Comprehensive Evaluation Method and System for the Appearance State of Antique Coins Based on the Joint Modeling of Image Recognition and Deep Learning

Abstract

The present invention discloses a comprehensive evaluation method and system for the appearance state of antique coins based on the joint modeling of image recognition and deep learning, which relates to the field of artificial intelligence. It includes: First, obtain an antique coin evaluation model with multiple network branches, allocate evaluation dimensions to each branch, and determine the training direction. Configure an evaluation training framework according to the training direction, which includes evaluation units corresponding to each branch. During training, call the evaluation units to execute the evaluation dimension training process according to the corresponding training direction to complete the model training. Import the image to be evaluated into the trained model to obtain the comprehensive evaluation result of the appearance state. This method improves the accuracy and efficiency of the evaluation of the appearance state of antique coins by scientifically configuring the model and the training process.

Description

Technical Field

[0001] The present invention relates to the technical field of artificial intelligence. Specifically, it relates to a comprehensive evaluation method and system for the appearance state of antique coins based on the joint modeling of image recognition and deep learning. Background Art

[0002] With the development of the antique market, the demand for accurate evaluation of the appearance state of antique coins is increasing day by day. Traditional methods for evaluating the appearance state of antique coins mostly rely on manual experience, with strong subjectivity and low efficiency. Although there have been some evaluation attempts based on image recognition or deep learning, there are problems such as single evaluation dimension and low model training efficiency, making it difficult to comprehensively and accurately evaluate the appearance state of antique coins. Summary of the Invention

[0003] The purpose of the present invention is to provide a comprehensive evaluation method and system for the appearance state of antique coins based on the joint modeling of image recognition and deep learning.

[0004] In a first aspect, an embodiment of the present invention provides a comprehensive evaluation method for the appearance state of antique coins based on the joint modeling of image recognition and deep learning, including:

[0005] Obtain an untrained antique coin evaluation model, where the antique coin evaluation model includes multiple antique coin evaluation network branches;

[0006] Perform evaluation dimension allocation processing on each of the antique coin evaluation network branches to obtain the training directions of the antique coins for each of the antique coin evaluation network branches; the training directions of the antique coins for each of the antique coin evaluation network branches are used to represent the evaluation dimensions and the training processes of the evaluation dimensions in the training stage for each of the antique coin evaluation network branches;

[0007] Based on the training directions of the antique coins for the multiple antique coin evaluation network branches, configure an evaluation training framework for the antique coin evaluation model, where the evaluation training framework includes multiple evaluation units; one evaluation unit corresponds to one antique coin evaluation network branch, and each evaluation unit is configured based on the training direction of the corresponding antique coin evaluation network branch;

[0008] When performing the training process on the antique coin evaluation model, call the multiple evaluation units in the evaluation training framework to respectively execute the evaluation dimension training processes of the corresponding antique coin evaluation network branches based on the training processes corresponding to the training directions of the antique coins for the corresponding antique coin evaluation network branches, so as to obtain a trained antique coin evaluation model;

[0009] Import the antique coin image to be evaluated into the trained antique coin evaluation model to obtain a comprehensive evaluation result of the appearance state of the antique coin image to be evaluated.

[0010] In a second aspect, an embodiment of the present invention provides a server system, including a server, which is used to execute the method described in the first aspect.

[0011] Compared with the prior art, the beneficial effects provided by the present invention include: adopting a comprehensive evaluation method and system for the appearance state of antique coins based on the joint modeling of image recognition and deep learning disclosed in the present invention, obtaining an antique coin evaluation model with multiple network branches, allocating evaluation dimensions to each branch, and determining the training direction. According to the training direction, an evaluation training framework is configured, which includes evaluation units corresponding to each branch. During training, the evaluation units are called to execute the evaluation dimension training process according to the corresponding training direction to complete the model training. The image to be evaluated is imported into the trained model to obtain the comprehensive evaluation result of the appearance state. This method improves the accuracy and efficiency of the evaluation of the appearance state of antique coins by scientifically configuring the model and the training process. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required to be used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0013] Figure 1 It is a schematic flow chart of the steps of the comprehensive evaluation method for the appearance state of antique coins based on the joint modeling of image recognition and deep learning provided by the embodiment of the present invention;

[0014] Figure 2 It is a schematic block diagram of the structure of the computer device provided by the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0015] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Usually, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0016] The following will describe in detail the specific embodiments of the present invention with reference to the drawings.

