Online trading system for identifying counterfeit goods

Abstract

An online trading system for identifying counterfeit goods is provided. The sale of counterfeit items has always been a problem in marketplaces, especially in auction houses. When sellers submit items for sale in marketplaces, there is a need for automated detection of counterfeit items. The disclosed technology receives information about the items to be traded. Information associated with the item includes metadata related to the item and an image of the item. The method uses a database of items known not to be sold (e.g., stock photos) or counterfeit items. The method matches the item's data with data in the database and identifies the item as counterfeit based on the matching result. The matching operation includes data analysis (e.g., image analysis using features of image data). Using the disclosed technology, counterfeit items can be detected automatically and efficiently when sellers submit items for sale in marketplaces, thereby improving the reliability of marketplaces from the buyer's perspective.

Description

[0001] Cross-references

[0002] This patent application claims priority to U.S. Patent Application No. 17 / 243,516, filed April 28, 2021, entitled “AN ONLINETRANSACTION SYSTEM FOR IDENTIFYING COUNTERFEITS”, which has been assigned to the assignee of this application and whose entire contents are expressly incorporated herein by reference. Technical Field

[0003] This disclosure relates to an online trading system, and more particularly to an online trading system for identifying counterfeit goods. Background Technology

[0004] Traditional marketplace systems facilitate the sale of items that sellers wish to sell to buyers. For example, auction systems receive items from sellers and place them in the bidding process. The sale of counterfeit items has become a problem when sellers knowingly or unknowingly place counterfeit items on the marketplace and buyers purchase them. Identifying counterfeit items involves visually comparing submitted items to a standard list used to determine whether an item is genuine or counterfeit. Therefore, identifying counterfeit items has always been a time-consuming task. In practice, buyers need to determine whether an item is counterfeit before purchasing. Buyers attempt to identify counterfeit items based on the descriptions of items posted on the system. Preventing the sale of counterfeit items improves the effective use of the system because more sellers and buyers have confidence in using it. Tracking and reversing the sale of counterfeit items after a transaction occurs increases operational costs. Therefore, developing a technology that can better meet the needs while minimizing trade-offs is desirable.

[0005] It is with regard to these and other general considerations that the aspects disclosed herein have been made. While relatively specific issues may be discussed, it should be understood that the examples are not limited to solving specific problems identified in the context of this disclosure or elsewhere. Summary of the Invention

[0006] According to this disclosure, the above and other problems are solved by using image analysis based on machine learning models to identify counterfeit items.

[0007] This disclosure relates to identifying counterfeit items in a sales transaction system. Specifically, an online shopping server receives items along with image data of the items for sale. An item matcher matches the image data of the items against a set of images in a database. The database stores image data of items belonging to the same category as the received item (e.g., stock photos). Based on the matched image data, a counterfeit item identifier identifies the received item as counterfeit. Additionally or alternatively, a transaction blocker disables the sale of the received item because the received item has been determined to be counterfeit.

[0008] The disclosed technology includes a machine learning model for predicting whether a received item is counterfeit. The machine learning model includes a neural network with trained parameters for predicting a match between the item and inventory image data. The machine learning model uses a probability distribution of the likelihood of an item matching one or more inventory image data sets based on features of the image data. Additionally or alternatively, the machine learning model includes a converter comprising an encoder and a decoder. The disclosed technology also includes disabling transactions for items already identified as counterfeit and sending warnings about counterfeit items.

[0009] The summary is provided to introduce, in simplified form, the selection of concepts further described below in the detailed description. The summary is not intended to identify key or essential features of the claimed subject matter, nor is it intended to limit the scope of the claimed subject matter. Additional aspects, features, and / or advantages of the examples will be set forth in part in the description which follows, and will be apparent in part from that description, or may be learned by practice of this disclosure. Attached Figure Description

[0010] The following figures illustrate examples of non-restrictive and non-exhaustive properties.

[0011] Figure 1 An overview of an example system for identifying articles as counterfeit is shown, according to aspects of this disclosure.

[0012] Figures 2A to 2B An example of a model based on aspects of this disclosure is shown.

[0013] Figure 3 An example of a data structure for item information according to aspects of this disclosure is shown.

[0014] Figure 4 An example of a data structure for seller information according to aspects of this disclosure is shown.

[0015] Figure 5 An example of a method for identifying counterfeit articles according to aspects of this disclosure is shown.

[0016] Figure 6 This is a block diagram illustrating an example physical component of a computing device that can be used to practice aspects of this disclosure. Detailed Implementation

[0017] Various aspects of this disclosure are described more fully below with reference to the accompanying drawings, which are taken from a part of this disclosure and illustrate specific example aspects. However, different aspects of this disclosure may be implemented in many different ways and should not be construed as limited to the aspects set forth herein; rather, these aspects are provided so that this disclosure will be comprehensive and complete, and will fully convey the scope of these aspects to those skilled in the art. The aspects may be practiced as methods, systems, or apparatuses. Thus, aspects may take the form of hardware implementations, entirely software implementations, or implementations combining software and hardware aspects. Therefore, the following detailed description should not be considered limiting.

[0018] Online shopping systems, especially online auction systems, rely on sellers to sell genuine, non-counterfeit items or to help buyers understand the risk that an item might be counterfeit. Traditional online shopping sites offer policy-based enforcement that prohibits the sale of counterfeit items. However, traditional systems merely provide sellers with an interface to interactively declare, for example, by clicking or using checkboxes, that an item is not counterfeit before placing it for sale. Traditional systems cannot automatically determine whether an item is counterfeit. The sale of counterfeit items can negatively impact not only the seller of the item but also the reputation of the online shopping service using the system. In some respects, sellers are unaware of when an item becomes counterfeit.

