Information processing, model training, feature library creation method and electronic equipment

Abstract

The embodiment of the present application provides an information processing, model training, feature library creation method and electronic device. Among them, the information processing method includes: obtaining image information of the target object; based on the image information, obtaining the category to which the target object belongs; determining the reference price of the target object based on the image information and the category to which the target object belongs; and displaying the reference price. The technical solution provided by the embodiment of the present application determines a reference price for the target object based on the image information of the target object. This reference price can be a specific price or a price range, so that the user who publishes the target object can set a more appropriate price for the target object based on the reference price.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to an information processing, model training, feature library creation method and electronic equipment. Background Art

[0002] In the secondhand market, a suitable price is crucial for facilitating rapid product turnover. However, it's difficult for average users to directly offer a suitable price, which in turn hinders sales. Furthermore, unlike e-commerce websites, the product information uploaded by users through client applications on secondhand market websites is often incomplete; for example, some products may only have images.

[0003] How to make reasonable price estimates based on product images is a problem that needs to be solved to help users set a more appropriate price. Summary of the Invention

[0004] Each embodiment of the present application provides an information processing, model training, feature library creation method and electronic device that solves or partially solves the existing technologies.

[0005] In one embodiment of the present application, an information processing method is provided. The method includes:

[0006] Obtain image information of the target object;

[0007] Based on the image information, obtaining the category to which the target object belongs;

[0008] Determining a reference price of the target object based on the image information and the category to which the target object belongs;

[0009] The reference price is displayed.

[0010] In another embodiment of the present application, an information processing method is provided. The method includes:

[0011] Get the image information of the product;

[0012] Using the image information as input to a calculation model, executing the calculation model to obtain a reference price for the product;

[0013] The calculation model is trained based on the commodity images and transaction prices of the traded commodities.

[0014] In another embodiment of the present application, an information processing method is provided. The method includes:

[0015] Get the image information of the product;

[0016] When the discriminant model is used to determine that the image information belongs to the difficult type, at least one reference sample that meets the similarity requirement with the image information is retrieved based on the image feature index library;

[0017] Determining a reference price for the product based on a price corresponding to each of the at least one reference samples;

[0018] The discriminant model is obtained by training based on sample images and the difficulty level of the sample images.

[0019] In another embodiment of the present application, a model training method is provided. The method includes:

[0020] Obtaining a first training sample; wherein the first training sample includes: a commodity image and a transaction price of a traded commodity;

[0021] Using the product image as input of a computational model to be trained, and executing the computational model to obtain a first output result;

[0022] Optimizing parameters of the calculation model according to the first output result and the transaction price;

[0023] The trained computing model is used to determine the reference price of the commodity based on the image information of the commodity.

[0024] In another embodiment of the present application, a model training method is provided. The training method includes:

[0025] Obtaining a second training sample, wherein the second training sample includes: an image of a traded commodity and a difficulty category of the commodity image;

[0026] Using the product image as an input of a discriminant model to be trained, and executing the discriminant model to obtain a third output result;

[0027] Optimizing the parameters of the discriminant model according to the third output result and the difficulty type;

[0028] The trained discriminant model is used to determine, based on the image information of the commodity, whether the image information can be used to calculate the reference price of the commodity based on the image information using the calculation model.

[0029] In another embodiment of the present application, a method for creating a feature library is provided. The method includes:

[0030] Collecting commodity information of traded commodities, wherein the commodity information includes commodity images and transaction prices;

[0031] Using the estimation model, we can identify some difficult product samples from the collected product information.

[0032] Based on difficult samples, an image feature index library is created to retrieve reference samples that meet similarity requirements with the image information of the target product, so as to determine the reference price of the target product based on the price corresponding to the reference sample.

[0033] In another embodiment of the present application, an electronic device is provided. The electronic device includes a memory, a processor, and a display; wherein,

[0034] The memory is used to store programs;

[0035] The processor is coupled to the memory and is configured to execute the program stored in the memory to:

[0036] Obtain image information of the target object;

[0037] Based on the image information, obtaining the category to which the target object belongs;

[0038] Determining a reference price of the target object based on the image information and the category to which the target object belongs;

[0039] The display is controlled to display the reference price.

[0040] In another embodiment of the present application, an electronic device is provided. The electronic device includes a memory and a processor; wherein,

[0041] The memory is used to store programs;

[0042] The processor is coupled to the memory and is configured to execute the program stored in the memory to:

[0043] Get the image information of the product;

[0044] Using the image information as input to a calculation model, executing the calculation model to obtain a reference price for the product;

[0045] The calculation model is trained based on the commodity images and transaction prices of the traded commodities.

[0046] In another embodiment of the present application, an electronic device is provided. The electronic device includes a memory and a processor; wherein,

[0047] The memory is used to store programs;

[0048] The processor is coupled to the memory and is configured to execute the program stored in the memory to:

[0049] Get the image information of the product;

[0050] When the discriminant model is used to determine that the image information belongs to the difficult type, at least one reference sample that meets the similarity requirement with the image information is retrieved based on the image feature index library;

[0051] Determining a reference price for the product based on the price corresponding to each of the at least one reference samples;

[0052] The discriminant model is obtained by training based on sample images and the difficulty level of the sample images.

[0053] In another embodiment of the present application, an electronic device is provided. The electronic device includes a memory and a processor; wherein,

[0054] The memory is used to store programs;

[0055] The processor is coupled to the memory and is configured to execute the program stored in the memory to:

[0056] Obtaining a first training sample; wherein the first training sample includes: a commodity image and a transaction price of a traded commodity;

[0057] Using the product image as input of a computational model to be trained, and executing the computational model to obtain a first output result;

[0058] Optimizing parameters of the calculation model according to the first output result and the transaction price;

[0059] The trained computing model is used to determine the reference price of the commodity based on the image information of the commodity.

[0060] In another embodiment of the present application, an electronic device is provided. The electronic device includes a memory and a processor; wherein,

[0061] The memory is used to store programs;

[0062] The processor is coupled to the memory and is configured to execute the program stored in the memory to:

[0063] Obtaining a second training sample, wherein the second training sample includes: an image of a traded commodity and a difficulty category of the commodity image;

[0064] Using the product image as an input of a discriminant model to be trained, and executing the discriminant model to obtain a third output result;

[0065] Optimizing the parameters of the discriminant model according to the third output result and the difficulty type;

[0066] The trained discriminant model is used to determine, based on the image information of the commodity, whether the image information can be used to calculate the reference price of the commodity based on the image information using the calculation model.

[0067] In another embodiment of the present application, an electronic device is provided. The electronic device includes a memory and a processor; wherein,

[0068] The memory is used to store programs;

[0069] The processor is coupled to the memory and is configured to execute the program stored in the memory to:

[0070] Collecting commodity information of traded commodities, wherein the commodity information includes commodity images and transaction prices;

[0071] Using the estimation model, we can identify some difficult product samples from the collected product information.

[0072] Based on difficult samples, an image feature index library is created to retrieve reference samples that meet similarity requirements with the image information of the target product, so as to determine the reference price of the target product based on the price corresponding to the reference sample.

[0073] Another embodiment of the present application further provides an information processing method. The information processing method includes:

[0074] Obtain image information of the target object;

[0075] Based on the image information, obtaining the category to which the target object belongs;

[0076] Determining a reference price for the target object and auxiliary information that enables the target object to meet a first preset condition based on the image information and the category to which the target object belongs;

[0077] The reference price and the auxiliary information are respectively associated with the image information and displayed.

[0078] The technical solution provided in the embodiment of the present application determines a reference price for the target object based on the image information of the target object. This reference price can be a specific price or a price range, so that the user who publishes the target object can set a more appropriate price for the target object based on the reference price.

[0079] Another embodiment of the present application provides a technical solution that utilizes a computational model to directly obtain a reference price for a commodity based on its image information; wherein the computational model is trained based on commodity images and transaction prices of commodities that have been traded in the past; and facilitates users who publish the commodity to set a more appropriate price for the commodity based on the reference price.

[0080] In the technical solution provided in another embodiment of the present application, a discriminant model is used to determine whether the image information of the product belongs to a simple type. When the image information of the product belongs to a simple type, a calculation model is used to complete the calculation of the product parameter data based on the image information. This solution increases the discrimination of the difficulty type of the image information, and then selectively selects a suitable calculation method to obtain the reference price of the product with high accuracy.