[0017] To solve the technical problems in the foregoing background art, Figure 1It is a schematic flowchart of the comprehensive evaluation method for the appearance state of antique coins based on the joint modeling of image recognition and deep learning provided by the embodiments of the present disclosure. The following is a detailed introduction to the comprehensive evaluation method for the appearance state of antique coins based on the joint modeling of image recognition and deep learning.

[0018] Step S201: Obtain an untrained antique coin evaluation model, where the antique coin evaluation model includes multiple antique coin evaluation network branches.

[0019] Step S202: Perform evaluation dimension allocation processing on each of the antique coin evaluation network branches to obtain the antique coin training directions of each of the antique coin evaluation network branches; the antique coin training directions of each of the antique coin evaluation network branches are used to characterize the evaluation dimensions and the training processes of the evaluation dimensions of each of the antique coin evaluation network branches during the training phase.

[0020] Step S203: Based on the antique coin training directions of the multiple antique coin evaluation network branches, configure the evaluation training framework of the antique coin evaluation model, where the evaluation training framework includes multiple evaluation units; one evaluation unit corresponds to one antique coin evaluation network branch, and each of the evaluation units is configured based on the antique coin training direction of the corresponding antique coin evaluation network branch.

[0021] Step S204: When performing the training process on the antique coin evaluation model, call the multiple evaluation units in the evaluation training framework to respectively execute the evaluation dimension training processes of the corresponding antique coin evaluation network branches based on the training processes corresponding to the antique coin training directions of the corresponding antique coin evaluation network branches, so as to obtain a trained antique coin evaluation model.

[0022] Step S205: Import the antique coin image to be evaluated into the trained antique coin evaluation model to obtain the comprehensive evaluation result of the appearance state of the antique coin image to be evaluated.

