives image data as input and outputs features; a distance calculation step of calculating angles between the first feature vector and the second feature vector based on a first feature vector consisting of a set of feature amounts stored in a feature amount storage unit that stores a set of card identification information and a set of feature amounts in association with each other, and a second feature vector consisting of a set of feature amounts of the image data calculated in the feature amount calculation step; a card discrimination step of discriminating a card based on whether or not each of the angles calculated in the distance calculation step matches a predetermined condition; an information acquisition step of acquiring, when the angle calculated by the distance calculation step meets a predetermined condition, the card identification information corresponding to the first feature vector that meets the predetermined condition from the feature amount storage unit, and acquiring, from a card information storage unit, card information associated with the acquired card identification information; A method for identifying trading cards.
4. 4. The trading card discrimination method according to claim 2 or claim 3, wherein the feature calculation step inputs image data of a genuine trading card into the trained model, and stores a set of features output from the trained model in the feature memory unit in correspondence with the card identification information.
5. the card information storage unit stores a hash value in association with the card identification information; 4. The trading card discrimination method according to claim 2 or 3, further comprising a hash value discrimination step of calculating a hash value from image data of the trading card to be discriminated, and discriminating whether or not the calculated hash value and the hash value stored in the card information storage unit meet a predetermined condition.
6. 4. The trading card discrimination method according to claim 2, further comprising a transmitting step of transmitting the card identification information and the card information of each of the discriminated trading cards to a POS system.
7. an image receiving step of receiving image data of all or a part of a trading card and imaging information at the time of imaging the image data; and an image adjustment step of adjusting the image data to that of the trading card to be discriminated based on the image data and the imaging information received in the image receiving step, 4. The trading card discrimination method according to claim 2, wherein the imaging information is information for correcting and adjusting an image.
8. A trading card discrimination system, comprising: The discrimination server is a feature storage means for storing card identification information and a set of feature values in association with each other; a card information storage means for storing the card identification information and card information in association with each other; a receiving means for receiving the image data of the trading card and the card identification information transmitted from a store terminal connected via a communication network; A feature calculation means for calculating a set of features from the image data received by the receiving means using a trained model that receives image data as input and outputs features; a distance calculation means for calculating a distance between each of the first feature vectors and the second feature vectors based on a first feature vector formed of a set of feature amounts associated with the card identification information stored in the feature storage means and a second feature vector formed of the set of feature amounts calculated by the feature calculation means; a card authenticity determination means for determining whether the image data is authentic or counterfeit based on whether the distance calculated by the distance calculation means matches a predetermined condition; a transmission means for transmitting the authenticity result determined by the card authenticity determination means to the store terminal, The store terminal includes: A display means for displaying the authenticity result transmitted from the determination server; A trading card discrimination system comprising:
9. 1. A computer-implemented trading card discrimination method, comprising: A feature calculation step of calculating feature amounts of image data of a trading card to be discriminated using a trained model that receives image data and outputs feature amounts; a distance calculation step of calculating a distance between each of the first feature vectors and the second feature vectors based on a first feature vector consisting of a set of feature amounts associated with the card identification information of the image data, the first feature vectors being stored in a feature amount storage unit that stores sets of card identification information and feature amounts in association with each other, and a second feature vector consisting of the set of feature amounts calculated in the feature amount calculation step; a card authenticity determination step of determining whether the image data is authentic or counterfeit based on whether the distance calculated by the distance calculation step matches a predetermined condition; A method for identifying trading cards.
10. 1. A computer-implemented trading card discrimination method, comprising: A feature calculation step of calculating feature amounts of image data of a trading card to be discriminated using a trained model that receives image data and outputs feature amounts; a distance calculation step of calculating angles between the first feature vector and the second feature vector based on each first feature vector composed of a set of feature amounts associated with the card identification information of the image data, which is stored in a feature amount storage unit that stores sets of card identification information and feature amounts in association with each other, and a second feature vector composed of the set of feature amounts calculated in the feature amount calculation step; a card authenticity determination step of determining whether the image data is authentic or counterfeit based on whether the angle calculated by the distance calculation step matches a predetermined condition; A method for identifying trading cards.
11. the feature amount storage unit stores authenticity information indicating whether the card is genuine or counterfeit in association with the card identification information; the distance calculation step calculates a first distance between the second feature vector and the third feature vector based on the second feature vector and a third feature vector consisting of a set of features of a genuine product that are associated with the card identification information of the image data and stored in the feature storage unit; calculating a second distance between the second feature vector and the fourth feature vector based on the second feature vector and a fourth feature vector consisting of a set of features of a counterfeit product that are associated with the card identification information of the image data and stored in the feature storage unit; 11. The trading card discriminating method according to claim 9, wherein the card authenticity discriminating step discriminates authenticity of the image data based on the first distance and the second distance calculated in the distance calculating step.
12. the distance calculation step includes calculating a first angle between the second feature vector and the third feature vector based on the second feature vector and the third feature vector; calculating a second angle between the second feature vector and the fourth feature vector based on the second feature vector and the fourth feature vector; 12. The trading card discriminating method according to claim 11, wherein said card authenticity discriminating step discriminates authenticity of said image data based on said first angle and said second angle calculated in said distance calculating step.
13. 12. The trading card discrimination method according to claim 11, further comprising the steps of: generating a trained model that is trained to increase a distance between a feature vector formed of a set of features of the image data of the genuine product and a feature vector formed of a set of features of the image data of the counterfeit product, using the image data of the genuine product and the image data of the counterfeit product as training data.
14. The discrimination server, an individual feature storage means for storing card identification information of each trading card and a set of features in association with each other; a feature updating means for storing a set of card identification information and feature amounts of a trading card for which a trade-card purchase is input in the individual feature storing means, and updating the feature amount for each of the card identification information stored in the feature storing means based on the set of feature amounts associated with the card identification information stored in the individual feature storing means; 9. The trading card discrimination system according to claim 1, further comprising:
15. an individual feature amount storage unit that stores card identification information and a set of feature amounts for each trading card in association with each other; a feature updating step of storing a set of card identification information and feature amounts of a trading card for which a trade-in is made, in the individual feature amount storage unit, and updating the feature amount for each of the card identification information stored in the feature amount storage unit based on the set of feature amounts associated with the card identification information stored in the individual feature amount storage unit; 11. The trading discrimination method according to claim 2, 3, 9, or 10, comprising:
16. A trading card discrimination program for causing a computer to execute the trading card discrimination method according to any one of claims 2 to 7, 9 to 13 and 15.
Citation Information
Patent Citations
Information processor and program
JP2018160184A
Trading card purchase assessment system
JP2019082870A
Image identification device, image identification method, and image identification program
JP2020035097A
Image processing device, image processing method, image processing program, program, and mobile terminal device with camera
WO2014092193A1