[0081] In the technical solution provided in another embodiment of the present application, when the discriminant model is used to determine that the image information of a commodity is of a difficult type, similar samples are retrieved based on the image feature index library; then, the reference price related to the commodity transaction is determined based on the reference price corresponding to the similar samples; this solution increases the discrimination of the difficulty type of the image information, and then selectively selects a suitable determination method to obtain the reference price of the commodity, with high accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0083] Figure 1a An interface diagram provided for an embodiment of the present application;

[0084] Figure 1b A flowchart of an information processing method provided in one embodiment of the present application;

[0085] Figure 2a Another interface diagram provided for an embodiment of the present application;

[0086] Figure 2b Another interface diagram provided for an embodiment of the present application;

[0087] Figure 2c A flowchart of an information processing method provided in another embodiment of the present application;

[0088] Figure 2d Another interface diagram provided for an embodiment of the present application;

[0089] Figure 3A flowchart of an information processing method provided in yet another embodiment of the present application;

[0090] Figure 4 A flowchart of an information processing method provided in yet another embodiment of the present application;

[0091] Figure 5 A flowchart of a model training method provided in one embodiment of the present application;

[0092] Figure 6 A flowchart of a model training method provided in another embodiment of the present application;

[0093] Figure 7 A flowchart of a method for creating a feature library according to an embodiment of the present invention is provided;

[0094] Figure 8 A flowchart of an information processing method provided in yet another embodiment of the present application;

[0095] Figure 9 A schematic diagram of the structure of an information processing device provided in one embodiment of the present application;

[0096] Figure 10 A schematic structural diagram of an information processing device provided in another embodiment of the present application;

[0097] Figure 11 A schematic structural diagram of an information processing device provided in yet another embodiment of the present application;

[0098] Figure 12 A schematic diagram of the structure of a model training device provided in one embodiment of the present application;

[0099] Figure 13 A schematic structural diagram of a model training device provided in another embodiment of the present application;

[0100] Figure 14 A schematic diagram of the structure of a feature library creation device provided in one embodiment of the present application;

[0101] Figure 15 A schematic diagram of the structure of an electronic device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0102] Users can upload pictures of items they want to sell, along with information such as the selling price, to the secondhand market application platform. Other users can then view the item's pictures and price information through the corresponding client application. Currently, users set their own prices for their items. If users lack familiarity with the secondhand market for a particular item, they may offer a price that is either too high or too low. If the price is too high, the item may be difficult to sell; if the price is too low, the user may suffer financial losses.

[0103] Some products on secondhand market platforms lack product information, such as images. Unlike products sold on e-commerce platforms, which often include not only images but also descriptive information such as brand, material, and model, this application aims to provide a solution for evaluating product prices based on image information.

[0104] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0105] In some processes described in the specification, claims and the above-mentioned figures of this application, multiple operations that appear in a specific order are included. These operations may not be executed in the order in which they appear in this document or may be executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish between different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequence, nor do they limit "first" and "second" to different types. In addition, the following embodiments are only some of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of this application.

[0106] Figure 1b FIG. 1 shows a flow chart of an information processing method provided by an embodiment of the present application. Figure 1b As shown, the information processing method includes:

[0107] 101. Obtain image information of the target object.

[0108] 102. Based on the image information, obtain the category to which the target object belongs.

[0109] 103. Determine a reference price of the target object based on the image information and the category to which the target object belongs.

[0110] 104. Display the reference price.

[0111] In combination with specific scenarios, the target object in the above 101 may be a commodity, such as a new commodity or a second-hand commodity. The image information may include images of one or more commodities.

[0112] In step 102 above, a category determination model can be used to determine the category to which the target object belongs. Specifically, the category determination model is run using the image information as input to determine the category to which the target object belongs. The category determination model is trained based on images of traded goods and the categories to which they belong. The training process for the category determination model will be described in detail in the following embodiments.

[0113] In a feasible technical solution, the above step 103 can be completed using a computing model. For example, Figure 1a As shown, the calculation model is trained based on the product images, product categories and transaction prices of traded products; or, using the image feature index library, at least one reference sample that meets the similarity requirements with the image information is retrieved; then, the reference price of the target object is determined in combination with the price corresponding to the reference sample.

[0114] The reference price determined for the target object may be a specific value or a price range, which is not specifically limited in this embodiment.

[0115] In the above 104, the reference price can be displayed in association with the image information, such as Figure 1a As shown, the reference price is displayed on the product image; of course, the reference price can also be displayed in other forms, which is not specifically limited in this embodiment.

[0116] The technical solution provided in this embodiment determines a reference price for the target object based only on the image information of the target object. This reference price can be a specific price or a price range. In this way, the user who publishes the target object can set a more appropriate price for the target object based on the reference price.

[0117] The execution subject of the method provided in this embodiment can be a server or a client. The server can be a physical server, a virtual server, a cloud service platform, etc., and this embodiment does not specifically limit this. The client can be any device such as a laptop, a desktop computer, a smart phone, a tablet computer, a smart wearable device (such as a smart watch). A client application is installed on the client, such as a second-hand item trading application. The client can establish a communication connection with the server via a wireless network or a wired network.

[0118] Assuming that the execution subject of the method provided in this embodiment is the server, then the image information of the target object described in 101 above can be uploaded to the server by the user through the client. After receiving the image information, the server triggers the operation of determining the reference price for the target object (i.e., the above steps 102 to 103); or, after receiving the image information, the server does not immediately determine the reference price; instead, after receiving the reference price calculation request for the target object triggered by the user through the client, it triggers the operation of determining the reference price for the target object. In specific implementation, the user can trigger the reference price calculation request by touching the corresponding control on the client application; the reference price calculation request can also be triggered by issuing a control voice that meets the requirements; and so on. This embodiment does not make specific limitations on this.

[0119] Assuming the method provided in this embodiment is executed by a client, the image information of the target object described in step 101 above can be imported by the user from a photo album into the corresponding client application, or captured by the user using the capture function on the client application. Similarly, the client can trigger the operation of determining a reference price for the target object after the user imports or successfully captures the image information of the target object; the client can also execute the operation of determining a reference price for the target object after the user triggers a reference price calculation request for the target object.

[0120] Furthermore, the above step 103 of “determining a reference price of the target object based on the image information and the category to which the target object belongs” may include:

[0121] When the discriminant model determines that the image information belongs to a simple type, the image information and the category to which the target object belongs are used as inputs of a calculation model, and the calculation model is executed to obtain the reference price;

[0122] The discrimination model is obtained by training based on sample images and the difficulty categories to which the sample images belong; and the calculation model is obtained by training based on the product images, product categories and transaction prices of traded products.

[0123] Furthermore, the above step 103 of “determining a reference price of the target object based on the image information and the category to which the target object belongs” may further include:

[0124] When the discriminant model is used to determine that the image information belongs to the difficult type, at least one reference sample that meets the similarity requirement with the image information is retrieved based on the image feature index library;

[0125] A reference price of the target object is determined according to the price corresponding to each of the at least one reference samples.

[0126] During specific implementation, there is also a situation where the reference sample may not be retrieved based on the image feature index library. In this case, the method provided in this embodiment may further include the following steps:

[0127] When no reference sample that meets the similarity requirement with the image information is retrieved based on the image feature index library, a prompt content is displayed to refuse to determine the reference price for the target object.

[0128] In one feasible solution, in the above 104 , a reference price, such as a price value or a price range, may be displayed on the image of the target object. Figure 1a The price value is displayed.

[0129] In actual applications, after the target object is appraised, if the appraisal price is low, the user may give up selling the target object and dispose of it as garbage due to cost considerations (e.g., shipping costs). To facilitate the user's garbage disposal, the method provided in this embodiment may further include the following steps:

[0130] Determine the garbage category to which the target object belongs according to the category to which the target object belongs;

[0131] Displays the garbage category to which the target object belongs.

[0132] The correspondence between multiple categories and garbage categories can be established in advance, and then the garbage category to which the target object belongs can be queried based on the category to which the target object belongs and the correspondence.

[0133] Furthermore, the method provided in this embodiment may further include the following steps:

[0134] According to the preset category list, determining whether the category to which the target object belongs meets a second preset condition;

[0135] Based on the judgment result, relevant suggestions are displayed.

[0136] In one embodiment, a preset category list may be set according to actual needs. When the category to which the target object belongs is found in the preset category list, it is determined that the category to which the target object belongs meets the second preset condition.

[0137] For example, a preset category list contains all product categories that can be sold on the online platform. If the target object's category is in the preset category list, a "available for sale" suggestion may be displayed; if the target object's category is not in the preset category list, a "not available for sale" suggestion may be displayed.

[0138] It should be noted that in practice, not all categories of goods can be sold on online platforms. For example, clothing cannot be sold on a used car sales platform; for another example, certain prohibited goods cannot be sold on online platforms.

[0139] For example, a preset category list includes all product categories that increase in value if held, such as gold and watches. If the target item belongs to a category in the preset category list, a suggestion of "holding on to increase value" may be displayed. If the target item does not belong to a category in the preset category list, a suggestion of "sell as soon as possible, as holding on to it will cause it to depreciate" may be displayed.

[0140] In practical applications, an object may be composed of multiple components. For example, a desktop computer consists of a monitor and a host computer; another example is a diamond ring consisting of a diamond and a platinum setting. To meet the user's needs for disassembly and resale, it is necessary to estimate the reference price of at least one component of the target object and provide it to the user for reference. Specifically, the above method may also include the following steps:

[0141] Determining a reference price of at least one component unit constituting the target object according to a reference price of the target object and a category to which the target object belongs;

[0142] A reference price of the at least one component unit is displayed.

[0143] Among them, at least one component unit can be a partial component unit or all component units that constitute the target object, which can be set according to actual needs and is not specifically limited in the embodiments of the present application.