[0023] In an embodiment of the present invention, exemplarily, the server obtains an antique coin evaluation model from a specific storage location or data source. For example, the model may be stored in a local database of the server or downloaded from a remote model storage server. This antique coin evaluation model contains multiple antique coin evaluation network branches, and each branch may focus on different aspects of the appearance state evaluation of antique coins. For example, some branches may be responsible for evaluating the wear degree of the coins, and some branches may be responsible for evaluating the clarity of the patterns on the coins, etc. The server performs evaluation dimension allocation processing on each antique coin evaluation network branch respectively. Taking one of the branches (the first antique coin evaluation network branch) as an example: Obtain the evaluation dimension feature domain: The server obtains the evaluation dimension feature domain of the first antique coin evaluation network branch, and this feature domain contains multiple standard antique coin training directions determined in advance for this branch. For example, the standard antique coin training directions may include multiple directions such as overall coin wear evaluation, pattern detail evaluation, and text clarity evaluation. Determine the evaluation period: Based on the standard evaluation configuration parameters corresponding to each standard antique coin training direction in the evaluation dimension feature domain to determine the evaluation period. Assume that the sample evaluation configuration parameters in the evaluation configuration parameters include the evaluation granularity (for example, dividing the antique coin image for evaluation according to different levels of fineness, such as being fine to individual pattern elements or rough to the entire coin surface area) and the sample division dimension (such as dividing samples according to different parts of the coin, such as the front, back, and edge). The architecture evaluation configuration parameters include the training session process segmentation dimension (such as dividing the training into different stages such as initial training and optimization training) and the feature analysis dimension (such as analyzing from different feature angles such as color and shape). Taking the first standard antique coin training direction as an example, the server standardizes the first antique coin evaluation network branch into a network entity in the first model training environment, processes a sample antique coin image with this network entity, and obtains the standard evaluation period on the first model training environment. Then, obtain the evaluation granularity in the standard evaluation configuration parameters corresponding to the first standard antique coin training direction, perform a multiplication operation with the standard evaluation period, and then perform a division operation on the multiplication result with the corresponding architecture evaluation configuration parameters to obtain the evaluation period in this standard training direction. Select the training direction: The server selects the standard antique coin training direction corresponding to the shortest evaluation period in the evaluation dimension feature domain as the antique coin training direction of the first antique coin evaluation network branch. In this way, the antique coin training direction is determined for each antique coin evaluation network branch, and this direction represents the evaluation dimension and training process in the training stage. Before determining the evaluation period, the server will also perform the following operations: Obtain the standard processing loads of multiple antique coin evaluation network branches on the same model training environment. For example, some branches have a larger load when processing complex pattern evaluations, and some have a smaller load when processing simple contour evaluations. Determine the target antique coin evaluation network branch with the largest standard processing load from multiple branches.If the target antique coin evaluation network branch is not the first antique coin evaluation network branch being processed, the server will adjust each standard antique coin training direction in the evaluation dimension feature domain of the first antique coin evaluation network branch according to the antique coin training direction of the target antique coin evaluation network branch. For example, obtain the multiplication operation result between the evaluation granularity corresponding to the antique coin training direction of the target antique coin evaluation network branch and the sample division dimension, and perform a division operation between this result and the sample division dimension corresponding to the first standard antique coin training direction to adjust the evaluation granularity corresponding to the first standard antique coin training direction; or perform a division operation between the multiplication operation result and the evaluation granularity corresponding to the first standard antique coin training direction to adjust the sample division dimension corresponding to the first standard antique coin training direction. Build an evaluation unit: The server builds an evaluation unit corresponding to each branch based on the antique coin training direction of each antique coin evaluation network branch. For example, for the first antique coin evaluation network branch, determine the number of model training environments in the first evaluation unit based on the sample division dimension and the architecture evaluation configuration parameters in the evaluation configuration parameters corresponding to its training direction. Then obtain the environment configurations in the preset model training environment set, such as environmental factors like different hardware configurations and software versions, and obtain the standard evaluation period of the first antique coin evaluation network branch on the model training environments with each environment configuration. Then, based on the evaluation granularity corresponding to the training direction, the architecture evaluation configuration parameters, and the standard evaluation period, determine the evaluation period on the model training environments with each environment configuration, and find the target environment configuration corresponding to the shortest evaluation period. Obtain the number of model training environments of the target environment configuration from the model training environment set. When the number of model training environments of the target environment configuration is greater than or equal to the number of model training environments in the first evaluation unit, use the target environment configuration as the environment configuration of the model training environments in the first evaluation unit, and standardize the first antique coin evaluation network branch in the model training environment set based on the training direction and these configurations, thereby configuring the first evaluation unit. Build the corresponding evaluation unit for each antique coin evaluation network branch in a similar manner. Establish a multi-modal feature fusion architecture: Assume that the antique coin evaluation model also includes a second antique coin evaluation network branch connected before and after the first antique coin evaluation network branch. The first branch corresponds to the first evaluation unit, and the second branch corresponds to the second evaluation unit. The server groups the first evaluation unit and the second evaluation unit, obtains the common evaluation base number between the sample division dimension corresponding to the antique coin training direction of the first antique coin evaluation network branch and the sample division dimension corresponding to the antique coin training direction of the second antique coin evaluation network branch, uses it as the number of evaluation unit pools, groups the two evaluation units based on this number to obtain multiple evaluation unit pools. Each evaluation unit pool includes at least one evaluation subunit obtained from the first evaluation unit and at least one evaluation subunit obtained from the second evaluation unit. Each evaluation subunit contains one or more model training environments.When establishing a multi-modal feature fusion architecture, configure dynamic resource ratios for each evaluation unit pool. For example, allocate different proportions of computing resources according to the requirements of different evaluation tasks, etc., and determine the feature processing sequence between the first antique coin evaluation network branch and the second antique coin evaluation network branch based on the front-to-back connection order. Among the evaluation subunits obtained by the evaluation unit with the earlier feature processing sequence, select a model training environment as the feature acquisition node; among the evaluation subunits obtained by the evaluation unit with the later feature processing sequence, select a model training environment as the feature fusion node, and establish a cross-modal connection path between the feature acquisition node and the feature fusion node. Generate the evaluation training framework of the antique coin evaluation model in this way. When the server executes the training process of the antique coin evaluation model, divide the training phase into multiple training steps in the training session, and one training step corresponds to the evaluation dimension of an antique coin evaluation network branch. For example, the training session may include steps such as initial feature extraction and feature optimization. Among them, the initial feature extraction step may correspond to the preliminary recognition and evaluation dimension of the coin pattern by a certain branch. When executing the training session, the server determines the target evaluation unit for executing the current training step. For example, when conducting the preliminary recognition and evaluation of the coin pattern, determine the corresponding evaluation unit as the evaluation unit responsible for this evaluation dimension (assume it is the first evaluation unit), and call the target evaluation unit to execute the evaluation dimension training process corresponding to the antique coin training direction of its corresponding antique coin evaluation network branch (the first antique coin evaluation network branch). Through multiple evaluation units respectively executing the evaluation dimension training processes of their corresponding branches, finally obtain the trained antique coin evaluation model. The server imports the antique coin image to be evaluated into the trained antique coin evaluation model, and the model analyzes and processes the image based on the capabilities obtained through training, comprehensively evaluates the appearance state of the antique coin from multiple evaluation dimensions, and finally obtains the comprehensive evaluation result of the appearance state of the antique coin image to be evaluated, such as giving evaluation information such as the wear level of the coin and the pattern clarity score.