[0019] Problems arise when online shopping sites need to identify and prevent counterfeit items from appearing in the listings of tens of millions of items on their sites. There may even be tens of millions more counterfeit items that need to be checked. Therefore, there is a trade-off between protecting online marketplaces from counterfeit goods and the cost of doing so. This application addresses this trade-off by providing a system that automatically identifies counterfeit items using a machine learning model with image analysis capabilities. This application is able to balance the competing needs of market security and efficiency through the use of an automatic counterfeit goods identifier.

[0020] As discussed in more detail below, this disclosure relates to a machine learning counterfeit detection model for identifying counterfeit goods in online shopping marketplaces. Specifically, the counterfeit detection model analyzes a set of image data of an item to predict whether the item is counterfeit.

[0021] Figure 1An overview of an example system 100 for identifying counterfeit goods is shown. System 100 represents a system for identifying counterfeit goods using image analysis based on image data of the goods, employing counterfeit goods identification models (or multiple counterfeit goods identification models). System 100 includes a client device 102, an application server 110, an online shopping server 120, a network 150, and an image data server 160. The client device 102 communicates with the application server 110, which includes one or more sets of instructions executed on the client device 102 as an application. The application server 110 includes an online shopping application 112 (i.e., an application). One or more sets of instructions in the application server 110 can provide an interactive user interface (not shown) via an interactive interface 104.

[0022] The online shopping server 120 includes an item receiver 122, an item matcher 124, a counterfeit item identifier 126, a transaction disabler 128, a model trainer 140, training data 142, and a training model 144. A network 150 provides network connectivity between the client device 102, the application server 110, the online shopping server 120, and the image data server 160. The image data server 160 includes an image database 162. The image database 162 stores image data of the items. The image data server 160 stores image data via the network 150 and provides it to the online shopping server 120.

[0023] Client device 102 connects to application server 110 via network 150 to execute an application including user interaction through interactive interface 104. Application server 110 interacts with client device 102 and online shopping server 120 via network 150 to perform online shopping as a seller or buyer of items.

[0024] Client device 102 is a general-purpose computer device that provides user input capabilities (e.g., online shopping via network 150 through interactive interface 104). In some aspects, client device 102 may optionally receive user input from sellers of items. Sellers upload information about items for sale in an online shopping marketplace. This information includes image data of the item, a brief description of the item, price information, quantity information, etc. For example, interactive interface 104 may present a graphical user interface associated with a web browser. In some aspects, client device 102 may communicate with application server 110 via network 150.

[0025] Application server 110 is a server that enables sellers (who can post items for sale) and buyers (who purchase items) to interactively use system 100 on client device 102. Application server 110 may include an application containing online shopping application 112. Online shopping application 112 can provide the presentation of items for users to purchase.

[0026] In some respects, the online shopping application 112 may connect to the item receiver 122 of the online shopping server 120 to post information about items for sale on an online shopping site (not shown). The information about the item may include its name, a brief description, quantity, price, and one or more image data depicting the item. Additionally or alternatively, the information about the item may include category information. For example, the item may be a laptop computer. The one or more images may include photographs of the laptop computer in different views and close-ups of a product information label with the item's product code and serial number. When the online shopping server 120 successfully receives information about the item and confirms that the item is not counterfeit, the online shopping application 112 may receive confirmation from the online shopping server 120. When the online shopping server 120 determines that the item is counterfeit, the transaction disabler 128 may remove the item from the list of items for sale on the online shopping site.

[0027] Online shopping server 120 represents an application / system for identifying counterfeit items. Online shopping server 120 controls the online shopping site by removing counterfeit items from a buyer's list of items for sale. Online shopping server 120 uses a counterfeit detection model to match items with inventory image data and determines that the item is counterfeit. Item receiver 122 receives information about items from the online shopping application 112 used by the seller via an interactive interface 104 on client device 102. This information includes the item's name, brief description, and image data. Item matcher 124 uses the counterfeit detection model to match the item's image data with a set of image data for various items. The counterfeit detection model (e.g., training model 144) is a trained model used to predict whether an input item is counterfeit by analyzing the item's image data and generating a probability distribution of the likelihood that the item's image data matches one or more images of counterfeit items.

[0028] In some aspects, analyzing image data may include pixel-level comparisons of image data of the item with one or more images of a counterfeit item. Additionally or alternatively, analyzing image data may include extracting features from the image data and comparing those features with known features of one or more images of a counterfeit item. The features used for comparison in the image data may depend on the category of the item. For example, when the category of the item is electronic equipment, the features of the image data of the item may include a photograph of a product label with a product serial number. In another example, when the category of the item is an accessory package, the feature may include a close-up photograph or printed pattern located on the side of the accessory package.

[0029] In some respects, the use of counterfeit identification models based on machine learning neural networks and / or encoder-decoder models enables efficient matching of image data about an item with a set of image data associated with genuine counterfeit items. For example, the use of machine learning models can include pattern matching based on feature vectors associated with predefined portions of the image data. The category of the item can determine the predefined portions used for matching. For example, when the item is an accessory kit, the model can be trained to focus on comparing printed logo patterns on the side view of the item. The effectiveness of training the model becomes considerable when there are tens of millions of items and even more image data to compare. Furthermore, parts of the predictive processing can be performed simultaneously to improve the performance of the matching process. For example, the model can simultaneously perform pattern matching predictions for different portions of the image data. In some respects, parallel processing of matching can be performed in real time when sellers submit information about items for sale on online shopping sites.