[0144] In specific implementations, the price ratios of the multiple components that make up an object within a certain category can be determined in advance based on big data statistics. A correspondence between categories and price ratios can then be established. Subsequently, the reference prices of each component within at least one component of the target object can be determined based on the reference price of the target object and the corresponding price ratios within the category to which it belongs. For example, for the cup category, the price ratio of the cup body and coasters is 3:1. If the reference price of the cup is predicted to be 20 yuan, the reference price of the cup body would be 15 yuan, and the reference price of the coaster would be 5 yuan.

[0145] The reference price of each component unit in the at least one component unit may be displayed in association with the image of each component unit on the image of the target object. Alternatively, the reference price of each component unit in the at least one component unit may be displayed one by one in the form of a list. Figure 2dAs shown, the cup is disassembled into two components: the cup body and the coaster, and the reference prices of the two components are displayed in the form of a list.

[0146] Taking the second-hand e-commerce application scenario as an example, users who sell second-hand goods usually have little selling experience. Some users only upload the image of the goods they want to sell through the client, and then set a selling price for the goods based on the reference price. For consumers, they hope to obtain more information about the goods, such as pictures from multiple angles, videos of how to use the goods, and purchase invoices for the goods (similar to those used to prove that the goods were purchased not long ago). If there are only product images and reference prices, consumers' desire to buy will not be too high. In order to promote the rapid sale of goods, the technical solution provided in the embodiment of the present application can provide auxiliary information for the goods in addition to providing a more reasonable reference price for the goods. That is, the present application provides such Figures 2a to 2c The technical solution shown.

[0147] Figure 2c FIG. 1 shows a flow chart of an information processing method provided by an embodiment of the present application. Figure 2c As shown, the method includes:

[0148] 101'. Obtain image information of the target object.

[0149] 102'. Based on the image information, obtain the category to which the target object belongs;

[0150] 103 ′: Determine a reference price for the target object and auxiliary information that enables the target object to meet a first preset condition based on the image information and the category to which the target object belongs.

[0151] 104 ′: associate the reference price and the auxiliary information with the image information and display them respectively.

[0152] For the contents of determining the reference price in the above steps 101 ′, 102 ′ and 103 ′, please refer to the description in the above embodiment, which will not be repeated here.

[0153] In the above 103 ′, the auxiliary information that enables the target object to meet the first preset condition may include but is not limited to at least one of the following:

[0154] Suggestions for modifications to the image information;

[0155] Recommendations on combination objects that can form a combination relationship with the target object;

[0156] Suggestions for improvements related to the stated target object;

[0157] Suggestions for adding descriptive information to the target object.

[0158] Figure 2a An interface diagram showing "modification suggestions on the image information" is shown, that is, in addition to the reference price, the interface diagram also shows a suggestion of "Is this just one picture? Can you add pictures from other angles?" Figure 2b It shows that "recommendations on combination objects that can form a combination relationship with the target object" are displayed, that is, in addition to the reference price, the interface diagram also shows the suggestion that "this product can be sold in combination with the following products *** teapot http: / / ***.YYY.&^&&..."

[0159] In one feasible technical solution, in step 103', "determining, for the target object, auxiliary information that causes the target object to meet a first preset condition based on the image information and the category to which the target object belongs" may include:

[0160] Determining modification suggestions for the image information based on the image information; and / or

[0161] Identify the product attributes of the target object based on the image information and the category to which the target object belongs; based on the product attributes, query whether there is a product in the pool of products to be sold that can be sold in combination with the target object; if so, obtain a sales link for the product, and generate a recommendation for a combination object that can form a combination relationship with the target object based on the sales link; and / or

[0162] Identify the product attributes of the target object based on the image information and the category to which the target object belongs; determine improvement suggestions for the target object based on the product attributes of similar products that have been sold historically; and / or

[0163] Identify whether the target object has defects based on the image information; if the identification result indicates that the target object has defects, generate improvement suggestions for the target object based on the identification result; and / or

[0164] When it is detected that the content related to the target object only contains the image information, a suggestion for adding description information for the target object is generated.

[0165] The aforementioned modification suggestions for the image information may specifically include suggestions for adding images from different angles, suggestions for adding animated images, etc. The modification suggestions for the image information may be determined based on the number of images included in the image information. For example, if the image information contains only one image, an improvement suggestion may be generated to add images from different angles.

[0166] The aforementioned product attributes may include product names, such as coffee cups and glasses. Locally, pre-set product combination rules can be used, such as teacups and teapots, charging cables and plugs, and quilts and quilt covers. In practice, pre-set product combination rules can be used to determine which products can be sold in combination for the target audience.

[0167] The aforementioned product attributes may include color attributes. For example, based on historically sold product attributes of similar products, such as the color of mobile phones, and statistics showing that black mobile phones have a high shipment volume, a recommendation may be generated based on the analysis results to suggest that the user replace the phone case of the phone shown in the image information with a black one.

[0168] The above-mentioned defect recognition of image information can be implemented using existing technologies and is not specifically limited in this embodiment. Taking a mobile phone as an example, assuming that a crack is detected on the screen of a mobile phone uploaded by a user, a recommendation can be generated based on the recognition result to recommend that the user replace the screen with a new one.

[0169] In specific implementation, the auxiliary information in the above step 103' can be obtained by Figure 2a and 2b The auxiliary information generation module shown in FIG is generated based on the input image information and the category to which the target object belongs. The auxiliary information generation module can be hardware with an embedded program, or application software, etc.

[0170] Figure 3 FIG. 1 shows a flow chart of an information processing method provided by an embodiment of the present application. Figure 3 As shown, the information processing method includes:

[0171] 201. Obtain image information of the product.

[0172] 202. Use the image information as input of a calculation model, and execute the calculation model to obtain a reference price of the product.

[0173] The calculation model is trained based on the commodity images and transaction prices of the traded commodities.

[0174] The execution subject of the method provided in this embodiment can be a server or a client. The server can be a physical server, a virtual server, a cloud service platform, etc., and this embodiment does not specifically limit this. The client can be any device such as a laptop, a desktop computer, a smart phone, a tablet computer, a smart wearable device (such as a smart watch). A client application is installed on the client, such as a second-hand item trading application. The client can establish a communication connection with the server via a wireless network or a wired network.

[0175] Assuming that the execution subject of the method provided in this embodiment is the server, then the image information of the product described in the above 201 can be uploaded to the server by the user through the client. After receiving the image information, the server triggers the operation of calculating the reference price of the product using the calculation model; or, after receiving the image information, the server does not immediately calculate the reference price of the product; instead, after receiving the reference price calculation request for the product triggered by the user through the client, it calculates the reference price of the product based on the image information of the product using the calculation model. In specific implementation, the user can trigger the reference price calculation request by touching the corresponding control on the client application; can also trigger the reference price calculation request by issuing a control voice that meets the requirements; and so on. This embodiment does not specifically limit this.

[0176] Assuming the method provided in this embodiment is executed by a client, the product image information described in step 201 can be imported from a user's photo album into the corresponding client application, or captured by the user using the capture function on the client application. Similarly, the client can trigger the calculation of the product's reference price using the computational model after the user has imported or successfully captured the product's image information. Alternatively, the client can use the computational model to calculate the product's reference price based on the product's image information after the user has triggered a request to calculate the product's reference price.

[0177] In the above 202, the calculated reference price can be a specific value or a value range.

[0178] In addition, it should be added here that the training process of the computing model will be described in detail below.

[0179] The technical solution provided in this embodiment utilizes a computing model to directly obtain a reference price for a commodity based on its image information; the computing model is trained based on the images and transaction prices of commodities that have been traded in the past; the user selling the commodity can set a more appropriate price for the commodity based on the reference price.

[0180] In one feasible technical solution, the above step 202 of "using the image information as input of a calculation model and executing the calculation model to obtain a reference price of the product" may specifically include the following steps:

[0181] 2021. Using the image information as input to a category determination model, executing the category determination model to obtain the category to which the product belongs;

[0182] 2022. Use the image information and the category to which the product belongs as inputs to the calculation model, and execute the calculation model to obtain a reference price for the product.

[0183] Accordingly, the calculation model is trained based on the product images, category information and transaction prices of the traded products.

[0184] The category to which a product belongs may include at least one first-level category identifier. For example, if a product is a refrigerator, then the category of the product may include at least two second-level category identifiers, such as the first-level category identifier "home appliances" and the second-level category identifier "refrigerator".

[0185] Adding the ability to determine the category to which a product belongs and using the category as an input to the calculation model will help improve the accuracy of the product reference price calculation.

[0186] For example, judging from product photos (i.e., image information), the products all resemble refrigerators. However, one product is a toy refrigerator, while the other is a home appliance refrigerator. There is still a significant price difference between toys and home appliances. Therefore, this embodiment first determines the product category and then combines the image information and category to determine the product's reference price, which results in higher accuracy.

[0187] Of course, in a specific implementation, the calculation model may include two modules, for example, a first sub-model and a second sub-model. The first sub-model is used to determine the category to which the product belongs based on the product image information; the second sub-model is used to calculate the reference price of the product, such as a price value or a price range, based on the image information and the category.