[0024] In the embodiment of the present invention, the antique coin training directions of each of the antique coin evaluation network branches are also used to represent the evaluation configuration parameters of each of the antique coin evaluation network branches, and each of the antique coin evaluation network branches is represented as the first antique coin evaluation network branch;

[0025] Perform evaluation dimension allocation processing on the first antique coin evaluation network branch to obtain the antique coin training direction of the first antique coin evaluation network branch, which can be implemented through the following examples.

[0026] Obtain the evaluation dimension feature domain of the first antique coin evaluation network branch, where the evaluation dimension feature domain of the first antique coin evaluation network branch includes multiple standard antique coin training directions predetermined for the first antique coin evaluation network branch;

[0027] Based on the standard evaluation configuration parameters corresponding to each of the standard antique coin training directions in the evaluation dimension feature domain of the first antique coin evaluation network branch, determine the evaluation period of the first antique coin evaluation network branch in each of the standard antique coin training directions;

[0028] Select the standard antique coin training direction corresponding to the shortest evaluation period in the evaluation dimension feature domain of the first antique coin evaluation network branch as the antique coin training direction of the first antique coin evaluation network branch;

[0029] Among them, the evaluation configuration parameters include sample evaluation configuration parameters and architecture evaluation configuration parameters; the sample evaluation configuration parameters include evaluation granularity and sample division dimension; the architecture evaluation configuration parameters include training session process segmentation dimension and feature analysis dimension.