[0030] Counterfeit Detector 126 uses the output from Item Matcher 124 to determine that an item is counterfeit. In some aspects, Counterfeit Detector 126 can notify the seller by sending the determination result to Online Shopping Application 112. In other aspects, Counterfeit Detector 126 sends the determination and information about the item to Transaction Disabler 128. Transaction Disabler 128 disables transactions associated with the item on the Online Shopping Site. For example, when an item is already listed on the Online Shopping Site, Transaction Disabler 128 can disable the sale of that item. Additionally or alternatively, Transaction Disabler 128 can disable listings of items that the seller has previously posted on the Online Shopping Site for sale. In yet another aspect, Transaction Disabler 128 can notify buyers who have previously purchased the item from the seller and warn them about potentially problematic activity.

[0031] Model trainer 140 represents an application / system for generating a counterfeit identification model and training that model for deployment as training model 144 in online shopping server 120. In some aspects, model trainer 140 retrieves a set of image data from image database 162 of image data server 160.

[0032] Image data server 160 includes image database 162. Image database 162 stores a collection of image data and metadata associated with one or more image data sets. For example, image database 162 may store image data of items that have been previously identified as counterfeit. Additionally or alternatively, image database 162 may store image data of items that have been previously identified as stolen. The image data server can provide an interface for registering and maintaining image data associated with counterfeit and / or stolen items.

[0033] As will be understood, regarding Figure 1 The various methods, devices, applications, features, etc., described are not intended to limit System 100 to being performed by the specific applications and features described. Therefore, additional controller configurations may be used to practice the methods and systems described herein and / or the features and applications described may be excluded without departing from the methods and systems disclosed herein.

[0034] Figures 2A to 2B An example of a counterfeit identification model according to aspects of this disclosure is shown. Figure 2A In the example, system 200A illustrates a counterfeit identification model 202A using a converter-based encoder-decoder model. Counterfeit identification model 202A includes a feature extractor 204, an encoder 206, and a decoder 208. Counterfeit identification model 202A receives information about an item. This information may include the item's name, brief description, price, and image data. The image data about the item may include one or more photographs of the item. The information about the item may include a description of the corresponding image data. For example, a photograph could be a front view of a laptop computer. Another photograph could be a close-up view of a product label with the item's serial number. In some other respects, the image data may not include a description provided by the seller.

[0035] Feature extractor 204 identifies and extracts features from image data about the object. Features of the image data may include edge information and color. In some aspects, features may include metadata about the image data. For example, metadata may include the location where the photo was taken, camera or scanner settings, and the date / time the photo was taken. Metadata may be part of an image file associated with information about the object.

[0036] Encoder 206 encodes image data and features into a multidimensional vector as encoded data. The multidimensional vector can represent each encoded feature of the image data. Decoder 208 decodes the encoded image data. Decoder 208 generates an output that includes probability distribution data. In some aspects, the counterfeit identification model 202A is based on a converter model. The probability distribution data indicates the likelihood that an item matches one or more image data representing a genuine counterfeit item. Additionally or alternatively, the counterfeit identification model 202A can be fine-tuned from the output of decoder 208 to identify counterfeits for a task-specific dataset representing items of a predetermined category. For example, items of a predetermined category may include handbag accessories, shoes, and vintage jeans. Each predetermined category may specify specific parts of the corresponding item for comparison in finer details than other parts. For example, texture, top button, rivets, belt tags, pagination labels, average market price, etc.

[0037] exist Figure 2B In this system, the counterfeit identification model 202B uses a neural network as a matching predictor. System 200B includes the counterfeit identification model 202B. The counterfeit identification model 202B includes an embedding generator 210, a matching predictor 212 (a trained neural network), and a matching provider 214. Similar to what has been described herein, the counterfeit identification model 202B receives information about an item. This information may include the item's name, a brief description, and image data. The image data about the item may include one or more photographs of the item. The information about the item may include a description of the corresponding image data. For example, a photograph could be a front view of a laptop computer. Another photograph could be a close-up view of a product label with the item's serial number. In some other respects, the image data may not include a description provided by the seller.

[0038] Embedding generator 210 generates embeddings of received information about the item, including image data. In some aspects, the embedding includes a multidimensional vector mapping of the image data. The embedding helps to identify the similarity between portions of the image data and corresponding portions of the trained image. Additionally or alternatively, the embedding includes a multidimensional representation of metadata and other aspects of the information about the item. The embedding of metadata can help to efficiently determine the category of the image data by processing vectors.

[0039] Match predictor 212 receives and generates predictions of matching items. Match predictor 212 may be a neural network used to generate a probability distribution of the likelihood that image data of an item matches one or more images associated with a counterfeit item. In some aspects, the neural network is a multi-layer convolutional neural network. In some other aspects, the neural network is a multi-layer recurrent neural network. Match predictor 212 outputs the most likely counterfeit item based on the matching image data and other information about the item, as well as a confidence value expressed as a percentage. For example, the output may indicate a counterfeit item X with a confidence (or probability) value of 65%.