[0188] It should be noted that all models mentioned in this article, such as the computational model, the category determination model, the discriminant model and the estimation model described below, as well as the two sub-models mentioned above, can be implemented using existing self-learning neural network models, such as fully connected neural network models and convolutional neural networks. The network layers included in each model can be determined based on the functional requirements of the model.

[0189] Furthermore, the information processing method provided in this embodiment may further include the following steps:

[0190] 203. Use the image information as input of a discriminant model, and execute the discriminant model to obtain a discriminant result.

[0191] 204. When the discrimination result indicates that the image information belongs to a simple type, triggering the calculation model to complete the calculation of the reference price of the commodity;

[0192] The discriminant model is trained based on the sample images and the difficulty level of the sample images. Similarly, the training process of the discriminant model will be described in detail below.

[0193] During the implementation of this embodiment, the inventors discovered that if all product images were calculated using a computational model without difficulty discrimination, the reference prices for some products would be significantly skewed. Therefore, this embodiment incorporates a step where the discriminant model is used to discriminate between the difficulty and ease of the image information. If the image information is classified as easy, the computational model is used to calculate the reference price. This addition effectively improves the accuracy of the computational model in determining the reference price.

[0194] Furthermore, the information processing method provided in this embodiment may further include the following steps:

[0195] 205. When the discrimination result indicates that the image information belongs to a difficult type, at least one reference sample that meets similarity requirements with the image information is retrieved based on an image feature index library.

[0196] 206. Determine a reference price for the product based on the price corresponding to each of the at least one reference samples.

[0197] In step 205 above, the image feature index library is created based on the product information of the traded goods. This product information may include, but is not limited to, product images, product categories, and transaction prices. The process of creating the image feature index library will be described in detail below.

[0198] In one feasible technical solution, the above step 205 of “retrieving at least one reference sample that meets the similarity requirement with the image information based on the image feature index library” may specifically include:

[0199] 2051. Extract image features based on the image information;

[0200] 2052. Retrieve at least one candidate sample from the image feature index library according to the image feature;

[0201] 2053. Calculate the similarity value between the image information and each candidate sample respectively;

[0202] 2054. Candidate samples with similarity values ​​greater than a second threshold are reference samples that meet similarity requirements with the image information.

[0203] In step 2052 above, when searching the image feature index library based on the image features, the category to which the product belongs may also be considered. That is, based on the image features and the category to which the product belongs, at least one candidate sample is retrieved from the image feature index library. The purpose of using the category as a search criterion is to ensure that the retrieved candidate sample, in addition to meeting the image feature similarity requirement, also belongs to the same category as the product. Using the product category as a search criterion can effectively improve search accuracy.

[0204] It should be noted here that the specific implementation process of image feature extraction, retrieving candidate samples in the image feature index library based on image features, and calculating the similarity value between image information and candidate samples can be found in the relevant content of the prior art and will not be repeated here.

[0205] In the above 206, the reference price can be determined in a variety of ways, such as the following:

[0206] 1. Determine the average of the prices corresponding to the at least one reference sample as the reference price.

[0207] Assume there are three reference samples that meet the similarity requirement with the image information: reference sample A, reference sample B, and reference sample C. Reference sample A corresponds to a price, reference sample B corresponds to b, and reference sample C corresponds to c. In practice, (a+b+c) / 3 can be determined as the reference price of the product.

[0208] 2. Determine the weighted average of the prices corresponding to the at least one reference sample as the reference price.

[0209] In specific implementations, the weights corresponding to the reference samples can be determined based on their similarity to the image information. Continuing with the above example, assuming that the weight corresponding to reference sample A is w1, the weight corresponding to reference sample B is w2, and the weight corresponding to reference sample C is w3; then (w1*a+w2*b+w3*c) can be determined as the reference price of the product.

[0210] 3. Determine a numerical range as a reference price based on the price corresponding to each of the at least one reference samples.

[0211] Continuing with the above example, assuming a>b>c, the numerical range c~a can be directly used as the reference price of the product.

[0212] Of course, other methods besides the above methods may also be used for implementation, and this embodiment does not specifically limit this.

[0213] Furthermore, the information processing method provided in this embodiment may further include:

[0214] 207. When no reference sample that meets the similarity requirement with the image information is retrieved based on the image feature index library, output a prompt content of refusing to determine a reference price for the product.

[0215] Furthermore, the information processing method provided in this embodiment may further include:

[0216] 208. Obtain historical information related to the product.

[0217] 209. Based on the historical information, modify the reference price.

[0218] In a specific implementation, the historical information includes but is not limited to at least one of the following: the historical publishing price of the user who published the product, and the historical publishing price of products belonging to the same category as the product.

[0219] By analyzing the prices of products that users have posted in the past, we can analyze the price posting preferences of the user, for example, we can analyze the price range of products that users mostly post, and / or whether the price preferences of products posted by users are high or low, etc. Based on the historical posting prices of other products of the same category, the price range of products in this category can be determined. For example, if the user prefers to set a low price, and the reference price is on the high side of this price range, the reference price can be adjusted lower; if the user prefers to set a high price, and the reference price is on the low side of this price range, the reference price can be adjusted higher; and so on. In specific implementation, the correction rules set based on actual needs can be used, or the corresponding model can be used to complete the above correction process; the embodiment of the present application does not specifically limit the correction process.

[0220] Figure 4 FIG. 1 shows a flow chart of an information processing method provided by another embodiment of the present application. Figure 4 As shown, the method includes:

[0221] 301. Obtain image information of the product.

[0222] 302. When the discriminant model is used to determine that the image information belongs to the difficult type, at least one reference sample that meets the similarity requirement with the image information is retrieved based on the image feature index library.

[0223] 303. Determine a reference price for the product based on the price corresponding to each of the at least one reference samples.

[0224] The discriminant model is trained based on the sample images and the difficulty level of the sample images. The training process of the discriminant model is described in detail below.

[0225] For the specific contents of 301 and 303 above, please refer to the corresponding description in the above embodiment.

[0226] In the above 302, the image feature index library can be simply understood as a list of relationships between image features and reference samples. Reference samples can be product information for traded goods, including but not limited to product images, category information, and transaction prices (e.g., transaction prices). Image features can include color, texture, shape, and spatial relationship features. Color features are global features that describe the surface properties of the scene corresponding to an image or image region. Texture features are also global features that describe the surface properties of the scene corresponding to an image or image region. Shape features can be represented in two ways: contour features and region features. Contour features focus on the outer boundaries of objects, while region features relate to the entire shape region. Spatial relationship features refer to the spatial position or relative orientation relationships between multiple objects segmented from an image. These relationships can be categorized into connection / adjacency, overlap / overlap, and containment / inclusion. The specific process of retrieving similar reference samples based on the image feature index library can be found in prior art image retrieval techniques and will not be elaborated upon here.

[0227] In one feasible technical solution, the above step 302 of “retrieving at least one reference sample that meets the similarity requirement with the image information based on the image feature index library” may include:

[0228] 3021. Extract image features based on the image information.

[0229] 3022. Retrieve at least one candidate sample from the image feature index library based on the image feature;

[0230] 3023. Calculate the similarity value between the image information and each candidate sample respectively;

[0231] 3033. Candidate samples with similarity values ​​greater than a second threshold value are reference samples that meet similarity requirements with the image information.

[0232] It should be noted that step 3022 may specifically include retrieving at least one candidate sample from the image feature index library based on the image features, the product category, and the product tag. For some products, users may upload descriptive text for the product in addition to the product image. Based on this text, at least one product tag may be determined for the product.

[0233] In the technical solution provided by this embodiment, when the discriminant model is used to determine that the image information of a product belongs to a difficult type, similar samples are retrieved based on the image feature index library; then the reference price of the product is determined based on the prices corresponding to the similar samples; this solution increases the discrimination of the difficulty type of the image information, and then selectively selects a suitable determination method to obtain the reference price of the product, with high accuracy.

[0234] Furthermore, the information processing method provided in this embodiment may further include the following steps:

[0235] 304. Collect commodity information of the traded commodity; wherein the commodity information includes: commodity image information and transaction price.

[0236] 305. The product information that meets the first preset requirement is regarded as a simple sample; the product information that meets the second preset requirement is regarded as a difficult sample.

[0237] 306. Construct the image feature index library based on the difficult samples.

[0238] The collected commodity information of traded goods may contain some information about goods with abnormal prices, such as a mobile phone sold for 1 yuan. This type of information is considered to be abnormally priced goods. Such product information needs to be deleted. During specific implementation, some samples with excessively high or low prices can be removed through price sample data distribution and manual intervention. In addition, during specific implementation, for samples in dense price ranges, downsampling can be used to sample them. That is, before the above-mentioned step 307, the following step can also be included: data cleaning processing can be performed on the collected commodity information of all traded goods. Among them, this data cleaning processing can include: removing abnormally priced goods information, sampling processing, etc.

[0239] In the above 305, the estimation model can be used to determine whether the product information meets the first preset requirement or the second preset requirement. The estimation model is obtained by training based on the product image and transaction price of the traded product, or is obtained by training based on the product image, the product category and the transaction price of the traded product. The estimation model here can be understood as: a primary reference price determination model; and the calculation model can be understood as: a secondary reference price determination model. The training process of the estimation model and the calculation model is similar, the difference is that the training samples are different. The training samples of the estimation model are all the collected product information of the traded products. And the training samples of the calculation model are: all simple samples. That is, the information processing method provided in this embodiment can also include the following steps:

[0240] 307. Obtain the estimation model.