[0030] In an embodiment of the present invention, exemplarily, the server obtains the evaluation dimension feature domain of the first antique coin evaluation network branch from a preset storage area. For example, in a database dedicated to storing parameters and data related to antique coin evaluation, there is a specific data table recording information related to each evaluation network branch. For the first antique coin evaluation network branch, its evaluation dimension feature domain contains multiple pre-determined standard antique coin training directions. Suppose these standard training directions are: evaluating from the perspective of the clarity of the characters on the coin, evaluating from the perspective of the integrity of the patterns on the coin, and evaluating from the perspective of the degree of edge wear of the coin. In terms of sample evaluation configuration parameters: taking the standard antique coin training direction of evaluating from the perspective of the clarity of the characters on the coin as an example, in its corresponding sample evaluation configuration parameters, the evaluation granularity may be set to finely evaluate the characters on the coin according to individual characters; the sample division dimension is set to use the characters on the front and back of the coin as different sample subsets respectively. In terms of architecture evaluation configuration parameters: the process segmentation dimension of the training session may divide the training into several stages such as data preprocessing, feature extraction, model training, and evaluation optimization; the feature analysis dimension is set to analyze from aspects such as the thickness of the strokes of the characters and the color contrast. Calculating the evaluation period: The server standardizes the first antique coin evaluation network branch into a network entity in a given first model training environment. Then, the server selects a sample antique coin image and allows this network entity to process it, thereby obtaining the standard evaluation period on the first model training environment, which is assumed to be 10 seconds. Then, a certain correlation calculation (such as a multiplication operation) is performed between the evaluation granularity (such as individual character evaluation) and the standard evaluation period (10 seconds) to obtain a preliminary result. Then, this preliminary result is divided by the architecture evaluation configuration parameters (for example, divided by a certain influence coefficient brought by the process segmentation dimension and the feature analysis dimension of the training session), and finally the evaluation period under this standard antique coin training direction is obtained, which is assumed to be 8 seconds after calculation. In the same way, the server calculates the evaluation periods under other standard antique coin training directions such as evaluating from the perspective of pattern integrity and evaluating from the perspective of edge wear degree. For example, the evaluation period calculated under the evaluation from the perspective of pattern integrity is 12 seconds, and the evaluation period calculated under the evaluation from the perspective of edge wear degree is 10 seconds. The server compares the evaluation periods calculated for each standard antique coin training direction. In the above example, 8 seconds (evaluation from the perspective of character clarity) < 10 seconds (evaluation from the perspective of edge wear degree) < 12 seconds (evaluation from the perspective of pattern integrity), so the server selects the standard antique coin training direction of evaluating from the perspective of character clarity as the antique coin training direction of the first antique coin evaluation network branch. Through such a process, the server determines the evaluation dimension and the corresponding training process of the first antique coin evaluation network branch in the training stage, and at the same time this training direction also represents the corresponding evaluation configuration parameter situation.

[0031] In an embodiment of the present invention, the multiple standard antique coin training directions include a first standard antique coin training direction; the evaluation configuration parameters include sample evaluation configuration parameters and architecture evaluation configuration parameters, and the sample evaluation configuration parameters include an evaluation granularity;

[0032] Based on the standard evaluation configuration parameters corresponding to the first standard antique coin training direction, determining the evaluation period of the first antique coin evaluation network branch in the first standard antique coin training direction can be implemented through the following examples.

[0033] In the first model training environment, standardize the first antique coin evaluation network branch into the network entity of the first antique coin evaluation network branch, and call the network entity of the first antique coin evaluation network branch to process a sample antique coin image to obtain the standard evaluation period of the first antique coin evaluation network branch in the first model training environment;

[0034] Obtain the first multiplication operation result between the evaluation granularity in the standard evaluation configuration parameters corresponding to the first standard antique coin training direction and the standard evaluation period of the first antique coin evaluation network branch in the first model training environment;

[0035] Take the first division operation result between the first multiplication operation result and the architecture evaluation configuration parameters corresponding to the first standard antique coin training direction as the evaluation period of the first antique coin evaluation network branch in the first standard antique coin training direction.