[0040] Match provider 214 provides output from match predictor 212. In some respects, the match provider can provide the probability, expressed as a percentage, that an item is counterfeit, along with a description of possible counterfeit items. For example, the match provider can provide that item X has a 75% probability of matching counterfeit item Z. Based on this output, a counterfeit determiner (e.g., Figure 1 The counterfeit determiner 126 in the code can determine that item X is counterfeit. In some other aspects, the match provider can provide an output indicating a 25% probability that item Y matches the most likely counterfeit item Z2. Based on this output, the counterfeit determiner can determine that the confidence value is less than a predetermined threshold (e.g., 60%). Therefore, the counterfeit determiner can determine that item Y is not counterfeit.

[0041] Figure 3 An example of a data structure for information about an item according to aspects of this disclosure is shown. Figure 3 Item table 300 is shown. Item table 300 includes item ID 302, status 304, item category 306, matched counterfeit item ID 308, and seller ID 310. Item ID 302 represents the identifier of the item submitted by the seller for a sales transaction on the online shopping site. Status 304 is the status of the item. In some aspects, the status of the item includes a counterfeit identifier (e.g., Figure 1 The counterfeit identifier 126) determines whether the item has been identified as counterfeit or non-saleable. In some respects, the item may be non-saleable when the item matcher matches the item as stolen. Non-saleable items can include both counterfeit and stolen items.

[0042] For example, item 5001, a laptop computer, is valid on the online shopping site. There are no matching counterfeit items for item 5001. Seller 1 has submitted the item for sale. Item 5002, an accessory kit, is valid on the online shopping site. There are no matching counterfeit items for this item. Seller 2 has submitted the item for sale.

[0043] Item 5003 has been suspended on the online shopping site. The counterfeit detector has identified the item as counterfeit, matching counterfeit item 8000. Matched counterfeit item 8000 represents a genuine counterfeit. Seller 2 has submitted items for sale. In some respects, the model trainer receives image data associated with genuine counterfeit or stolen items as training data from an image data server (e.g., image database 162 in image data server 160).

[0044] Item 5004 has been suspended on online shopping sites because of item matchers (e.g., Figure 1 Item matcher 124 in the system matches the item with the stolen item (item 8002). Seller 2 submits the item. Item 5005, accessory pack, is valid on the online shopping site. There is no matching counterfeit item for this item. Seller 2 submits the item.

[0045] Figure 4 An example of maintaining a seller's data structure according to aspects of this disclosure is shown. Figure 4 Seller table 400 is shown. Seller table 400 includes seller ID 402, seller status 404, items for sale on the online shopping site 406, and counterfeit items 408. The seller ID identifies each seller as registered with the online shopping site. Seller status 404 indicates the seller's status. For example, seller status indicates whether the online shopping site allows the seller to participate in sales transactions and / or post items for sale. When a seller is not allowed to participate in transactions due to submitting counterfeit items, the seller status can include "suspended". Items on online shopping site 406 include item IDs of items from sellers on the online shopping site. Counterfeit items 408 include one or more items submitted by a seller that have been identified as counterfeit by a counterfeit item determiner.

[0046] For example, Seller 1 is effectively participating in transactions on an online shopping site. Seller 1 has item 5001 available for sale. Seller 1 does not have any items identified as counterfeit by the counterfeit item detector. Seller 2 has been suspended from participating in transactions on the online shopping site. In some respects, submitting an item identified as counterfeit by the counterfeit item detector results in Seller 2's seller status being suspended. Suspension may be one of the measures taken by the online shopping site to prevent the sale of illegal or suspicious items. Seller 1 has posted two items on the online shopping site: item 5002 and item 5005. Seller 1 has two items identified as counterfeit by the counterfeit item detector: item 5003 and item 5004. Additionally or alternatively, Seller Table 400 may include a list of items that have been identified as stolen. In some respects, when a seller submits stolen items for sale, the seller's status may become suspended.

[0047] Figure 5This is an example of a method for identifying an item as non-saleable (e.g., counterfeit, stolen item) according to aspects of this disclosure. The general sequence of operations in method 500 is as follows: Figure 5 As shown in the diagram. Typically, method 500 begins with start operation 502 and ends with end operation 518. Method 500 may include more or fewer steps, or may be combined with... Figure 5 The different sequential arrangement steps shown are in order. Method 500 can be executed as a set of computer-executable instructions executed by a computer system and encoded or stored on a computer-readable medium. Furthermore, method 500 can be executed by gates or circuits associated with a processor, ASIC, FPGA, SOC, or other hardware device. References will be incorporated herein by reference. Figure 1 , Figure 2A , Figure 2B , Figure 3 , Figure 4 and Figure 6 Method 500 is described by describing the system, components, devices, modules, software, data structures, data characteristic representations, signaling diagrams, and methods.

[0048] After initiating operation 502, method 500 begins with receiving operation 504, which receives the item via an online shopping site. In some aspects, the seller uses an online shopping application as the seller and uploads information about the item to the online shopping server. The information about the item may include the item's name, a brief description, category, and one or more image data representing the item. For example, the one or more image data may be a collection of photographs of the item from different views. The one or more image data may include close-up photographs of product labels with product codes and serial numbers.

[0049] The identification operation 506 identifies one or more features of an item from information about the item. The identification operation 506 uses category information, a brief description, and one or more image data to identify features or characteristics of the item. For example, the identification operation 506 may identify that the item belongs to the category of accessory bags with handles and is brown. In some aspects, the identification operation 506 may analyze image data to determine features. These features may include logos printed on the item and damage to parts of the item.

