Image processing method, computer equipment and computer readable storage medium

Abstract

The embodiment of the invention provides an image processing method, computer equipment and a computer readable storage medium, and the method comprises the steps: obtaining annotation object type information and an original image, and the original image comprises a to-be-annotated object; performing format adjustment processing on the original image to obtain a to-be-labeled image, the image quality of the to-be-labeled image being matched with the image quality of the original image; the annotation object type information and the to-be-annotated image are input into an image annotation model to be processed, target annotation information of the to-be-annotated object is obtained, and the target annotation information comprises one or more of name information, type information, object number information and object position indication information. Through the method provided by the embodiment of the invention, the efficiency and accuracy of image annotation can be effectively improved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to an image processing method, a computer device, and a computer-readable storage medium. Background Art

[0002] Image annotation refers to the process of marking the content in an image. The image annotation information can enable the computer system to quickly identify the image. With the development of computer technology, more and more image processing technologies need to use image annotation information.

[0003] Usually, the manual annotation method is used to determine the annotation information of the image. This method requires professional technicians to observe the image and manually add labels or annotations to the image. This method heavily relies on the professional technical level and annotation experience of professional technicians, resulting in low efficiency of image annotation and difficulty in meeting the accuracy requirements. Summary of the invention

[0004] The embodiments of the present application provide an image processing method, a computer device, and a computer-readable storage medium, which can effectively improve the efficiency and accuracy of image annotation.

[0005] On the one hand, an embodiment of the present application provides an image processing method, the method comprising:

[0006] Acquire the type information of the labeled object and the original image, wherein the original image includes the object to be labeled;

[0007] Performing format adjustment processing on the original image to obtain an image to be annotated, wherein the image quality of the image to be annotated matches the image quality of the original image;

[0008] Inputting the labeled object type information and the image to be labeled into an image labeling model for processing to obtain target labeling information of the object to be labeled, wherein the target labeling information includes one or more of name information, type information, object quantity information, and object position indication information;

[0009] Among them, the image annotation model is obtained by training the initial annotation model using the adjusted sample annotation information, the adjusted sample annotation information is obtained by adjusting the sample annotation information in the rendered image, the rendered image is obtained by rendering the sample annotation information and the initial sample image, the sample annotation information is obtained by inputting the sample image to be annotated and the annotation object type information into the initial annotation model for processing, and the sample image to be annotated is obtained by formatting the initial sample image.

[0010] On the one hand, an embodiment of the present application provides an image processing device, the device comprising:

[0011] An acquisition unit, used to acquire the type information of the labeled object and the original image, wherein the original image includes the object to be labeled;

[0012] An adjustment unit, configured to perform format adjustment processing on the original image to obtain an image to be annotated, wherein the image quality of the image to be annotated matches the image quality of the original image;

[0013] a processing unit, configured to input the labeled object type information and the image to be labeled into an image labeling model for processing to obtain target labeling information of the object to be labeled, wherein the target labeling information includes one or more of name information, type information, object quantity information, and object position indication information;

[0014] Among them, the image annotation model is obtained by training the initial annotation model using the adjusted sample annotation information, the adjusted sample annotation information is obtained by adjusting the sample annotation information in the rendered image, the rendered image is obtained by rendering the sample annotation information and the initial sample image, the sample annotation information is obtained by inputting the sample image to be annotated and the annotation object type information into the initial annotation model for processing, and the sample image to be annotated is obtained by formatting the initial sample image.

[0015] On the one hand, an embodiment of the present application provides a computer device, comprising: a processor, a communication interface and a memory, wherein the processor, the communication interface and the memory are interconnected, wherein the memory stores an executable program code, and the processor is used to call the executable program code to implement the image processing method provided in the embodiment of the present application.

[0016] Accordingly, an embodiment of the present application further provides a computer-readable storage medium, in which instructions are stored. When the computer-readable storage medium is executed on a computer, the computer implements the image processing method provided by the embodiment of the present application.

[0017] Accordingly, the embodiment of the present application further provides a computer program product, the computer program product includes a computer program or a computer instruction, and the computer program or the computer instruction is stored in a computer-readable storage medium. The processor of the computer device reads the computer program or the computer instruction from the computer-readable storage medium, and the processor executes the computer program or the computer instruction, so that the computer device implements the image processing method provided in the embodiment of the present application.

[0018] In the present application, the type information of the labeled object and the original image including the object to be labeled can be obtained, and the format of the original image can be adjusted to obtain the image to be labeled; the type information of the labeled object and the image to be labeled are input into the image annotation model for processing, and the target annotation information of the object to be labeled can be obtained, and the target annotation information includes one or more of the name information, type information, object quantity information and object position indication information. Through the image processing method provided by the embodiment of the present application, the format of the original image can be adjusted to obtain the image to be labeled, which ensures the universality of the method provided by the present application; the image annotation model can be used to quickly determine the name information, type information, object quantity information and object position indication information of the object in the image, which can effectively improve the efficiency of image annotation; because the image annotation model is obtained by model training based on the adjusted sample annotation information, the prediction accuracy of the image annotation model is high, and the annotation information obtained using the image annotation model has high accuracy; because the method does not rely on professional and technical personnel, the cost is low and the resource consumption is low. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0020] Figure 1 is a schematic diagram of the system architecture of an image processing system provided in an embodiment of the present application;

[0021] Figure 2 It is a flowchart of an image processing method provided in an embodiment of the present application;

[0022] Figure 3 is a schematic diagram of an image processing system provided by an embodiment of the present application;

[0023] Figure 4 It is a flowchart of a model training method provided in an embodiment of the present application;

[0024] Figure 5 is a schematic diagram of a rendered image provided by an embodiment of the present application;

[0025] Figure 6 is a schematic diagram of a labeling information adjustment interface provided in an embodiment of the present application;

[0026] Figure 7 is a schematic diagram of a model training method provided in an embodiment of the present application;

[0027] Figure 8 is a structural block diagram of an image processing device provided in an embodiment of the present application;

[0028] Fig. 9 It is a structural block diagram of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0029] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0030] It should be noted that the descriptions of "first", "second", etc. involved in the embodiments of the present application are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the technical features defined as "first" or "second" may explicitly or implicitly include at least one of the features.

[0031] In some solutions, the way to obtain image annotation information is usually for professional technicians to observe the image and manually add labels or annotations to the image. This method is inefficient and inaccurate in obtaining annotation information, and is prone to mislabeling or missing labels.

[0032] Based on this, the embodiment of the present application provides an image processing method, which can obtain the type information of the annotation object and the original image, the original image includes the object to be annotated; the format of the original image is adjusted to obtain the image to be annotated, and the image quality of the image to be annotated matches the image quality of the original image; the annotation object type information and the image to be annotated are input into the image annotation model for processing to obtain the target annotation information of the object to be annotated, and the target annotation information includes one or more of the name information, type information, object quantity information and object position indication information; wherein the image annotation model is obtained by training the initial annotation model using the adjusted sample annotation information, the adjusted sample annotation information is obtained by adjusting the sample annotation information in the rendered image, the rendered image is obtained by rendering the sample annotation information and the initial sample image, the sample annotation information is obtained by inputting the sample image to be annotated and the annotation object type information into the initial annotation model for processing, and the sample image to be annotated is obtained by formatting the initial sample image. Through the method provided by the embodiment of the present application, the efficiency of obtaining the annotation information of the image can be effectively improved, and the accuracy of the obtained annotation information can also be effectively improved.