[0241] 308. Use the product image as input to the estimation model, and execute the estimation model to obtain estimation parameters.

[0242] 309. When the difference between the estimated parameter and the transaction price is less than a first threshold, the product information meets the first preset requirement; when the difference between the estimated parameter and the transaction price is greater than or equal to the first threshold, the product information meets the second preset requirement.

[0243] The purpose of the estimation model is to classify training samples into easy and difficult categories. The classified difficult and easy samples can be used to train the discriminant model. The easy samples can be used to train the computational model mentioned in the above embodiment; while the difficult samples can be used to create an image feature index library.

[0244] As mentioned above, the inventors discovered that if all product image information is not classified as difficult or easy, and the calculation model is used to calculate the reference price of each product, the results of the reference price calculation for some products will have large deviations. Therefore, in order to improve the accuracy of the reference price of the product, this embodiment first uses the estimation model to divide the training samples into difficult samples and simple samples, and then uses the difficult and simple samples to train the discriminant model. Using the trained discriminant model to determine the difficulty type of the product image information, different reference price determination schemes are then adopted based on the difficulty type of the product image information, which helps to improve the accuracy of the reference price determination.

[0245] Figure 5 The flow chart of the model training method provided by an embodiment of the present application is shown. This embodiment provides a training scheme for the computing model mentioned in the above embodiment. Specifically, Figure 5 As shown, the method includes:

[0246] 401. Obtain a first training sample; wherein the first training sample includes: a commodity image and a transaction price of a traded commodity.

[0247] 402. Use the product image as input of a computational model to be trained, and execute the computational model to obtain a first output result.

[0248] 403. Optimize parameters of the calculation model according to the first output result and the transaction price.

[0249] The trained computing model is used to determine the reference price of the commodity based on the image information of the commodity.

[0250] In step 403 above, a loss function may be used to calculate the difference between the first output result and the transaction price. Then, the parameters of the calculation model are optimized based on the difference. Specifically, the optimization process may be as follows:

[0251] Using the first output result and the transaction price as input parameters of a first loss function to calculate a first loss value;

[0252] When the first loss value does not meet the first setting requirement, optimizing the parameters of the calculation model based on the first loss value;

[0253] When the first loss value meets the first setting requirement, the computing model completes training.

[0254] Furthermore, the first training sample also includes: the category to which the traded product belongs. Accordingly, the above step 402 "using the image information as the input of the calculation model and executing the calculation model to obtain the first output result" can be specifically as follows:

[0255] The image information and the category to which the traded commodity belongs are used as inputs of the calculation model, and the calculation model is executed to obtain a first output result.

[0256] Furthermore, the information processing method provided in this embodiment may further include the following steps:

[0257] 404. Use the image information as input of a category determination model to be trained, and execute the category determination model to obtain a second output result.

[0258] 405. Optimize parameters of the category determination model based on the second output result and the category to which the traded product belongs.

[0259] The trained category determination model is used to determine the category information of the product based on the product image information. The above steps 404 and 405 are the training process of the category determination model mentioned in the above embodiment.

[0260] In the above 405, “optimizing parameters of the category determination model according to the second output result and the category to which the traded product belongs” may specifically include:

[0261] Using the second output result and the category to which the traded product belongs as input parameters of a second loss function, and calculating a second loss value;

[0262] When the second loss value does not meet the second setting requirement, optimizing the parameters of the category determination model based on the second loss value;

[0263] When the second loss value meets the set requirements, the category determination model completes training.

[0264] During specific implementation, the above-mentioned loss functions may be selected from functions in the prior art, such as the Huber Loss function or the Truncated Loss function, etc., which is not specifically limited in this embodiment.

[0265] Furthermore, the information processing method provided in this embodiment may further include the following steps:

[0266] 406. Collect product information of the traded product; the product information includes: product image and transaction price.

[0267] 407. The product information that meets the first preset requirement is regarded as a simple sample; the product information that meets the second preset requirement is regarded as a difficult sample.

[0268] 408. Use the simple sample as the first training sample.

[0269] For the contents of 406 to 407, please refer to the corresponding description in the above embodiment.

[0270] In addition, it should be noted that the estimation model mentioned in the above embodiment can also be obtained using the training process described in steps 401 to 402. However, the first training samples used in training the computational model are simple samples, while the training samples used in training the estimation model are collected commodity information of traded commodities. In specific implementations, the estimation model is trained using the first batch of training samples. The trained estimation model is then used to distinguish the difficulty of the second batch of training samples, dividing them into simple samples and difficult samples. The computational model is trained using the simple samples from the second batch of training samples.

[0271] Figure 6 FIG2 shows a flow chart of a model training method provided by another embodiment of the present application. This embodiment provides a training scheme for the discriminant model mentioned in the above embodiment. Figure 6 As shown, the method includes:

[0272] 501. Obtain a second training sample, wherein the second training sample includes: an image of a traded commodity and a difficulty category of the commodity image.

[0273] 502. Use the product image as input of a discriminant model to be trained, and execute the discriminant model to obtain a third output result.

[0274] 503. Optimize the parameters of the discriminant model according to the third output result and the difficulty type;

[0275] The trained discriminant model is used to determine whether the image information of the commodity belongs to a simple type based on the image information of the commodity. When the image information belongs to a simple type, the calculation model can be used to calculate the reference price of the commodity based on the image information.

[0276] In an achievable technical solution, the above 503 “optimizing the parameters of the discrimination model according to the third output result and the difficulty level” may specifically include:

[0277] Using the third output result and the difficulty type as input parameters of a third loss function, and calculating a third loss value;

[0278] When the third loss value does not meet the third setting requirement, optimizing the parameters of the discriminant model based on the third loss value;

[0279] When the third loss value meets the set requirements, the discriminant model completes training.

[0280] It should be noted that the various loss functions mentioned herein, such as the first, second, and third loss functions mentioned above, can be implemented using existing loss functions in the prior art, and this embodiment does not specifically limit this. Furthermore, the process of optimizing model parameters based on loss values ​​can be found in the relevant prior art and will not be elaborated upon herein.

[0281] Furthermore, the model training method provided in this embodiment may also include:

[0282] 504. Collect product information of the traded product; wherein the product information includes: product image and transaction price.

[0283] 505. Use the product image as input to the estimation model, and execute the estimation model to obtain estimation parameters.

[0284] 506. When the difference between the estimated parameter and the transaction price is less than a first threshold, mark the product image as a simple type; when the difference between the estimated parameter and the transaction price is greater than or equal to the first threshold, mark the product image as a difficult type.

[0285] Figure 7 FIG. 1 shows a flow chart of a method for creating a feature library according to an embodiment of the present application. Figure 7 As shown, the feature library creation method includes:

[0286] 601. Collect commodity information of traded commodities; wherein the commodity information includes commodity images and transaction prices.

[0287] 602. Using the estimation model, distinguish some product information as difficult samples from the collected multiple product information.

[0288] 603. Based on the difficult samples, an image feature index library is created to retrieve reference samples that meet similarity requirements with the image information of the target product, so as to determine the reference price of the target product based on the price corresponding to the reference sample.

[0289] In one feasible technical solution, step 602 of "using the estimation model to distinguish some product information as difficult samples from the collected multiple product information" can be implemented by the following steps:

[0290] 6021. Use a product image of product information as input to the estimation model, and execute the estimation model to obtain estimation parameters.

[0291] 6022. When the difference between the estimated parameter and the transaction price of the product information is greater than or equal to a first threshold, the product information is treated as a difficult sample.

[0292] For the specific contents of 601 to 603 , please refer to the corresponding descriptions in the above embodiments.

[0293] Furthermore, the above-mentioned product information may also include but is not limited to at least one of the following: product category information, product tags, etc. These information can be used as image features to create an image feature index library.

[0294] In summary, the technical solutions provided by the embodiments of this application include the following parts:

[0295] 1. Sample Collection

[0296] Step 1.1: Collect recent transaction samples.

[0297] Specifically, collect images of traded goods, their final transaction prices, and the categories they belong to. Some products may have multiple images; when there are multiple images, select one as a sample, use all images, or select only the main image from the multiple images.

[0298] Step 1.2: Filter abnormal price samples to obtain training samples.

[0299] For example, through price sample data distribution and manual intervention, some samples with excessively high or low prices can be removed, and samples in dense price ranges can be downsampled.

[0300] Step 1.3: Use the estimation model to divide the training samples into easy samples and difficult samples.

[0301] 2. Model Training

[0302] 2.1 Category Determination Model Training

[0303] Step 2.1.1: Use the product images in the training samples as input to the category determination model to be trained, and execute the category determination model to obtain the category result.

[0304] Step 2.2.2: Optimize the parameters of the category determination model based on the category results and the categories associated with the product images.

[0305] 2.2 Training of computational models

[0306] Step 2.2.1: Obtain all simple samples; simple samples include product images, transaction prices, and categories.