[0050] Matching operation 508 matches the item with one or more items in the item set. For example, matching operation 508 can use a counterfeit detection model that has been trained with real counterfeit items. The counterfeit detection model can use a machine learning model. Examples of machine learning models can include encoder-decoder models and / or deep neural networks. In some aspects, the trained counterfeit detection model takes information about the item as input and generates a probability distribution indicating the likelihood that the item is a counterfeit based on matching with image data of real counterfeit items. For example, matching operation 508 provides three real counterfeit items as matching results, and the highest probability (or confidence level of the match) among the three is 45%. In another example, matching operation 508 provides a real counterfeit item with a 90% probability. Additionally or alternatively, to predict counterfeit items, the model can be trained using image data about genuinely stolen items and output whether the item matches one or more stolen items.

[0051] In some aspects, the matching operation 508 may use a non-sale item identification model for matching the item with one or more items in the non-sale item set. Non-sale items may include one or both counterfeit and stolen items. In some aspects, stolen items may be genuine rather than counterfeit, but not sold on the online shopping site. The non-sale item identification model may be trained using real non-sale items as training data. In some other aspects, the matching operation 508 may use a stolen item identification model for matching the item with one or more stolen items. The stolen item identification model may be trained using a real set of stolen items as training data.

[0052] The determination operation 510 identifies the item as non-saleable (e.g., counterfeit and stolen items). In some aspects, the determination operation 510 compares the probability distributions that are the output of the matching operation. When the probability distribution data indicates a probability higher than a predetermined threshold, the determination operation 510 identifies the item as non-saleable (e.g., counterfeit). For example, when the highest value of the probability distribution data that is the output of the model is higher than 80%, the determination operation 510 identifies the item as non-saleable (e.g., counterfeit).

[0053] Operation 512 identifies the seller of the item. For example, Figure 3 Seller 2 submitted item 5003, which has been identified as counterfeit by the counterfeit detector. In some respects, sellers of items include sellers who have already submitted items for sale. When an item is already listed on an online shopping site, the online shopping server can suspend the listing of that item to prevent the sale of items that are highly likely to be counterfeit.

[0054] Update operation 514 updates the seller's transaction status (e.g., license). When a seller submits an item that has been identified as counterfeit by the counterfeit detector, the online shopping server can suspend the seller's participation in further transactions.

[0055] Warning action 516 alerts the seller to a confirmed status that the item is counterfeit. Additionally or alternatively, warning action 516 alerts the seller to an updated status (e.g., pause) based on the confirmed counterfeit item. In some respects, warning action 516 can add information about the item to the training data of a counterfeit detection model for future training of that model.

[0056] Figure 6 A simplified block diagram of an apparatus that can be used to practice aspects of this disclosure is shown. One or more of the embodiments described herein may be implemented in operating environment 600. This is merely one example of a suitable computing environment and is not intended to imply any limitation on the scope of functionality or use. Other well-known computing systems, environments, and / or configurations that may be suitable for use include, but are not limited to, personal computers, server computers, handheld or laptop devices, multiprocessor systems, microprocessor-based systems, programmable consumer electronic devices such as smartphones, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the aforementioned systems or devices, etc.

[0057] In its most basic configuration, the operating environment 600 typically includes at least one processing unit 602 and memory 604. Depending on the exact configuration and type of computing device, the memory 604 (as described herein, executing instructions for identifying non-tradable items) can be volatile (e.g., RAM), non-volatile (e.g., ROM, flash memory, etc.), or some combination of both. This most basic configuration in... Figure 6 The image is shown by dashed line 606. Furthermore, the operating environment 600 may also include storage devices (removable, 608, and / or non-removable, 610), including but not limited to disks, optical discs, or magnetic tapes. Similarly, the operating environment 600 may also have input devices 614 such as a keyboard, mouse, pen, voice input, onboard sensors, etc., and / or output devices 616 such as a display, speaker, printer, motor, etc. The environment may also include one or more communication connections 612, such as LAN, WAN, near-field communication networks, point-to-point, etc.

[0058] Operating environment 600 typically includes at least some form of computer-readable medium. Computer-readable medium can be any available medium accessible by at least one processing unit 602 or other devices including the operating environment. By way of example and not limitation, computer-readable medium can include computer storage media and communication media. Computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information such as computer-readable instructions, data structures, program modules, or other data. Computer storage media includes: RAM, ROM, EEPROM, flash memory or other memory technologies, CDROM, Digital Universal Optical Disc (DVD) or other optical disc storage devices, magnetic tape cassettes, magnetic tape, disk storage devices or other magnetic storage devices, or any other tangible, non-transitory medium that can be used to store desired information. Computer storage media does not include communication media. Computer storage media does not include carrier waves or other propagated or modulated data signals.

[0059] Communication media embody computer-readable instructions, data structures, program modules, or other data in the form of modulated data signals (such as carrier waves or other transmission mechanisms), and include any information transmission medium. The term "modulated data signal" means a signal having one or more characteristics set or altered in a manner that encodes information in the signal. By way of example and not limitation, communication media include wired media such as wired networks or direct wired connections, and wireless media such as acoustic, RF, infrared, and other wireless media.