[0033] The image processing method provided in the embodiment of the present application can also be applied to the field of cloud computing. Cloud computing is a computing model that distributes computing tasks on a resource pool composed of a large number of computers, so that various application systems can obtain computing power, storage space and information services as needed. The cloud computing resource pool mainly includes: computing devices (virtualized machines, including operating systems), storage devices, and network devices. Cloud computing technology can be used to implement the image annotation model in this application: the annotation object type information and the original object can be obtained, and the original object can be formatted to obtain the image to be annotated; the annotation object type information is input into the image annotation model implemented by cloud computing technology for processing to obtain the annotation information of the original image. Through the method provided in the embodiment of the present application, the efficiency of obtaining annotation data can be effectively improved, and at the same time, the accuracy of the acquired annotation data can be improved.

[0034] The architecture of the image processing system provided in the embodiments of the present application will be introduced below in conjunction with the accompanying drawings.

[0035] See also Figure 1 , which is a schematic diagram of the system architecture of an image processing system provided in an embodiment of the present application, the image processing system includes a terminal device 101, an image processing device 102 and a database 103, the image processing device 102 can exchange data with the terminal device 101 and the database 103, and the image processing device 102 includes an image annotation module. Among them:

[0036] The terminal device 101 can send the original image and the annotated object type information to the image processing device 102. The terminal device 101 can take images and receive images sent by other devices. The terminal device 101 can be a handheld device with communication function (such as a smart phone, a tablet computer), a computing device (such as a personal computer (PC), a vehicle terminal, an intelligent voice interaction device, a wearable device or other intelligent device, but is not limited thereto.

[0037] The image processing device 102 can receive the original image and the annotation object type information sent by the terminal device 101, and process the original image to obtain the annotation information. The image processing device 102 includes an image annotation module, which is used to generate the annotation information of the image. The image annotation module includes an image annotation model, which is obtained by training the initial annotation model using the adjusted sample annotation information, the adjusted sample annotation information is obtained by adjusting the sample annotation information in the rendered image, the rendered image is obtained by rendering the sample annotation information and the initial sample image, the sample annotation information is obtained by inputting the sample image to be annotated and the annotation object type information into the initial annotation model for processing, and the sample image to be annotated is obtained by format adjustment processing of the initial sample image. The image processing device 102 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs), and basic cloud computing services such as big data and artificial intelligence platforms.

[0038] The database 103 is used to store relevant data of the image processing device 102, such as: annotation object type information, original image, annotation information, etc. The database 103 can be a local database in the image processing device 102, or a cloud database associated with the image processing device 102 (i.e., a database deployed in the cloud). Specifically, it can be deployed based on any of private clouds, public clouds, hybrid clouds, edge clouds, etc., so that the cloud database focuses on different functions. For example, the database deployed in a private cloud has the user's personal equipment as the basic cloud hardware, and focuses more on serving a small number of users, while the database deployed in the public cloud is deployed based on a cloud platform provided by a third party, which allows the data stored in the database to achieve data sharing. Any user's data can be stored in the database, and any user can also use the data in the database.

[0039] The following will explain in detail Figure 1 The working principle of the image processing system shown:

[0040] The terminal device 101 captures or receives an original image (the original image includes the object to be annotated), and sends the original image and the type information of the annotated object to the image processing device 102; the image processing device 102 receives the type information of the annotated object and the original image, and performs format adjustment processing on the original image to obtain the image to be annotated, and the image quality of the image to be annotated matches the image quality of the original image; the image processing device 102 can input the type information of the annotated object and the image to be annotated into the image annotation module for processing to obtain the target annotation information of the object to be annotated, and the target annotation information can include one or more of name information, type information, object quantity information and object position indication information. The image processing device 102 stores the original image and the target annotation information in the database 103. The image processing device can determine the credit certification data of the terminal device 101 according to the name information, type information and object quantity information in the target annotation information, and determine the credit limit according to the credit certification data of the terminal device 101, and send the credit limit to the terminal device 101. The image processing method provided in the embodiment of the present application can effectively improve the efficiency of image annotation and the accuracy of image annotation information. At the same time, the credit limit can be determined based on the image annotation information, which is conducive to achieving reasonable credit.

[0041] It is to be understood that the schematic diagram of the image processing system described in the embodiment of the present application is to more clearly illustrate the image processing method of the embodiment of the present application, and does not constitute a limitation on the image processing method provided in the embodiment of the present application. For example, the image processing method provided in the embodiment of the present application can be performed by the image processing device 102, and can also be performed by other devices that are different from the image processing device 102 and can communicate with the terminal device 101 and the database 103. It is known to those skilled in the art that Figure 1 The number of terminal devices 101, image processing devices 102, and databases 103 is merely illustrative. Any number of devices and nodes may be configured according to business implementation requirements. Moreover, with the evolution of system architecture and the emergence of new business scenarios, the image processing method provided in the embodiment of the present application is also applicable to similar technical problems.

[0042] See also Figure 2 , Figure 2 A schematic diagram of a flow chart of an image processing method provided in an embodiment of the present application. The image processing method can be implemented by the above-mentioned image processing device 102, or by other devices. The flow chart of the image processing method provided in an embodiment of the present application includes but is not limited to:

[0043] S201: Acquire annotation object type information and an original image, where the original image includes an object to be annotated.

[0044] In the embodiment of the present application, the original image may be an image taken for a specific scene, for example, the original image may be an image of a production workshop, an image of a warehouse where items are placed, or an image of a farm. The original image may include one or more objects to be annotated, and the annotated object type information is used to indicate the type of the object to be annotated, for example, the original image is an image of a warehouse where bicycles are placed, the annotated object type information may be "bicycle", and the object to be annotated is the bicycle in the original image. The object to be annotated in the original image may be a means of transportation such as a bicycle or an electric vehicle, may be a household appliance, or may be a breeding product, such as a cow or a sheep. The annotated object type information may be externally input, may be received together with the original image, or may be configured by the image processing device itself. In some cases, it is necessary to understand the annotation information such as the number and position of the object to be annotated in the original image, then the method provided in the embodiment of the present application may be used to obtain the annotation information of the object to be annotated, effectively improving the efficiency of obtaining the annotation information.

[0045] In one embodiment, the implementation method of obtaining the original image may be: determining the access information of the image database, the access information includes the access address and access authentication information of the image database; performing detection processing on the image database according to the access information to obtain the detection result; if the detection result indicates that there is a new image in the image database, the new image is determined as the original image. The original image can be obtained by detecting the database: the access information of the image database can be determined, and the access information may include the access address and access authentication information of the image database, for example: the access information may include the Internet Protocol Address (IP address) of the image database, the access user name, the access password and other information. The image database is the image data source from which the original image can be obtained. The access information can be used to access the image database, and the image database can be detected and processed to obtain the detection result; if the detection result indicates that there is a new image in the image database, the new image can be determined as the original image, thereby achieving the acquisition of the original image. In some cases, data related to obtaining the original image can also be configured, for example: the database query statement, unique identification field, file name, file system network address and other data required to obtain the original image are configured, so that the original image data can be accurately obtained. Through the method provided in the embodiment of the present application, the decoupling of the image data source and the image processing system can be achieved, and the original image can be actively acquired without modifying the image data source, thereby ensuring the processing efficiency of the system.