[0307] Step 2.2.2: Use the product images and categories as inputs to the computational model to be trained, and execute the computational model to obtain computational results.

[0308] Step 2.2.3: Optimize the parameters of the calculation model based on the calculation results and the transaction price.

[0309] 3. Feature Library Creation

[0310] Step 3.1: Get all difficult samples.

[0311] Step 3.2: Based on all collected difficult samples, build an image feature index library.

[0312] After completing the above preparations, you can proceed to estimate the price of the product based on the product's image information.

[0313] 4. The basic process of online estimation, such as Figure 8 As shown, including:

[0314] Step 4.1: Image information preprocessing.

[0315] The preprocessing of the image information may include operations such as scaling and normalizing the image.

[0316] Step 4.2: Use the image category prediction model to determine the category to which the product corresponding to the image information belongs.

[0317] Step 4.3: Use the image information as the input of the discriminant model, and execute the discriminant model to obtain the discrimination result.

[0318] Step 4.4: When the discrimination result indicates that the image information belongs to a simple type, the image information and the category to which the corresponding product belongs are used as inputs of a calculation model, and the calculation model is executed to obtain a reference price of the product.

[0319] Step 4.5: When the discrimination result indicates that the image information belongs to the difficult type, at least one reference sample that meets the similarity requirement with the image information is retrieved based on the image feature index library.

[0320] Step 4.6: Determine a reference price for the product based on the price corresponding to each of the at least one reference samples.

[0321] Step 4.7: When no reference sample that meets the similarity requirement with the image information is retrieved based on the image feature index library, a prompt message is outputted indicating that a reference price is refused to be determined for the product.

[0322] Step 4.8: Obtain historical information related to the product.

[0323] Step 4.9: Based on the historical information, the reference price of the commodity is revised.

[0324] Step 4.10: Output the revised reference price.

[0325] Figure 9 FIG. 1 shows a schematic diagram of the structure of an information processing device provided by an embodiment of the present application. Figure 9 As shown, the information processing device includes an acquisition module 11, a determination module 12, and a display module 13. The acquisition module 11 is configured to acquire image information of a target object and, based on the image information, determine the category to which the target object belongs. The determination module 12 is configured to determine a reference price for the target object based on the image information and the category to which the target object belongs. The display module 13 is configured to display the reference price.

[0326] The technical solution provided in this embodiment determines a reference price for the target object based only on the image information of the target object. This reference price can be a specific price or a price range. In this way, the user who publishes the target object can set a more appropriate price for the target object based on the reference price.

[0327] Furthermore, the determination module 12 is further configured to: when the discriminant model determines that the image information belongs to the simple type, use the image information and the category to which the target object belongs as inputs to a calculation model, and execute the calculation model to obtain the reference price. The discriminant model is trained based on sample images and the difficulty category to which the sample images belong; and the calculation model is trained based on images of previously traded goods, their categories, and transaction prices.

[0328] Furthermore, the determination module 12 is also used to: when the discriminant model is used to determine that the image information belongs to the difficult type, based on the image feature index library, retrieve at least one reference sample that meets the similarity requirements with the image information; and determine the reference price of the target object according to the price corresponding to each of the at least one reference sample.

[0329] Furthermore, the display module 13 is further configured to display a prompt indicating that a reference price is refused to be determined for the target object when no reference sample that meets the similarity requirement with the image information is retrieved based on the image feature index library.

[0330] It should be noted here that the information processing device provided in the above embodiments can implement the technical solutions described in the above method embodiments. The specific implementation principles of the above modules or units can be found in the corresponding contents of the above method embodiments, which will not be repeated here.

[0331] Another embodiment of the present application also provides an information processing device. The structure of the information processing device is similar to the above Figure 9 The structure shown is the same. Specifically, the information processing device provided in this embodiment includes: an acquisition module, a determination module and a display module. The acquisition module is used to acquire image information of the target object; based on the image information, acquire the category to which the target object belongs. The determination module is used to determine a reference price for the target object and auxiliary information that causes the target object to meet a first preset condition based on the image information and the category to which the target object belongs. The display module is used to associate the reference price and the auxiliary information with the image information and display them respectively.

[0332] Furthermore, the auxiliary information includes at least one of the following:

[0333] Suggestions for modifications to the image information;

[0334] Recommendations on combination objects that can form a combination relationship with the target object;

[0335] Suggestions for improvements related to the stated target object;

[0336] Suggestions for adding descriptive information to the target object.

[0337] Figure 10 FIG. 1 shows a schematic diagram of the structure of an information processing device provided by an embodiment of the present application. Figure 10 As shown, the information processing device includes an acquisition module 21 and an execution module 22. The acquisition module 21 is configured to acquire product image information. The execution module 22 is configured to use the image information as input to a calculation model and execute the calculation model to obtain a reference price for the product. The calculation model is trained based on images and transaction prices of previously traded products.

[0338] The technical solution provided in this embodiment utilizes a computing model to directly obtain a reference price for a commodity based on its image information; the computing model is trained based on the images and transaction prices of commodities that have been traded in the past; the user who publishes the commodity can set a more appropriate price for the commodity based on the reference price.

[0339] Furthermore, the execution module 22 is further configured to:

[0340] Using the image information as input to the category determination model, executing the category determination model to obtain category information to which the product belongs;

[0341] The image information and the category information are used as inputs of the calculation model, and the calculation model is executed to obtain the reference price.

[0342] Furthermore, the information processing device further includes:

[0343] The execution module 22 is further configured to use the image information as input to a discrimination model and execute the discrimination model to obtain a discrimination result;

[0344] a triggering module, configured to trigger the calculation model to complete the calculation of the reference price of the commodity when the discrimination result indicates that the image information belongs to a simple type;

[0345] The discriminant model is obtained by training based on sample images and the difficulty level of the sample images.

[0346] Furthermore, the information processing module also includes:

[0347] a retrieval module configured to retrieve, based on an image feature index library, at least one reference sample that meets similarity requirements with the image information when the discrimination result indicates that the image information belongs to a difficult type;

[0348] The determination module is used to determine the reference price of the commodity according to the price corresponding to each of the at least one reference samples.

[0349] Furthermore, the information processing module also includes:

[0350] The output module is used to output a prompt content of refusing to determine the reference price for the commodity when no reference sample that meets the similarity requirement with the image information is retrieved based on the image feature index library.

[0351] Furthermore, the information processing module also includes:

[0352] The acquisition module 21 is used to acquire historical information related to the product;

[0353] A correction module is used to correct the reference price based on the historical information.

[0354] Furthermore, the historical information includes at least one of the following: a historical publishing price of the user who published the product, and a historical publishing price of products belonging to the same category as the product.

[0355] It should be noted here that the information processing device provided in the above embodiments can implement the technical solutions described in the above method embodiments. The specific implementation principles of the above modules or units can be found in the corresponding contents of the above method embodiments, which will not be repeated here.

[0356] Figure 11 FIG. 1 shows a schematic diagram of the structure of an information processing device provided by another embodiment of the present application. Figure 11 As shown, the information processing device includes: an acquisition module 31, a retrieval module 32, and a determination module 33. The acquisition module 31 is used to acquire image information of a product. The retrieval module 32 is used to retrieve at least one reference sample that meets the similarity requirement with the image information based on an image feature index library when the discriminant model determines that the image information belongs to the difficult type. The determination module 33 is used to determine the reference price of the product based on the price corresponding to each of the at least one reference sample. The discriminant model is trained based on sample images and the difficulty type to which the sample images belong.

[0357] In the technical solution provided by this embodiment, when the discriminant model is used to determine that the image information of a product belongs to a difficult type, similar samples are retrieved based on the image feature index library; then the reference price of the product is determined based on the reference prices corresponding to the similar samples; this solution increases the discrimination of the difficulty type of the image information, and then selectively selects a suitable determination method to obtain the reference price of the product, with high accuracy.

[0358] Furthermore, the retrieval module 32 is further configured to:

[0359] extracting image features based on the image information;

[0360] Retrieving at least one candidate sample from the image feature index library according to the image feature;

[0361] Calculating the similarity value between the image information and each candidate sample respectively;

[0362] The candidate samples whose similarity values ​​are greater than the second threshold value are reference samples that meet the similarity requirement with the image information.

[0363] Furthermore, the information processing device provided in this embodiment may further include:

[0364] The collection module is used to collect commodity information of the traded commodities, wherein the commodity information includes: commodity image information and transaction price;

[0365] A determination module, configured to use product information that meets a first preset requirement as a simple sample; and use product information that meets a second preset requirement as a difficult sample;

[0366] A construction module is used to construct the image feature index library according to the difficult samples.

[0367] Furthermore, the determining module 33 is further configured to:

[0368] Get the estimated model;

[0369] Taking the product image as input of the estimation model, executing the estimation model to obtain estimation parameters;

[0370] When the difference between the estimated parameter and the transaction price is less than a first threshold, the product information meets the first preset requirement;

[0371] When the difference between the estimated parameter and the transaction price is greater than or equal to the first threshold, the product information meets the second preset requirement.