[0060] Operating environment 600 can be a single computer operating in a network environment using logical connections to one or more remote computers. Remote computers can be personal computers, servers, routers, network PCs, peer-to-peer devices, or other public network nodes, and typically include many or all of the aforementioned elements as well as other unmentioned elements. Logical connections can include any method supported by available communication media. This type of network environment is common in offices, enterprise-wide computer networks, intranets, and the Internet.

[0061] The descriptions and illustrations of one or more aspects provided in this application are not intended to limit or restrict the scope of the claimed disclosure in any way. The aspects, examples, and details provided in this application are considered sufficient to convey ownership and enable others to make and use the claimed disclosure as the best mode. The claimed disclosure should not be construed as limited to any aspect, such as, or the details provided in this application. Whether shown and described in combination or separately, various features (structural and methodological) are intended to be selectively included or omitted to produce embodiments with a particular set of features. Having provided the descriptions and illustrations of this application, those skilled in the art can contemplate variations, modifications, and alternatives falling within the spirit of the broader aspects of the overall inventive concept embodied in this application without departing from the broader scope of the claimed disclosure.

[0062] The descriptions and illustrations of one or more aspects provided in this application are not intended to limit or restrict the scope of the claimed disclosure in any way. The aspects, examples, and details provided in this application are considered sufficient to convey ownership and enable others to make and use the claimed disclosure as the best mode. The claimed disclosure should not be construed as limited to any aspect, such as, or the details provided in this application. Whether shown and described in combination or separately, various features (structural and methodological) are intended to be selectively included or omitted to produce embodiments with a particular set of features. Having provided the descriptions and illustrations of this application, those skilled in the art can contemplate variations, modifications, and alternatives falling within the spirit of the broader aspects of the overall inventive concept embodied in this application without departing from the broader scope of the claimed disclosure.

[0063] This disclosure relates to systems and methods for identifying an item as counterfeit based at least according to the examples provided in the following sections. The method includes receiving information about an item associated with a transaction, wherein the information includes first data associated with the item; matching the first data with at least second data in a dataset based on a counterfeit identification model, wherein the dataset is associated with a set of counterfeit items in a database, wherein the counterfeit identification model includes a set of encoders and decoders, each trained using a training dataset of counterfeit items in the database, and wherein the counterfeit identification model generates a probability distribution of the likelihood that the item is counterfeit; identifying the item as counterfeit based on the matching of the first data with the at least second data; and providing a warning associated with the item being counterfeit, causing a suspension of the transaction associated with the item. The first data includes image data, and the method further includes determining one or more features of the image data based on the category of the item; and performing a pixel-level comparison of at least a portion of the image data with a set of stock photos associated with the set of counterfeit items based on the one or more features of the image data. The method further includes determining one or more features of the first data, wherein the one or more features include one or more of the following: the type of the item, one or more portions of the first data associated with the item, the color of the item, or the storage conditions of the item; matching the first data with at least one item represented in the dataset using at least one of a trained neural network or a trained converter model; identifying the seller of the item based on a probability distribution associated with the match; and suspending the transaction of the item and one or more bidding processes associated with the seller. The transaction includes online bidding, and wherein the set of items includes multiple counterfeit items. The method further includes receiving item information associated with the transaction from the seller; and suspending the seller's participation in the transaction of the item based on the identification that the item is counterfeit. The counterfeit identification model includes a trained encoder-decoder model, and wherein the counterfeit identification model includes a converter model fine-tuned to one or more predetermined categories of items for counterfeit item identification. Image analysis includes one or more of the following: extracting features of the first data, using a probability distribution to predict whether the first data matches one or more pieces of real counterfeit data, and identifying the item as counterfeit based on the probability distribution.

[0064] Another aspect of this technology relates to a system for identifying items that are not for sale. The system includes a processor; and a memory storing computer-executable instructions that, when executed by the processor, cause the system to: receive information about an item associated with a transaction, wherein the information includes first data associated with the item; match the first data with at least second data in a dataset based on an item identification model, wherein the dataset is associated with a set of items not for sale in a database, wherein the item identification model includes a set of encoders and decoders, each trained using a training dataset of items not for sale in the database, and wherein the item identification model generates a probability distribution of the likelihood that the item is not for sale; identify the item as not for sale based on the matching of the first data with the at least second data; and provide a warning associated with the item being not for sale, causing a suspension of the transaction associated with the item. The first data includes image data, and the computer-executable instructions, when executed, further cause the system to: determine one or more features of the image data based on the category of the item; and perform a pixel-level comparison of at least a portion of the image data with a set of stock photos associated with the non-sale items set, based on the one or more features of the image data, wherein the non-sale items set includes at least one of counterfeit or stolen items. The computer-executable instructions, when executed, further cause the system to: determine one or more features of the first data, wherein the one or more features include one or more of the following: the type of the item, one or more portions of the first data associated with the item, the color of the item, or the storage condition of the item; match the first data with at least one item represented in the dataset using at least one of a trained neural network or a trained converter model; identify the seller of the item based on a probability distribution associated with the match; and suspend the transaction of the item and one or more bidding processes associated with the seller. The transaction includes online bidding, and wherein the item set includes multiple non-sale items. When executed, the computer-executable instructions also cause the system to: receive item information associated with a transaction from the seller; and suspend the seller's participation in the transaction of the item based on the identified item being non-saleable. The counterfeit item identification model includes a trained encoder-decoder model, and wherein the counterfeit item identification model includes a converter model fine-tuned for one or more predetermined categories of items for counterfeit item identification. Image analysis includes one or more of the following: extracting features from first data, using a probability distribution to predict whether the first data matches one or more genuine non-saleable item data, and determining the item as non-saleable based on the probability distribution.