[0046] It should be noted that when the original image is obtained from the image database through the above embodiment, corresponding exception handling operations can also be performed: fault tolerance mechanisms such as detection frequency, error retry times, error retry duration, etc. can be configured to further improve the efficiency of obtaining the original image.

[0047] In one embodiment, the implementation method of acquiring the original image can also be: deploying an image acquisition module in the image data source; receiving the original image data sent by the image acquisition module. The image data source can be modified, and the image acquisition module can be deployed in the image data source. The image acquisition module can be an application (Application, APP) that acquires images in real time. The image acquisition module can continuously acquire the original image in the image data source and send the original image to the image processing device. Through the method provided in the embodiment of the present application, the original image can be passively acquired, and the efficiency of acquiring the original image can be improved.

[0048] S202: Perform format adjustment processing on the original image to obtain an image to be annotated, wherein the image quality of the image to be annotated matches the image quality of the original image.

[0049] In an embodiment of the present application, the original image can be formatted to obtain an image to be annotated, and the image quality of the image to be annotated matches the image quality of the original image. The image format, image size, image resolution and other attributes of the acquired original image may be different. In order to improve the processing speed of the image annotation model, the original image can be formatted, thereby saving computing resources and improving processing efficiency. It should be noted that in some cases, the original image can be formatted without being processed, and the original image and the annotation object type information can be directly input into the image annotation model for processing. Through the method provided in the embodiment of the present application, the image format can be unified, which is conducive to improving the processing efficiency of the image annotation model.

[0050] In one embodiment, the implementation method of performing format adjustment processing on the original image to obtain the image to be annotated may be: obtaining a reference image format and image compression parameters; performing image format conversion processing on the original image according to the reference image format to obtain a target format image, the image format of the target format image matches the reference image format; performing image compression processing on the target format image according to the image compression parameters to obtain the image to be annotated. The reference image format may be determined according to the processing capability of the image annotation model, and the image compression parameters may include parameters such as image compression ratio, pixel size, and storage format. The reference image format and image compression parameters may be determined according to actual application requirements so that the image to be annotated meets the application requirements. The image format conversion processing may be performed on the original image according to the reference image format to obtain a target format image, the image format of which matches the reference image format; the image compression processing may be performed on the target format image according to the image compression parameters to obtain the image to be annotated. Through the method provided in the embodiment of the present application, the format adjustment of the image may be achieved according to the reference image format and the image compression parameters, thereby meeting the application requirements and having good flexibility.

[0051] S203: Input the labeled object type information and the image to be labeled into an image labeling model for processing to obtain target labeling information of the object to be labeled, where the target labeling information includes one or more of name information, type information, object quantity information, and object position indication information.

[0052] In an embodiment of the present application, the type information of the labeled object and the image to be labeled can be input into the image labeling model for processing to obtain the target labeling information of the object to be labeled, and the target labeling information can include one or more of the name information, type information, object quantity information and object position indication information. For example, if the type information of the labeled object is "bicycle" and the image to be labeled includes 3 bicycles, the target labeling information of the object to be labeled obtained by using the image labeling model can include name information ("X brand bicycle"), type information ("bicycle"), object quantity information ("3") and object position indication information (indicating the positions of the 3 bicycles in the original image). The image labeling model is obtained by training the initial labeling model using the adjusted sample labeling information, the adjusted sample labeling information is obtained by adjusting the sample labeling information in the rendered image, the rendered image is obtained by rendering the sample labeling information and the initial sample image, the sample labeling information is obtained by inputting the sample image to be labeled and the type information of the labeled object into the initial labeling model for processing, and the sample image to be labeled is obtained by formatting the initial sample image. The image labeling model is obtained after multiple model trainings and has good prediction accuracy. Through the method provided in the embodiment of the present application, the image annotation model can be used to obtain annotation information, thereby effectively improving the efficiency of image annotation. At the same time, since the prediction accuracy of the image annotation model is relatively high, the accuracy of the annotation information can be effectively improved.

[0053] In one embodiment, the original image is taken and submitted by the target merchant. After determining the target annotation information of the object to be annotated, it is also possible to: determine the credit certification data of the target merchant based on the name information, type information and object quantity information of the target annotation information; determine the credit limit based on the credit certification data of the target merchant, and send the credit limit to the target merchant. The credit certification data of the target merchant can be obtained by calculation and processing based on the name information, type information and object quantity information in the target annotation information. The credit certification data can reflect the number of resources held by the target merchant to a certain extent. In some cases, the unit price of the object to be annotated can also be determined based on the name information and type information, and the credit certification data of the target merchant can be determined based on the object quantity information. The credit limit can be determined based on the credit certification data of the target merchant, and the credit limit can be sent to the target merchant. For example: the name information in the target annotation information is "Y brand car", and the type information is "car". The unit price of the object to be annotated can be determined as "2" based on the name information and type information, and the object quantity information is "4". The credit certification data of the target merchant can be determined as "8". The credit limit can be further determined based on the credit certification data of the target merchant, making the credit limit more reasonable. In some cases, the target annotation information of the object to be annotated can be output in the form of JS key-value pair data (JavaScript Object Notation, JSON), so that other data modules can directly use the target annotation information of the object to be annotated for subsequent processing. The method provided by the embodiment of the present application can effectively improve the rationality of the credit limit.

[0054] See also Figure 3, which is a schematic diagram of an image processing system provided by an embodiment of the present application. The image processing system is mainly divided into two parts: basic services and business services, wherein the basic services are divided into two parts: storage and system: the storage part is used to indicate the data storage type used by the image processing system provided by the present application, and the image processing system can use a relational database (e.g., MySQL) to store data such as the processing time of the model, the adjusted sample annotation information, etc., and can use a non-relational database (e.g., Redis) to store relevant data within the system, and can also use a file storage system to store data such as original images; the system part is used to indicate that the image processing system can provide services to terminal devices in the form of cloud services, and can also be deployed in a graphics processing unit (GPU) cluster. The business service includes a calling part and a processing part, wherein the calling part includes a display end and an external call, and the display end can display the original image and the corresponding image annotation information, which is convenient for interaction with the user; the external call can provide a calling interface for other processes, so that other processes can obtain the image annotation information determined by the image processing system through external calls. The processing part of the business service includes a data acquisition interface, a data detection service, an image compression service, an image annotation model, an annotation output interface, and an annotation adjustment interface, wherein the data acquisition interface can receive the original image sent by the image acquisition module deployed in the image data source; the data detection service can actively detect and process the image data source and obtain the original image; the image compression service can adjust the format of the original image to obtain the image to be annotated; the image annotation model can process according to the annotation object type information and the image to be annotated to obtain the annotation information of the image; the annotation output interface can send the determined annotation information to other systems; the annotation adjustment interface can adjust the determined annotation information, thereby further ensuring the accuracy of the annotation information. The method provided by the embodiment of the present application can effectively improve the efficiency of image annotation while ensuring the accuracy of image annotation information.