[0372] It should be noted here that the information processing device provided in the above embodiments can implement the technical solutions described in the above method embodiments. The specific implementation principles of the above modules or units can be found in the corresponding contents of the above method embodiments, which will not be repeated here.

[0373] Figure 12 The model training device provided by an embodiment of the present application is shown. The model training device includes: a first acquisition module 41, a first execution module 42 and a first optimization module 43. The first acquisition module 41 is used to obtain a first training sample; wherein, the first training sample includes: a product image and a transaction price of a traded product. The first execution module 42 is used to use the product image as the input of the computational model to be trained, and execute the computational model to obtain a first output result. The first optimization module 43 is used to optimize the parameters of the computational model according to the first output result and the transaction price. The computational model that has completed training is used to determine a reference price for the product based on the image information of the product.

[0374] Furthermore, the first training sample also includes category information of the traded goods. Accordingly, the first execution module 42 is further configured to use the image information and the category information as inputs of the calculation model and execute the calculation model to obtain a first output result.

[0375] Furthermore, the first execution module 42 is further configured to: use the image information as input of a category determination model to be trained, execute the category determination model to obtain a second output result;

[0376] The first optimization module 43 is further configured to: optimize parameters of the category determination model according to the second output result and the category information;

[0377] The trained category determination model is used to determine the category information of a product based on its image information.

[0378] Furthermore, the model training device provided in this embodiment also includes:

[0379] A collection module is used to collect commodity information of traded commodities, wherein the commodity information includes: commodity images and transaction prices;

[0380] The determination module is configured to use the commodity information that meets the first preset requirement as a simple sample; use the commodity information that meets the second preset requirement as a difficult sample; and use the simple sample as the first training sample.

[0381] Furthermore, the first acquisition module 41 is also used to: obtain an estimation model; the first execution module 42 is also used to: use the product image as the input of the estimation model, execute the estimation model to obtain estimation parameters; the determination module is also used to: when the difference between the estimation parameter and the transaction price is less than a first threshold, the product information meets the first preset requirement; when the difference between the estimation parameter and the transaction price is greater than or equal to the first threshold, the product information meets the second preset requirement.

[0382] It should be noted here that the model training device provided in the above embodiments can implement the technical solutions described in the above method embodiments. The specific implementation principles of the above modules or units can be found in the corresponding contents in the above method embodiments, which will not be repeated here.

[0383] Figure 13 FIG. 1 shows a schematic diagram of the structure of a model training device provided in an embodiment of the present application. Figure 13As shown, the model training device includes: a second acquisition module 51, a second execution module 52 and a second optimization module 53. The second acquisition module 51 is used to obtain a second training sample, wherein the second training sample includes: a product image of a traded product and the difficulty type to which the product image belongs. The second execution module 52 is used to use the product image as the input of the discriminant model to be trained, and execute the discriminant model to obtain a third output result. The second optimization module 53 is used to optimize the parameters of the discriminant model based on the third output result and the difficulty type. The trained discriminant model is used to determine whether the image information belongs to a simple type based on the image information of the product, and when the image information belongs to a simple type, the calculation model can be used to complete the calculation of the reference price of the product based on the image information.

[0384] Furthermore, the model training device provided in this embodiment also includes:

[0385] A collection module is used to collect commodity information of traded commodities, wherein the commodity information includes: commodity images and transaction prices;

[0386] The second execution module 52 is further configured to use the product image as input to the estimation model and execute the estimation model to obtain estimation parameters;

[0387] The marking module is used to mark the product image as a simple type when the difference between the estimated parameter and the transaction price is less than a first threshold; and mark the product image as a difficult type when the difference between the estimated parameter and the transaction price is greater than or equal to the first threshold.

[0388] It should be noted here that the model training device provided in the above embodiments can implement the technical solutions described in the above method embodiments. The specific implementation principles of the above modules or units can be found in the corresponding contents in the above method embodiments, which will not be repeated here.

[0389] Figure 14 FIG. 1 shows a schematic diagram of the structure of a feature library creation device provided by an embodiment of the present application. Figure 14 As shown, the feature library creation device includes: a collection module 61, a differentiation module 62, and a creation module 63. The collection module 61 is used to collect product information of traded products; wherein the product information includes product images and transaction prices. The differentiation module 62 is used to use an estimation model to distinguish some product information as difficult samples from the collected multiple product information. The creation module 63 is used to create an image feature index library based on the difficult samples, which is used to retrieve reference samples that meet the similarity requirements with the image information of the target product, so as to determine the reference price of the target product based on the price corresponding to the reference samples.

[0390] Furthermore, the distinguishing module 62 is further configured to:

[0391] Taking a product image of product information as input of the estimation model, executing the estimation model to obtain estimation parameters;

[0392] When the difference between the estimated parameter and the transaction price of the product information is greater than or equal to a first threshold, the product information is used as a difficult sample.

[0393] It should be noted here that the feature library creation device provided in the above embodiment can implement the technical solutions described in the above method embodiments. The specific implementation principles of the above modules or units can be found in the corresponding contents in the above corresponding method embodiments, which will not be repeated here.

[0394] Figure 15 FIG. 1 shows a schematic diagram of the structure of an electronic device provided by an embodiment of the present application. Figure 15 As shown, the electronic device includes a memory 71, a processor 72 and a display 74; wherein,

[0395] The memory 71 is used to store programs;

[0396] The processor 72 is coupled to the memory 71 and is configured to execute the program stored in the memory 71 to:

[0397] Obtain image information of the target object;

[0398] Based on the image information, obtaining the category to which the target object belongs;

[0399] Determining a reference price of the target object based on the image information and the category to which the target object belongs;

[0400] The display is controlled to display the reference price.

[0401] The memory 71 can be configured to store various other data to support operations on the electronic device. Examples of such data include instructions for any application or method operating on the electronic device. The memory 71 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disk.

[0402] When executing the program in the memory 71 , the processor 72 may implement other functions in addition to the above functions. For details, please refer to the description of the previous embodiments.

[0403] Further, if Figure 15 As shown, the electronic device also includes: a communication component 73, a display 74, a power component 75, an audio component 76 and other components. Figure 15 Only some components are shown schematically, which does not mean that the electronic device only includes Figure 15 Components shown.

[0404] Another embodiment of the present application provides an electronic device. The structure of the electronic device is similar to the above electronic device embodiment, and can be seen in the above Figure 15 The electronic device includes a memory and a processor; wherein,

[0405] The memory is used to store programs;

[0406] The processor is coupled to the memory and is configured to execute the program stored in the memory to:

[0407] Get the image information of the product;

[0408] Using the image information as input to a calculation model, executing the calculation model to obtain a reference price for the product;

[0409] The calculation model is trained based on the commodity images and transaction prices of the traded commodities.

[0410] In addition to the above functions, the processor can also implement other functions when executing the program in the memory. For details, please refer to the description of the previous embodiments.

[0411] Another embodiment of the present application provides an electronic device. The structure of the electronic device is similar to the above electronic device embodiment, and can be seen in the above Figure 15 The electronic device includes a memory and a processor; wherein,

[0412] The memory is used to store programs;

[0413] The processor is coupled to the memory and is configured to execute the program stored in the memory to:

[0414] Get the image information of the product;

[0415] When the discriminant model is used to determine that the image information belongs to the difficult type, at least one reference sample that meets the similarity requirement with the image information is retrieved based on the image feature index library;

[0416] Determining a reference price for the product based on a price corresponding to each of the at least one reference samples;

[0417] The discriminant model is obtained by training based on sample images and the difficulty level of the sample images.

[0418] In addition to the above functions, the processor can also implement other functions when executing the program in the memory. For details, please refer to the description of the previous embodiments.

[0419] Accordingly, an embodiment of the present application further provides a computer-readable storage medium storing a computer program, which, when executed by a computer, can implement the steps or functions of the information processing method provided in the above embodiments.

[0420] Another embodiment of the present application provides an electronic device. The structure of the electronic device is similar to the above electronic device embodiment, and can be seen in the above Figure 15 The electronic device includes a memory and a processor; wherein,

[0421] The memory is used to store programs;

[0422] The processor is coupled to the memory and is configured to execute the program stored in the memory to:

[0423] Obtaining a first training sample; wherein the first training sample includes: a commodity image and a transaction price of a traded commodity;

[0424] Using the product image as input of a computational model to be trained, and executing the computational model to obtain a first output result;

[0425] Optimizing parameters of the calculation model according to the first output result and the transaction price;

[0426] The trained computing model is used to determine the reference price of the commodity based on the image information of the commodity.

[0427] In addition to the above functions, the processor can also implement other functions when executing the program in the memory. For details, please refer to the description of the previous embodiments.

[0428] Another embodiment of the present application provides an electronic device. The structure of the electronic device is similar to that of the above electronic device embodiment. Figure 15 The electronic device includes: a memory and a processor; wherein,

[0429] The memory is used to store programs;

[0430] The processor is coupled to the memory and is configured to execute the program stored in the memory to:

[0431] Obtaining a second training sample, wherein the second training sample includes: an image of a traded commodity and a difficulty category of the commodity image;

[0432] Using the product image as an input of a discriminant model to be trained, and executing the discriminant model to obtain a third output result;

[0433] Optimizing the parameters of the discriminant model according to the third output result and the difficulty type;

[0434] The trained discriminant model is used to determine, based on the image information of the commodity, whether the image information can be used to calculate the reference price of the commodity based on the image information using the calculation model.