[0065] In another aspect, the present technology relates to a computer-readable non-transitory recording medium storing computer-executable instructions. When executed by a processor, the computer-executable instructions cause a computer system to receive information about an item associated with a transaction, wherein the information includes first data associated with the item; match the first data with at least second data in a dataset based on a non-sale item identification model, wherein the dataset is associated with a set of non-sale items in a database, wherein the non-sale item identification model includes a set of encoders and decoders, each encoder and decoder being trained using a non-sale item training dataset in the database, and wherein the non-sale item identification model generates a probability distribution of the likelihood that the item is a non-sale item; identify the item as a non-sale item based on the matching of the first data with the at least second data; and provide a warning associated with the item being a non-sale item, causing a suspension of the transaction associated with the item. The first data includes image data, and the computer-executable instructions, when executed, further cause the system to: determine one or more features of the image data based on the category of the item; and perform a pixel-level comparison of at least a portion of the image data with a set of stock photos associated with the non-sale items set, based on the one or more features of the image data, wherein the non-sale items set includes at least one of counterfeit or stolen items. The computer-executable instructions, when executed, further cause the system to: determine one or more features of the first data, wherein the one or more features include one or more types of the item, one or more portions of the first data for detailed analysis, the color of the item, or the storage conditions of the item; determine one or more features of the first data, wherein the one or more features include one or more of the following: the type of the item, one or more portions of the first data associated with the item, the color of the item, or the storage conditions of the item; match the first data with at least one item represented in the dataset using at least a trained converter model, wherein the trained converter model is fine-tuned to the at least one or more types of items; identify the seller of the item based on a probability distribution associated with the match; and suspend the transaction of the item and one or more bidding processes associated with the seller. The transaction involves online bidding, and the collection of items includes multiple non-sale items. The computer-executable instructions, when executed, also cause the system to: receive item information associated with the transaction from the seller; and suspend the seller's participation in the transaction of the item based on the identified non-sale item. Image analysis includes one or more of the following: extracting features from first data, using a probability distribution to predict whether the first data matches one or more real non-sale item data, and determining the item as non-sale based on the probability distribution.

[0066] Any one of the above-mentioned aspects in combination with any other aspect of the above-mentioned aspects. Any one of the one or more aspects described herein.

Claims

1. A method executed by an online transaction system for identifying items for sale as counterfeit, the method comprising: Receive information on items for sale associated with a transaction in the online trading system, wherein the information includes first image data associated with the items for sale provided by the seller of the items for sale; The encoder of the counterfeit identification model generates a multidimensional vector by encoding the first image data, the multidimensional vector mapping the first image data, the counterfeit identification model including a converter model, the converter model being fine-tuned for one or more predetermined categories of items for counterfeit identification; The multidimensional vector is decoded by the decoder of the counterfeit identification model to generate probability distribution data indicating the likelihood that the first image data matches at least a second image data in the dataset, wherein the dataset is associated with a set of counterfeit items in the database; The counterfeit identification model fine-tunes the probability distribution data from the decoder based on items associated with a predetermined category, wherein the predetermined category specifies one or more portions of an item that are compared using one or more feature vectors associated with a predetermined portion of the first image data. Based on the probability distribution data, the items for sale are identified as counterfeit. Provide a warning associated with the item being sold being a counterfeit; as well as Suspend transactions associated with the items for sale.

2. The method of claim 1, wherein, The method further includes: Based on the category of the items for sale, determine one or more features of the first image data; and Based on one or more features of the first image data, perform a pixel-level comparison of at least a portion of the first image data with a set of stock photos associated with the set of counterfeit items.

3. The method according to claim 1, further comprising: Determine one or more features of the first image data, wherein the one or more features include one or more of the following: the type of the item for sale, one or more portions of the first image data associated with the item for sale, the color of the item for sale, or the storage conditions of the item for sale; The first image data is matched with at least one item represented in the dataset using at least one of a trained neural network or a trained converter model. Based on the probability distribution associated with the match, the seller of the item for sale is identified; as well as Suspend the transaction of the item for sale and one or more bidding processes associated with the seller.

4. The method of claim 1, wherein, The transaction includes online bidding, and the set of counterfeit items includes multiple counterfeit items.

5. The method according to claim 1, further comprising: Receive information about items for sale associated with a transaction from the seller; as well as Based on the identification that the item for sale is counterfeit, the seller's participation in the transaction of the item for sale is suspended.

6. The method of claim 1, wherein, The counterfeit identification model includes a trained encoder-decoder model.

7. The method of claim 1, wherein, The image analysis includes one or more of the following: Extract features from the first image data. A probability distribution is used to predict whether the first image data matches one or more real counterfeit data, and Based on the probability distribution, the item for sale is determined to be counterfeit.