[0055] Through the image processing method provided in the embodiment of the present application, the original image data can be obtained in a variety of ways, effectively ensuring the security of the data source and the image acquisition efficiency; the original image can be formatted and processed, so that the format of the image to be annotated is the same, which is conducive to improving the processing efficiency of the image annotation model; the image annotation model can be used to process the annotation object type information and the image to be annotated to obtain annotation information, effectively improving the efficiency of image annotation, and realizing automatic detection and annotation of objects in the image; and because the image annotation model is obtained by performing multiple model training based on the initial annotation model, the annotation information obtained according to the image annotation model has good accuracy, avoiding the problems of wrong labeling, missing labeling, etc. that may occur during manual annotation, and effectively reducing the image annotation cost.

[0056] See also Figure 4 , Figure 4 A flow chart of a model training method provided in an embodiment of the present application. The model training method can be implemented by the above-mentioned image processing device 102, or by other devices. The flow of the model training method provided in the embodiment of the present application includes but is not limited to:

[0057] S401: Acquire an initial sample image, where the initial sample image includes a sample object, and the sample object is of the same type as the object to be labeled.

[0058] In an embodiment of the present application, for a certain type of object, an initial annotation model can be trained using an initial sample image to obtain an image annotation model, thereby effectively ensuring the prediction accuracy of the image annotation model. For example, for objects of the vehicle type, an initial annotation model can be trained using initial sample images containing various vehicles, so that the image annotation model has good accuracy in the annotation information obtained by processing the original image containing the vehicle. Multiple initial sample images can be obtained, each of which can include one or more sample objects, and the sample objects are of the same type as the object to be annotated. Through the method provided in the embodiment of the present application, the prediction accuracy of the image annotation model can be improved, thereby further improving the accuracy of the annotation information obtained according to the image annotation model.

[0059] S402: Perform format adjustment processing on the initial sample image to obtain a sample image to be labeled, wherein the image quality of the sample image to be labeled matches the image quality of the initial sample image.

[0060] In an embodiment of the present application, the initial sample image can be formatted to obtain a sample image to be annotated, and the image quality of the sample image to be annotated matches the image quality of the initial sample image. The image format, image size, image resolution and other attributes of the acquired initial sample image may be different. In order to improve the processing speed of the initial annotation model, the initial sample image can be formatted, thereby saving computing resources and improving the model training efficiency. It should be noted that in some cases, the initial sample image may not be formatted, and the initial sample image and the annotation object type information may be directly input into the initial annotation model for processing. Through the method provided in the embodiment of the present application, the image format can be unified, which is conducive to improving the training efficiency of the initial annotation model.

[0061] In one embodiment, the implementation method of performing format adjustment processing on the initial sample image to obtain the sample image to be annotated may be: obtaining a reference image format and an image compression parameter; performing image format conversion processing on the initial sample image according to the reference image format to obtain a sample image in a target format, wherein the image format of the sample image in the target format matches the reference image format; performing image compression processing on the sample image in the target format according to the image compression parameter to obtain the sample image to be annotated. The reference image format may be determined according to the processing capability of the image annotation model, and the image compression parameter may include parameters such as image compression ratio, pixel size, and storage format. The reference image format and image compression parameter may be determined according to actual application requirements so that the sample image to be annotated meets the application requirements. The initial sample image may be subjected to image format conversion processing according to the reference image format to obtain a sample image in a target format, wherein the image format of the sample image in the target format matches the reference image format; and the sample image in the target format may be subjected to image compression processing according to the image compression parameter to obtain the sample image to be annotated. Through the method provided in the embodiment of the present application, the format adjustment of the image may be realized according to the reference image format and the image compression parameter, thereby meeting the application requirements and having good flexibility.

[0062] S403: Input the labeled object type information and the sample image to be labeled into an initial labeling model for processing to obtain sample labeling information of the sample object.

[0063] In an embodiment of the present application, the annotation object type information and the sample image to be annotated can be input into the initial annotation model for processing to obtain sample annotation information of the sample object, and the sample annotation information may include one or more of sample name information, sample type information, sample object quantity information, and sample object position indication information. When there are multiple sample images to be annotated corresponding to the initial sample objects, each sample image to be annotated and the annotation object type information can be input into the initial annotation model in turn for processing to obtain sample annotation information of the sample object in each sample image to be annotated. Through the method provided in the embodiment of the present application, the sample annotation information of the sample object can be determined, which is convenient for subsequent model training based on the sample annotation information, and is conducive to improving the efficiency of model training.

[0064] In one embodiment, the implementation method of inputting the type information of the labeled object and the sample image to be labeled into the initial labeling model for processing to obtain the sample labeling information of the sample object can be: inputting the sample image to be labeled into the image encoding module in the initial labeling model for encoding processing to obtain the image encoding data to be labeled; using the information encoding module in the initial labeling model to encode the type information of the labeled object to obtain the type information encoding data; inputting the image encoding data to be labeled and the type information encoding data into the decoding module in the initial labeling model for decoding processing to obtain the sample labeling information of the sample object. The initial labeling model includes an image encoding module, an information encoding module and a decoding module. The sample image to be labeled can be input into the image encoding module in the initial labeling model for encoding processing to obtain the image encoding data to be labeled; the type information of the labeled object can be input into the information encoding module in the initial labeling model for encoding processing to obtain the type information encoding data; the image encoding data to be labeled and the type information encoding data can be input into the decoding module in the initial labeling model for decoding processing to obtain the sample labeling information of the sample object. Through the method provided in the embodiment of the present application, the sample labeling information can be determined according to the image to be labeled and the type information, so as to facilitate the subsequent model training of the initial labeling model according to the sample labeling information.

[0065] In one embodiment, the initial annotation model can be a Segment Anything Model (SAM). The SAM model can generate a one-time code for the image through an image encoder, and then use a lightweight encoder to convert any prompt information into a prompt code in real time, and then combine the two information sources of image code and prompt code in a lightweight decoder for predicting the segmentation mask and finally outputting the annotation information. The SAM model has the characteristics of an interactive image segmentation model (can receive two interactive prompt words, sparse prompt and dense prompt), and zero-sample learning; it has the characteristics of using a combination of three encoders, a prompt word encoder, an image encoder, and a mask decoder, to extract picture feature information; and it has the characteristics of a large amount of model training data. The method provided by the embodiment of the present application can effectively improve the accuracy of the annotation information obtained according to the image annotation model.

[0066] It should be noted that the initial annotation model can be a pre-trained model, and the model training method provided in this application can be used to fine-tune the model parameters in the initial annotation model to obtain an image annotation model with higher prediction accuracy, thereby effectively improving the model training efficiency and saving resources.

[0067] S404: Perform image rendering processing on the sample annotation information and the initial sample image to obtain a rendered image, and adjust the sample annotation information in the rendered image to obtain adjusted sample annotation information.