[0435] In addition to the above functions, the processor can also implement other functions when executing the program in the memory. For details, please refer to the description of the previous embodiments.

[0436] Accordingly, an embodiment of the present application also provides a computer-readable storage medium storing a computer program, which, when executed by a computer, can implement the steps or functions of the model training method provided in the above embodiments.

[0437] Another embodiment of the present application provides an electronic device. The structure of the electronic device is similar to the above electronic device embodiment, and can be seen in the above Figure 15 The electronic device includes a memory and a processor; wherein,

[0438] The memory is used to store programs;

[0439] The processor is coupled to the memory and is configured to execute the program stored in the memory to:

[0440] Collecting commodity information of traded commodities, wherein the commodity information includes commodity images and transaction prices;

[0441] Using the estimation model, we can identify some difficult product samples from the collected product information.

[0442] Based on difficult samples, an image feature index library is created to retrieve reference samples that meet similarity requirements with the image information of the target product, so as to determine the reference price of the target product based on the price corresponding to the reference sample.

[0443] In addition to the above functions, the processor can also implement other functions when executing the program in the memory. For details, please refer to the description of the previous embodiments.

[0444] Accordingly, an embodiment of the present application further provides a computer-readable storage medium storing a computer program, which, when executed by a computer, can implement the steps or functions of the feature library creation method provided in the above embodiments.

[0445] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0446] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, or of course, by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the existing technology can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or certain parts of the embodiments.

[0447] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. An information processing method, characterized in that: include: Obtain image information of the target object; Based on the image information, obtaining the category to which the target object belongs; Determining a reference price of the target object according to the image information and the category to which the target object belongs, including: When the discriminant model determines that the image information belongs to a simple type, the image information and the category to which the target object belongs are used as inputs of a calculation model, and the calculation model is executed to obtain a reference price of the target object; wherein the calculation model is trained based on the product image, product category, and transaction price of the first traded product; When the discriminant model determines that the image information belongs to the difficult type, at least one reference sample that meets the similarity requirement with the image information is retrieved based on the image feature index library; and a reference price of the target object is determined based on the price corresponding to each of the at least one reference sample; The simple type is used to indicate that the target object is suitable for price prediction using the calculation model, and the difficult type is used to indicate that the target object is not suitable for price prediction using the calculation model; The reference price is displayed.

2. The method according to claim 1, characterized in that The discriminant model is trained based on the sample image of the second traded commodity and the difficulty type of the sample image; when the difference between the transaction price of the second traded commodity and the estimated parameter estimated by the estimation model based on the sample image of the second traded commodity is less than a first threshold, the difficulty type to which the sample image of the second traded commodity belongs is a simple type; when the difference between the transaction price of the second traded commodity and the estimated parameter estimated by the estimation model based on the sample image of the second traded commodity is greater than or equal to the first threshold, the difficulty type to which the sample image of the second traded commodity belongs is a difficult type.

3. The method according to claim 2, characterized in that Also includes: When no reference sample that meets the similarity requirement with the image information is retrieved based on the image feature index library, a prompt content is displayed to refuse to determine the reference price for the target object.

4. The method according to any one of claims 1 to 3, characterized in that Also includes: Determine the garbage category to which the target object belongs according to the category to which the target object belongs; Displays the garbage category to which the target object belongs.

5. The method according to any one of claims 1 to 3, characterized in that Also includes: According to the preset category list, determining whether the category to which the target object belongs meets a second preset condition; Based on the judgment result, relevant suggestions are displayed.

6. The method according to any one of claims 1 to 3, characterized in that Also includes: Determining a reference price of at least one component unit constituting the target object according to a reference price of the target object and a category to which the target object belongs; A reference price of the at least one component unit is displayed.

7. An information processing method, characterized in that: include: Get the image information of the product; When the discriminant model determines that the image information belongs to a simple type, the image information is used as input to a calculation model, and the calculation model is executed to obtain a reference price of the product; wherein the calculation model is trained based on the product image and transaction price of the first traded product; When the discriminant model determines that the image information belongs to the difficult type, searching for at least one reference sample that meets the similarity requirement with the image information based on the image feature index library; determining a reference price of the product based on the price corresponding to each of the at least one reference sample; The simple type is used to indicate that the commodity is suitable for price prediction using the calculation model, and the difficult type is used to indicate that the commodity is not suitable for price prediction using the calculation model.

8. The method according to claim 7, characterized in that Also includes: Using the image information as input to a category determination model, executing the category determination model to obtain the category to which the product belongs; When the discriminant model is used to determine that the image information belongs to a simple type, the image information is used as an input of a calculation model, and the calculation model is executed to obtain a reference price of the product, including: When the discriminant model determines that the image information belongs to a simple type, the image information and the category to which the product belongs are used as inputs of a calculation model, and the calculation model is executed to obtain a reference price of the product; The calculation model is trained based on the product images, product categories and transaction prices of the traded products.

9. The method according to claim 7 or 8, characterized in that The discriminant model is trained based on sample images and the difficulty type to which the sample images belong. When the difference between the transaction price of the second traded commodity and the estimated parameter estimated by the estimation model based on the sample image of the second traded commodity is less than a first threshold, the difficulty type to which the sample image of the second traded commodity belongs is a simple type; when the difference between the transaction price of the second traded commodity and the estimated parameter estimated by the estimation model based on the sample image of the second traded commodity is greater than or equal to the first threshold, the difficulty type to which the sample image of the second traded commodity belongs is a difficult type.

10. The method according to claim 9, characterized in that Also includes: When no reference sample that meets the similarity requirement with the image information is retrieved based on the image feature index library, a prompt content of refusing to determine the reference price for the product is output.

11. The method according to claim 7 or 8, characterized in that Also includes: Obtain historical information related to the product; Based on the historical information, the reference price is revised.

12. The method according to claim 11, characterized in that The historical information includes at least one of the following: the historical publishing price of the user who published the product, and the historical publishing price of products belonging to the same category as the product.

13. An electronic device, characterized in that: including a memory, a processor and a display; wherein, The memory is used to store programs; The processor is coupled to the memory and is configured to execute the program stored in the memory to: Obtain image information of the target object; Based on the image information, obtaining the category to which the target object belongs; Determining a reference price of the target object according to the image information and the category to which the target object belongs, including: When the discriminant model determines that the image information belongs to a simple type, the image information and the category to which the target object belongs are used as inputs of a calculation model, and the calculation model is executed to obtain a reference price of the target object; wherein the calculation model is trained based on the product image, product category, and transaction price of the first traded product; When the discriminant model determines that the image information belongs to the difficult type, at least one reference sample that meets the similarity requirement with the image information is retrieved based on the image feature index library; and a reference price of the target object is determined based on the price corresponding to each of the at least one reference sample; The simple type is used to indicate that the target object is suitable for price prediction using the calculation model, and the difficult type is used to indicate that the target object is not suitable for price prediction using the calculation model; The display is controlled to display the reference price.

14. An electronic device, characterized in that: including a memory and a processor; wherein, The memory is used to store programs; The processor is coupled to the memory and is configured to execute the program stored in the memory to: Get the image information of the product; When the discriminant model determines that the image information belongs to a simple type, the image information is used as an input of a calculation model, and the calculation model is executed to obtain a reference price of the product; The calculation model is trained based on the product image and transaction price of the first traded product; When the discriminant model determines that the image information belongs to the difficult type, searching for at least one reference sample that meets the similarity requirement with the image information based on the image feature index library; determining a reference price of the product based on the price corresponding to each of the at least one reference sample; The simple type is used to indicate that the commodity is suitable for price prediction using the calculation model, and the difficult type is used to indicate that the commodity is not suitable for price prediction using the calculation model.

15. An information processing method, characterized in that: include: Obtain image information of the target object; Based on the image information, obtaining the category to which the target object belongs; Determining a reference price for the target object and auxiliary information that causes the target object to meet a first preset condition based on the image information and the category to which the target object belongs, including: When the discriminant model determines that the image information belongs to a simple type, the image information and the category to which the target object belongs are used as inputs of a calculation model, and the calculation model is executed to obtain a reference price of the target object; wherein the calculation model is trained based on the product image, product category, and transaction price of the first traded product; When the discriminant model determines that the image information belongs to the difficult type, at least one reference sample that meets the similarity requirement with the image information is retrieved based on the image feature index library; and a reference price of the target object is determined based on the price corresponding to each of the at least one reference sample; The simple type is used to indicate that the target object is suitable for price prediction using the calculation model, and the difficult type is used to indicate that the target object is not suitable for price prediction using the calculation model; The reference price and the auxiliary information are respectively associated with the image information and displayed.

16. The method according to claim 15, characterized in that The auxiliary information includes at least one of the following: Suggestions for modifications to the image information; Recommendations on combination objects that can form a combination relationship with the target object; Suggestions for improvements related to the stated target object; Suggestions for adding descriptive information to the target object.

Images