8. An online transaction system for identifying non-sale items, the online transaction system comprising: processor; as well as The memory stores computer-executable instructions that, when executed by the processor, enable the online trading system to: Receive information on items for sale associated with a transaction in the online trading system, wherein the information includes first image data associated with the items for sale provided by the seller of the items for sale; The encoder of the non-sale item identification model generates a multidimensional vector by encoding the first image data, the multidimensional vector mapping the first image data, the non-sale item identification model including a converter model, the converter model being fine-tuned for one or more predetermined categories of items for counterfeit item identification; The multidimensional vector is decoded by the decoder of the non-sale item recognition model to generate probability distribution data indicating the likelihood that the first image data matches at least a second image data in the dataset, wherein the dataset is associated with a set of non-sale items in the database; The non-sale item identification model fine-tunes the probability distribution data from the decoder based on items associated with a predetermined category, wherein the predetermined category specifies one or more portions of an item that are compared using one or more feature vectors associated with a predetermined portion of the first image data. Based on the probability distribution data, the items for sale are identified as non-sale items; Provide a warning associated with the item being offered for sale being a non-sale item; and Suspend transactions associated with the items for sale.

9. The online transaction system of claim 8, wherein, The computer-executable instructions, when executed, also enable the online trading system to: Based on the category of the items for sale, determine one or more features of the first image data; as well as Based on one or more features of the first image data, perform a pixel-level comparison of at least a portion of the first image data with a set of stock photos associated with the set of non-sale items, wherein the set of non-sale items includes at least one of counterfeit or stolen items.

10. The online trading system of claim 8, wherein the computer-executable instructions, when executed, further cause the online trading system to: Determine one or more features of the first image data, wherein the one or more features include one or more of the following: the type of the item for sale, one or more portions of the first image data associated with the item for sale, the color of the item for sale, or the storage conditions of the item for sale; The first image data is matched with at least one item represented in the dataset using at least one of a trained neural network or a trained converter model. Based on the probability distribution associated with the match, the seller of the item for sale is identified; as well as Suspend the transaction of the item for sale and one or more bidding processes associated with the seller.

11. The online transaction system of claim 8, wherein, The transaction includes online bidding, and the set of non-sale items includes multiple non-sale items.

12. The online trading system according to claim 8, wherein the computer-executable instructions, when executed, further cause the online trading system to: Receive information about items for sale associated with a transaction from the seller; and Based on the fact that the identified item for sale is not for sale, the seller's participation in the transaction of the item for sale is suspended.

13. The online transaction system of claim 8, wherein, The non-sale item identification model includes a trained encoder-decoder model.

14. The online transaction system of claim 8, wherein, The image analysis includes one or more of the following: Extract features from the first image data. A probability distribution is used to predict whether the first image data matches one or more real non-sale items, and Based on the probability distribution, the item to be sold is determined to be a non-sale item.

15. A computer-readable non-transitory recording medium storing computer-executable instructions, which, when executed by a processor, enable an online trading system implemented by a computer system: Receive information on items for sale associated with a transaction in the online trading system, wherein the information includes first image data associated with the items for sale provided by the seller of the items for sale; The encoder of the non-sale item identification model generates a multidimensional vector by encoding the first image data, the multidimensional vector mapping the first image data, the non-sale item identification model including a converter model, the converter model being fine-tuned for one or more predetermined categories of items for counterfeit item identification; The multidimensional vector is decoded by the decoder of the non-sale item recognition model to generate probability distribution data indicating the likelihood that the first image data matches at least a second image data in the dataset, wherein the dataset is associated with a set of non-sale items in the database; The non-sale item identification model fine-tunes the probability distribution data from the decoder based on items associated with a predetermined category, wherein the predetermined category specifies one or more portions of an item that are compared using one or more feature vectors associated with a predetermined portion of the first image data. Based on the probability distribution data, the items for sale are identified as non-sale items; Provide a warning associated with the item being offered for sale being a non-sale item; and Suspend transactions associated with the items for sale.

16. The computer-readable non-transitory recording medium of claim 15, wherein, The computer-executable instructions, when executed, also enable the online trading system to: Based on the category of the items for sale, determine one or more features of the first image data; as well as Based on one or more features of the first image data, perform a pixel-level comparison of at least a portion of the first image data with a set of stock photos associated with the set of non-sale items, wherein the set of non-sale items includes at least one of counterfeit or stolen items.

17. The computer-readable non-transitory recording medium of claim 15, wherein, The computer-executable instructions, when executed, also enable the online trading system to: Determine one or more features of the first image data, wherein the one or more features include one or more types of the item for sale, one or more portions of the first image data for detailed analysis, the color of the item for sale, or the storage conditions of the item for sale; Determine one or more features of the first image data, wherein the one or more features include one or more of the following: the type of the item for sale, one or more portions of the first image data associated with the item for sale, the color of the item for sale, or the storage conditions of the item for sale; Using at least a trained converter model, the first image data is matched with at least one item represented in the dataset, wherein the trained converter model is at least fine-tuned to the one or more types of items; Based on the probability distribution associated with the match, the seller of the item for sale is identified; as well as Suspend the transaction of the item for sale and one or more bidding processes associated with the seller.

18. The computer-readable non-transitory recording medium of claim 15, wherein, The transaction includes online bidding, and the set of non-sale items includes multiple non-sale items.

19. The computer-readable non-transitory recording medium of claim 15, wherein the computer-executable instructions, when executed, further cause the online trading system to: Receive information about items for sale associated with a transaction from the seller; and Based on the fact that the identified item for sale is not for sale, the seller's participation in the transaction of the item for sale is suspended.

20. The computer-readable non-transitory recording medium of claim 15, wherein, Image analysis includes one or more of the following: Extract features from the first image data. A probability distribution is used to predict whether the first image data matches one or more real non-sale items, and Based on the probability distribution, the item to be sold is determined to be a non-sale item.

Images