[0068] In the embodiment of the present application, the sample annotation information and the initial sample image can be subjected to image rendering to obtain a rendered image, which can more intuitively display the sample annotation information, facilitating subsequent discrimination and adjustment; the sample annotation information in the rendered image can be adjusted to obtain adjusted sample annotation information. Through the method provided in the embodiment of the present application, the adjusted sample annotation information can be obtained, thereby achieving the adjustment of the prediction results of the initial annotation model, facilitating subsequent model training of the initial annotation model, and improving the model training efficiency.

[0069] In one embodiment, the sample annotation information may include sample object quantity information and sample object position indication information, wherein the sample object quantity information may indicate that the initial sample image includes multiple sample objects; the sample annotation information and the initial sample image are subjected to image rendering processing to obtain a rendered image, which may be implemented as follows: determining the position information corresponding to each sample object in the initial sample image according to the sample object position indication information, wherein the position information corresponding to the sample object is used to indicate the image area where the sample object is located in the initial sample image; determining the layer corresponding to each sample object in the rendered image; and rendering the image area corresponding to each sample object in the initial sample image in the layer corresponding to each sample object in sequence according to the position information corresponding to each sample object to obtain a rendered image. After the sample annotation information is determined, the sample image can be rendered using the sample annotation information: the sample annotation information includes sample object quantity information and sample object position indication information. The position information corresponding to each sample object in the initial sample image can be determined using the sample object position indication information. The position information corresponding to the sample object can be used to indicate the image region where the sample object is located in the initial sample image. For example, assuming that the region corresponding to the initial sample image is the region connected and surrounded by four points (0,0), (0,1), (1,0), (1,1); the sample object quantity information is 1, and the position information corresponding to the sample object can be determined according to the sample object position indication information as "a rectangular image region connected and surrounded by four points (0.1,0.1), (0.3,0.1), (0.1,0.3), (0.3,0.3) in the initial sample image". The layer corresponding to each sample object in the rendered image can be determined, and one sample object corresponds to one layer. According to the position information corresponding to each sample object, the image region corresponding to each sample object in the initial sample image can be sequentially rendered in the image corresponding to each sample object, thereby obtaining a rendered image. Through the method provided in the embodiment of the present application, rendering can be performed according to the hierarchical structure of the sample object, which effectively ensures the rendering effect of the rendered image and improves the recognizability of the rendered image.

[0070] In one embodiment, LeaferJS technology can be used to render the initial sample image according to the sample annotation information to obtain a rendered image. LeaferJS can realize the graphics rendering function, and can be combined with artificial intelligence (AI) drawing, interface generation and other functions. LeaferJS has the characteristics of small amount of code, fast loading and parsing speed; it has the characteristics of using graphics processing unit (GPU) for image rendering and high rendering performance; it has the characteristics of organizing and rendering image data in a hierarchical structure. The layered rendering in LeaferJS can make the drawing and interaction of images more efficient, thereby effectively improving the rendering speed.

[0071] See also Figure 5 , which is a schematic diagram of a rendered image provided by an embodiment of the present application. The sample annotation information is determined by the method provided by the embodiment of the present application. The sample annotation information includes sample object quantity information and sample object position indication information. Image rendering processing can be performed based on the sample annotation information and the initial sample image to obtain a rendered image, such as Figure 5 shown. Figure 5 In the example, the rendered image includes three sample objects (the sample objects are bicycles), and the rendered image includes the number of sample objects (i.e. Figure 5 The number "3" in the circle in the upper left corner of the image), the position information corresponding to each sample object can also be displayed in the rendered image ( Figure 5 The image area contained in the dotted box is the location information of the corresponding sample object, that is, the image area where the sample object is located in the image). Through the method provided in the embodiment of the present application, the relationship between the sample annotation information and the image can be intuitively displayed using the rendered image, which facilitates the subsequent adjustment of the sample annotation information according to the rendered image and improves the training efficiency of the model.

[0072] In one embodiment, the sample annotation information in the rendered image is adjusted to obtain the adjusted sample annotation information, which can be realized by: displaying an annotation information adjustment interface, which includes the name information and type information of the rendered image and each sample object; receiving adjustment information for the annotation information adjustment interface, which includes one or more of the name adjustment information, type adjustment information, object quantity adjustment information and object position adjustment information; integrating the adjustment information and the sample annotation information to obtain the adjusted sample annotation information. The rendered image is determined according to the initial sample image and the sample object quantity information and the sample object position indication information in the sample annotation information, and the sample annotation information can be adjusted according to the rendered image: the annotation information adjustment interface can be displayed, which includes the name information and class information of each sample object in the rendered image and the sample annotation information; the adjustment information input for the annotation information adjustment interface can be received, which includes one or more of the name adjustment information, type adjustment information, object quantity adjustment information and object position adjustment information; the adjustment information and the sample annotation information can be integrated to obtain the adjusted sample annotation information. The method provided in the embodiment of the present application can facilitate the rapid adjustment of sample annotation information, greatly reduce the learning cost of annotation adjustment personnel, improve the work efficiency of annotation adjustment personnel, and also improve the training efficiency of image annotation models.

[0073] See also Figure 6 , which is a schematic diagram of a label information adjustment interface provided by an embodiment of the present application. Figure 6 The annotation information adjustment interface shown includes a rendered image (the rendered image includes the position information and number of sample objects corresponding to each sample object), the name information, and the type information of the sample object. The position information of any sample object can be adjusted through the annotation information adjustment interface (that is, the size and position of the dotted box in the rendered image can be adjusted); the name information and type information of any sample object can be adjusted (for example: Figure 6 Enter the adjusted name information in the input box shown); you can also adjust the number of sample objects (for example: Figure 6 The label information adjustment interface includes a label information list, which includes the label information corresponding to each sample object, such as Figure 6, the annotation information list includes the annotation information of sample objects 1-3, wherein the name information of sample object 3 is "Y brand bicycle" and the category information is "bicycle". The annotation information adjustment interface also includes a confirmation control. After determining that the sample annotation information has been adjusted, the confirmation control can be triggered; after receiving the trigger instruction, the adjusted sample annotation information can be determined according to the input adjustment information. In some cases, the annotation information adjustment interface can also provide functions such as annotation information deletion, annotation information locking, annotation information hiding, and annotation management. Through the method provided in the embodiment of the present application, the adjustment efficiency of the sample annotation information can be effectively improved, thereby improving the training efficiency of the image annotation model.

[0074] S405 . Adjust model parameters of the initial annotation model according to the adjusted sample annotation information and the sample annotation information to obtain the image annotation model.

[0075] In an embodiment of the present application, loss calculation can be performed based on the adjusted sample annotation information and the sample annotation information, loss parameters can be determined, and model parameters of the initial annotation model can be adjusted based on the loss parameters. The model training method provided in the present application is used to adjust the model parameters of the initial annotation model multiple times. When the number of model parameter adjustments of the initial annotation model reaches a preset number of times, or the prediction accuracy of the initial annotation model reaches a preset requirement, it can be determined that the model training of the initial annotation model is completed, and the initial annotation model at this time is determined as the image annotation model. The method provided in the embodiment of the present application can effectively improve the prediction accuracy of the image annotation model, so that the image annotation model can accurately determine the precise annotation information of objects with large scale differences or severe occlusion in the image to be annotated, which meets the actual application needs.

[0076] See also Figure 7 , which is a schematic diagram of a model training method provided in an embodiment of the present application. Figure 7The initial annotation model is trained in the process, and the initial annotation model includes an image encoding module, an information encoding module and a decoding module. An initial sample image can be obtained, and the initial sample image includes a sample object; the format of the initial sample image is adjusted to obtain a sample image to be annotated; the sample image to be annotated is input into the image encoding module in the initial annotation model for encoding processing to obtain the image encoding data to be annotated; the annotation object type information is input into the information encoding module in the initial annotation model for encoding processing to obtain the type information encoding data; the image encoding data to be annotated and the type information encoding data are input into the decoding module in the initial annotation model for decoding processing to obtain the sample annotation information of the sample object; the rendering image is determined according to the sample annotation information and the initial sample image, and the adjusted sample annotation information is determined according to the rendering image; the model parameters of the initial annotation model can be adjusted according to the adjusted sample annotation information and the sample annotation information to obtain the image annotation model. The method provided in the embodiment of the present application can improve the training efficiency of the model, and effectively improve the prediction accuracy of the image annotation model.

[0077] The model training method provided by the embodiment of the present application can use the initial sample image to train the initial annotation model, so that the initial annotation model can accurately predict a certain type of object, effectively improving the annotation accuracy; the sample annotation information and the initial sample image can be used to obtain a rendered image, and the rendered image can be used to determine the adjusted sample annotation information, which can be conducive to quickly realizing the adjustment of the sample annotation information, greatly reducing the learning cost of the annotation adjustment personnel, improving the work efficiency of the annotation adjustment personnel, and also improving the training efficiency of the image annotation model; the initial annotation image can be trained according to the adjusted sample annotation information to obtain the image annotation model, so that the image annotation model can accurately determine the annotation information of objects with large scale differences or severe occlusion in the image to be annotated, which meets the actual application needs.

[0078] See also Figure 8 , Figure 8 This is a structural block diagram of an image processing device provided in an embodiment of the present application. The device includes:

[0079] An acquisition unit 801 is used to acquire the type information of the labeled object and the original image, wherein the original image includes the object to be labeled;

[0080] An adjusting unit 802 is used to perform format adjustment processing on the original image to obtain an image to be annotated, wherein the image quality of the image to be annotated matches the image quality of the original image;

[0081] A processing unit 803 is used to input the annotation object type information and the image to be annotated into an image annotation model for processing to obtain target annotation information of the object to be annotated, where the target annotation information includes one or more of name information, type information, object quantity information, and object position indication information;

[0082] Among them, the image annotation model is obtained by training the initial annotation model using the adjusted sample annotation information, the adjusted sample annotation information is obtained by adjusting the sample annotation information in the rendered image, the rendered image is obtained by rendering the sample annotation information and the initial sample image, the sample annotation information is obtained by inputting the sample image to be annotated and the annotation object type information into the initial annotation model for processing, and the sample image to be annotated is obtained by formatting the initial sample image.

[0083] In one embodiment, the processing unit 803 is further used to: obtain an initial sample image, the initial sample image includes a sample object, and the sample object is of the same type as the object to be labeled; perform format adjustment processing on the initial sample image to obtain a sample image to be labeled, and the image quality of the sample image to be labeled matches the image quality of the initial sample image; input the labeled object type information and the sample image to be labeled into an initial labeling model for processing to obtain sample labeling information of the sample object; perform image rendering processing on the sample labeling information and the initial sample image to obtain a rendered image, and adjust the sample labeling information in the rendered image to obtain adjusted sample labeling information; adjust model parameters of the initial labeling model according to the adjusted sample labeling information and the sample labeling information to obtain the image labeling model.

[0084] In one embodiment, when the processing unit 803 inputs the annotation object type information and the sample image to be annotated into the initial annotation model for processing to obtain the sample annotation information of the sample object, it is specifically used to: input the sample image to be annotated into the image encoding module in the initial annotation model for encoding processing to obtain the image encoding data to be annotated; use the information encoding module in the initial annotation model to encode the annotation object type information to obtain type information encoding data; input the image encoding data to be annotated and the type information encoding data into the decoding module in the initial annotation model for decoding processing to obtain the sample annotation information of the sample object.

[0085] In one embodiment, the sample annotation information includes sample object quantity information and sample object position indication information, and the sample object quantity information indicates that the initial sample image includes a plurality of the sample objects; when the processing unit 803 performs image rendering processing on the sample annotation information and the initial sample image to obtain a rendered image, it is specifically used to: determine the position information corresponding to each of the sample objects in the initial sample image according to the sample object position indication information, and the position information corresponding to the sample object is used to indicate the image area where the sample object is located in the initial sample image; determine the layer corresponding to each of the sample objects in the rendered image; and render the image area corresponding to each of the sample objects in the initial sample image in the layer corresponding to each of the sample objects in sequence according to the position information corresponding to each of the sample objects to obtain the rendered image.

[0086] In one embodiment, when the processing unit 803 adjusts the sample annotation information according to the rendered image to obtain the adjusted sample annotation information, it is specifically used to: display an annotation information adjustment interface, the annotation information adjustment interface includes the rendered image and the name information and type information of each of the sample objects; receive adjustment information for the annotation information adjustment interface, the adjustment information includes one or more of name adjustment information, type adjustment information, object quantity adjustment information and object position adjustment information; integrate the adjustment information and the sample annotation information to obtain the adjusted sample annotation information.

[0087] In one embodiment, when the adjustment unit 802 performs format adjustment processing on the original image to obtain the image to be annotated, it is specifically used to: obtain a reference image format and image compression parameters; perform image format conversion processing on the original image according to the reference image format to obtain a target format image, and the image format of the target format image matches the reference image format; perform image compression processing on the target format image according to the image compression parameters to obtain the image to be annotated.

[0088] In one embodiment, the acquisition unit 801 is also used to: determine access information of the image database, the access information including the access address and access authentication information of the image database; perform detection processing on the image database according to the access information to obtain a detection result; if the detection result indicates that there is a new image in the image database, then determine the new image as the original image.

[0089] In one embodiment, the original image is taken and submitted by the target merchant, and the acquisition unit 801 is also used to: determine the credit certification data of the target merchant based on the name information, type information and object quantity information of the target annotation information; determine the credit limit based on the credit certification data of the target merchant, and send the credit limit to the target merchant.

[0090] It can be understood that the functions of each functional unit of the image processing device in the embodiment of the present application can be specifically implemented according to the image processing method in the above method embodiment, and its specific implementation process can refer to the relevant description in the above image processing method embodiment, which will not be repeated here.

[0091] Through the image processing device provided by the embodiment of the present application, the original image data can be obtained in a variety of ways, which effectively ensures the security of the data source and the image acquisition efficiency; the format of the original image can be adjusted, so that the format of the image to be annotated is the same, which is conducive to improving the processing efficiency of the image annotation model; the image annotation model can be used to process the annotation object type information and the image to be annotated to obtain the annotation information, effectively improving the efficiency of image annotation, and realizing the automatic detection and annotation of objects in the image; and because the image annotation model is obtained by multiple model training based on the initial annotation model, the annotation information obtained according to the image annotation model has good accuracy, avoiding the problems of mislabeling and missing labels that may occur during manual annotation, and effectively reducing the image annotation cost; during model training, the sample annotation information and the initial sample image can be used to obtain the rendered image, and the rendered image can be used to determine the adjusted sample annotation information, which can be conducive to quickly realizing the adjustment of the sample annotation information, greatly reducing the learning cost of the annotation adjustment personnel, improving the work efficiency of the annotation adjustment personnel, and also improving the training efficiency of the image annotation model, so that the trained image annotation model can accurately determine the annotation information of objects with large scale differences or severe occlusion in the image to be annotated, which meets the actual application needs.

[0092] See also Fig. 9 , Fig. 9 A block diagram of a computer device provided in an embodiment of the present application. The computer device described in the embodiment of the present application includes: a processor 901, a communication interface 902, and a memory 903. The processor 901, the communication interface 902, and the memory 903 may be connected via a bus or other means, and the embodiment of the present application takes the connection via a bus as an example.

[0093] Among them, the processor 901 (or CPU (Central Processing Unit)) is the computing core and control core of the computer device, which can parse various instructions in the computer device and process various data of the computer device. For example, the CPU can be used to parse the power on and off instructions sent by the user to the computer device, and control the computer device to perform power on and off operations; for another example, the CPU can transmit various interactive data between the internal structures of the computer device, and so on. The communication interface 902 can optionally include a standard wired interface, a wireless interface (such as Wi-Fi, a mobile communication interface, etc.), which is controlled by the processor 901 to send and receive data. The memory 903 (Memory) is a memory device in the computer device for storing programs and data. It can be understood that the memory 903 here can include both the built-in memory of the computer device and the extended memory supported by the computer device. The memory 903 provides a storage space, which stores the operating system of the computer device, which may include but is not limited to: Android system, iOS system, Windows Phone system, etc., and this application does not limit this.

[0094] In the embodiment of the present application, the processor 901 performs the following operations by running the executable program code in the memory 903:

[0095] Acquire the type information of the labeled object and the original image, wherein the original image includes the object to be labeled;

[0096] Performing format adjustment processing on the original image to obtain an image to be annotated, wherein the image quality of the image to be annotated matches the image quality of the original image;

[0097] Inputting the labeled object type information and the image to be labeled into an image labeling model for processing to obtain target labeling information of the object to be labeled, wherein the target labeling information includes one or more of name information, type information, object quantity information, and object position indication information;

[0098] Among them, the image annotation model is obtained by training the initial annotation model using the adjusted sample annotation information, the adjusted sample annotation information is obtained by adjusting the sample annotation information in the rendered image, the rendered image is obtained by rendering the sample annotation information and the initial sample image, the sample annotation information is obtained by inputting the sample image to be annotated and the annotation object type information into the initial annotation model for processing, and the sample image to be annotated is obtained by formatting the initial sample image.

[0099] In one embodiment, the processor 901 is further used to: obtain an initial sample image, the initial sample image includes a sample object, and the sample object is of the same type as the object to be labeled; perform format adjustment processing on the initial sample image to obtain a sample image to be labeled, and the image quality of the sample image to be labeled matches the image quality of the initial sample image; input the labeled object type information and the sample image to be labeled into an initial labeling model for processing to obtain sample labeling information of the sample object; perform image rendering processing on the sample labeling information and the initial sample image to obtain a rendered image, and adjust the sample labeling information according to the rendered image to obtain adjusted sample labeling information; adjust model parameters of the initial labeling model according to the adjusted sample labeling information and the sample labeling information to obtain the image labeling model.

[0100] In one embodiment, when the processor 901 inputs the annotation object type information and the sample image to be annotated into the initial annotation model for processing to obtain sample annotation information of the sample object, it is specifically used to: input the sample image to be annotated into the image encoding module in the initial annotation model for encoding processing to obtain image encoding data to be annotated; use the information encoding module in the initial annotation model to encode the annotation object type information to obtain type information encoding data; input the image encoding data to be annotated and the type information encoding data into the decoding module in the initial annotation model for decoding processing to obtain sample annotation information of the sample object.

[0101] In one embodiment, the sample annotation information includes sample object quantity information and sample object position indication information, and the sample object quantity information indicates that the initial sample image includes multiple sample objects; when the processor 901 performs image rendering processing on the sample annotation information and the initial sample image to obtain a rendered image, it is specifically used to: determine the position information corresponding to each of the sample objects in the initial sample image according to the sample object position indication information, and the position information corresponding to the sample object is used to indicate the image area where the sample object is located in the initial sample image; determine the layer corresponding to each of the sample objects in the rendered image; and render the image area corresponding to each of the sample objects in the initial sample image in the layer corresponding to each of the sample objects in sequence according to the position information corresponding to each of the sample objects to obtain the rendered image.

[0102] In one embodiment, when the processor 901 adjusts the sample annotation information according to the rendered image to obtain the adjusted sample annotation information, it is specifically used to: display an annotation information adjustment interface, the annotation information adjustment interface includes the rendered image and the name information and type information of each of the sample objects; receive adjustment information for the annotation information adjustment interface, the adjustment information includes one or more of name adjustment information, type adjustment information, object quantity adjustment information and object position adjustment information; integrate the adjustment information and the sample annotation information to obtain the adjusted sample annotation information.

[0103] In one embodiment, when the processor 901 performs format adjustment processing on the original image to obtain the image to be annotated, it is specifically used to: obtain a reference image format and image compression parameters; perform image format conversion processing on the original image according to the reference image format to obtain a target format image, and the image format of the target format image matches the reference image format; perform image compression processing on the target format image according to the image compression parameters to obtain the image to be annotated.

[0104] In one embodiment, the processor 901 is further used to: determine access information of an image database, the access information including an access address and access authentication information of the image database; perform detection processing on the image database according to the access information to obtain a detection result; if the detection result indicates that a new image exists in the image database, the new image is determined as the original image.

[0105] In one embodiment, the original image is taken and submitted by the target merchant, and the processor 901 is also used to: determine the credit certification data of the target merchant based on the name information, type information and object quantity information of the target annotation information; determine the credit limit based on the credit certification data of the target merchant, and send the credit limit to the target merchant.

[0106] In a specific implementation, the processor 901, communication interface 902 and memory 903 described in the embodiment of the present application can execute the implementation method of the image processing device described in an image processing method provided in an embodiment of the present application, and can also execute the implementation method described in an image processing device provided in an embodiment of the present application, which will not be repeated here.

[0107] Through the computer device provided in the embodiment of the present application, the original image data can be obtained in a variety of ways, which effectively ensures the security of the data source and the image acquisition efficiency; the format of the original image can be adjusted, so that the format of the image to be annotated is the same, which is conducive to improving the processing efficiency of the image annotation model; the image annotation model can be used to process the annotation object type information and the image to be annotated to obtain the annotation information, effectively improving the efficiency of image annotation, and realizing the automatic detection and annotation of objects in the image; and because the image annotation model is obtained by multiple model training based on the initial annotation model, the annotation information obtained according to the image annotation model has good accuracy, avoiding the problems of mislabeling and missing labels that may occur during manual annotation, and effectively reducing the image annotation cost; during model training, the sample annotation information and the initial sample image can be used to obtain the rendered image, and the rendered image can be used to determine the adjusted sample annotation information, which can be conducive to quickly realizing the adjustment of the sample annotation information, greatly reducing the learning cost of the annotation adjustment personnel, improving the work efficiency of the annotation adjustment personnel, and also improving the training efficiency of the image annotation model, so that the trained image annotation model can accurately determine the annotation information of objects with large scale differences or severe occlusion in the image to be annotated, which meets the actual application needs.

[0108] The present application also provides a computer-readable storage medium, wherein a computer program is stored in the computer-readable storage medium, and when the computer-readable storage medium is run on a computer, the computer executes the image processing method as described in the present application. The specific implementation method can be referred to the above description, and will not be repeated here.

[0109] The embodiment of the present application also provides a computer program product, which includes a computer program or a computer instruction, and the computer program or the computer instruction is stored in a computer-readable storage medium. The processor of the computer device reads the computer program or the computer instruction from the computer-readable storage medium, and the processor executes the computer program or the computer instruction, so that the computer device performs the image processing method as described in the embodiment of the present application. The specific implementation method can be referred to the above description, which will not be repeated here.

[0110] It should be noted that, for the above-mentioned various method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that this application is not limited by the order of the actions described, because according to this application, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by this application.

[0111] In the above embodiments, the description of each embodiment has its own emphasis. For the part that is not described in detail in a certain embodiment, please refer to the relevant description of other embodiments. The technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a computer, a server or a network device, etc., specifically a processor in a computer device) to perform all or part of the steps of the above methods of each embodiment of the present application. Among them, the aforementioned storage medium may include: U disk, mobile hard disk, magnetic disk, optical disk, read-only memory (English: Read-Only Memory, abbreviated: ROM) or random access memory (English: Random Access Memory, abbreviated: RAM) and other media that can store program codes.

[0112] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, it should be understood that the technical solutions recorded in the aforementioned embodiments can still be modified, or some of the technical features therein can be replaced by equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. An image processing method, characterized in that: The method comprises: Acquire the type information of the labeled object and the original image, wherein the original image includes the object to be labeled; Performing format adjustment processing on the original image to obtain an image to be annotated, wherein the image quality of the image to be annotated matches the image quality of the original image; Inputting the labeled object type information and the image to be labeled into an image labeling model for processing to obtain target labeling information of the object to be labeled, wherein the target labeling information includes one or more of name information, type information, object quantity information, and object position indication information; Among them, the image annotation model is obtained by training the initial annotation model using the adjusted sample annotation information, the adjusted sample annotation information is obtained by adjusting the sample annotation information in the rendered image, the rendered image is obtained by rendering the sample annotation information and the initial sample image, the sample annotation information is obtained by inputting the sample image to be annotated and the annotation object type information into the initial annotation model for processing, and the sample image to be annotated is obtained by formatting the initial sample image.

2. The method according to claim 1, characterized in that: The method further comprises: Acquire an initial sample image, wherein the initial sample image includes a sample object, and the sample object is of the same type as the object to be labeled; Performing format adjustment processing on the initial sample image to obtain a sample image to be labeled, wherein the image quality of the sample image to be labeled matches the image quality of the initial sample image; Inputting the labeled object type information and the sample image to be labeled into an initial labeling model for processing to obtain sample labeling information of the sample object; Performing image rendering processing on the sample annotation information and the initial sample image to obtain a rendered image, and adjusting the sample annotation information according to the rendered image to obtain adjusted sample annotation information; The model parameters of the initial annotation model are adjusted according to the adjusted sample annotation information and the sample annotation information to obtain the image annotation model.

3. The method according to claim 2, characterized in that The step of inputting the labeled object type information and the sample image to be labeled into an initial labeling model for processing to obtain sample labeling information of the sample object includes: Inputting the sample image to be labeled into the image coding module in the initial labeling model for coding processing to obtain coded data of the image to be labeled; Using the information encoding module in the initial annotation model to encode the annotation object type information to obtain type information encoding data; The image encoding data to be annotated and the type information encoding data are input into a decoding module in the initial annotation model for decoding processing to obtain sample annotation information of the sample object.

4. The method according to claim 2, characterized in that: The sample annotation information includes sample object quantity information and sample object position indication information, wherein the sample object quantity information indicates that the initial sample image includes a plurality of the sample objects; The performing image rendering processing on the sample annotation information and the initial sample image to obtain a rendered image includes: Determine, according to the sample object position indication information, position information corresponding to each of the sample objects in the initial sample image, wherein the position information corresponding to the sample object is used to indicate an image region where the sample object is located in the initial sample image; Determine the layer corresponding to each of the sample objects in the rendered image; According to the position information corresponding to each of the sample objects, the image regions corresponding to each of the sample objects in the initial sample image are sequentially rendered in the layers corresponding to each of the sample objects to obtain the rendered image.

5. The method according to claim 4, characterized in that The step of adjusting the sample annotation information according to the rendered image to obtain the adjusted sample annotation information includes: Displaying a labeling information adjustment interface, wherein the labeling information adjustment interface includes the name information and type information of the rendered image and each of the sample objects; receiving adjustment information for the annotation information adjustment interface, the adjustment information including one or more of name adjustment information, type adjustment information, object quantity adjustment information, and object position adjustment information; The adjustment information and the sample annotation information are integrated to obtain adjusted sample annotation information.

6. The method according to any one of claims 1 to 5, characterized in that: The step of performing format adjustment processing on the original image to obtain the image to be annotated includes: Obtain reference image format and image compression parameters; Performing image format conversion processing on the original image according to the reference image format to obtain a target format image, wherein the image format of the target format image matches the reference image format; The target format image is compressed according to the image compression parameters to obtain an image to be labeled.

7. The method according to any one of claims 1 to 5, characterized in that: The method further comprises: Determining access information of an image database, wherein the access information includes an access address and access authentication information of the image database; Performing detection processing on the image database according to the access information to obtain a detection result; If the detection result indicates that there is a new image in the image database, the new image is determined as the original image.

8. The method according to any one of claims 1 to 5, characterized in that: The original image is taken and submitted by the target merchant, and the method further includes: Determine the credit certification data of the target merchant according to the name information, type information and object quantity information of the target annotation information; A credit limit is determined according to the credit certification data of the target merchant, and the credit limit is sent to the target merchant.

9. A computer device, characterized in that: include: A processor, a communication interface and a memory, wherein the processor, the communication interface and the memory are connected to each other, wherein the memory stores an executable program code, and the processor is used to call the executable program code to implement the image processing method according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, which, when executed on a computer, enable the computer to implement the image processing method according to any one of claims 1 to 8.

Images