[0036] In an embodiment of the present invention, exemplarily, the server runs in a specific first model training environment, which includes established hardware configurations (such as a certain model of CPU, GPU, a certain amount of memory, etc.) and software settings (a specific version of the deep learning framework, operating system, etc.). The server standardizes the first antique coin evaluation network branch to make it an executable network entity, just like transforming a design blueprint into an actually operable machine. Then, the server randomly selects an image from the database storing sample antique coin images. For example, the database stores a large number of antique coin images of different ages and types, and the server selects the front image of a Qing Dynasty copper coin. Then, the server calls the network entity of the standardized first antique coin evaluation network branch to process this sample antique coin image. During the processing, the network entity analyzes the image according to its internal algorithms and structures. After the processing is completed, the server records the time taken for the entire processing process to obtain the standard evaluation period of the first antique coin evaluation network branch in the first model training environment. Suppose this period is 15 seconds. The server obtains the evaluation granularity in the standard evaluation configuration parameters corresponding to the first standard antique coin training direction. Suppose the first standard antique coin training direction is to evaluate the detail degree of the coin pattern, and its evaluation granularity is set to divide the pattern according to the smallest element unit for evaluation. For example, the dragon pattern on the coin is refined to each dragon scale. The evaluation granularity can be represented by a quantitative value. Suppose it is 10 here (representing that the pattern is divided into 10 smallest evaluation units). The server multiplies the obtained evaluation granularity 10 by the standard evaluation period of 15 seconds of the first antique coin evaluation network branch in the first model training environment, that is, 10×15 = 150, to obtain the first multiplication result 150. The server obtains the architecture evaluation configuration parameters corresponding to the first standard antique coin training direction. For example, these architecture evaluation configuration parameters include the process segmentation dimension of the training link (dividing the training into four stages: data preparation, feature extraction, model training, and optimization adjustment) and the feature analysis dimension (analyzing from three aspects: the shape, color, and texture of the pattern). These factors can be represented by a coefficient. Suppose it is 5. The server divides the first multiplication result 150 by the coefficient 5 represented by the architecture evaluation configuration parameters corresponding to the first standard antique coin training direction, that is, 150÷5 = 30, and takes the obtained result 30 as the evaluation period of the first antique coin evaluation network branch in the first standard antique coin training direction. Through such a series of operations, the server accurately determines the evaluation period of this evaluation network branch in a specific standard training direction, providing key data for subsequent selection of the optimal training direction.

[0037] Before determining the evaluation period of the first antique coin evaluation network branch in each of the standard antique coin training directions based on the standard evaluation configuration parameters corresponding to the standard antique coin training directions in the evaluation dimension feature domain of the first antique coin evaluation network branch, the following implementation manners are also provided.

[0038] Obtain the standard processing loads of the multiple antique coin evaluation network branches in the same model training environment;

[0039] Determine the target antique coin evaluation network branch with the largest standard processing load from the multiple antique coin evaluation network branches;

[0040] If the target antique coin evaluation network branch is not the first antique coin evaluation network branch, adjust each of the standard antique coin training directions in the evaluation dimension feature domain of the first antique coin evaluation network branch according to the antique coin training direction of the target antique coin evaluation network branch.

[0041] In an embodiment of the present invention, exemplarily, the server is in a unified model training environment, which has specific hardware resources (such as multi-core CPUs, high-performance GPUs, and a certain amount of memory, etc.) and software conditions (specific versions of deep learning frameworks, operating systems, etc.). In this environment, the server runs multiple antique coin evaluation network branches simultaneously. For example, assume there are three antique coin evaluation network branches: the first antique coin evaluation network branch is responsible for evaluating the text-related features of the coins, the second antique coin evaluation network branch focuses on the recognition of coin patterns, and the third antique coin evaluation network branch mainly deals with the state evaluation of the coin edges. The server obtains the resource occupancy of each branch during operation in real time through a monitoring system, including indicators such as CPU usage rate, GPU usage rate, and memory occupancy. After a period of monitoring, the server calculates the standard processing load of each branch based on these indicators. Assume that the standard processing load of the first antique coin evaluation network branch is 30% (indicating that it occupies 30% of the system resources), the standard processing load of the second antique coin evaluation network branch is 50%, and the standard processing load of the third antique coin evaluation network branch is 20%. The server compares the standard processing loads of the obtained antique coin evaluation network branches. In the above example, 50% (the second antique coin evaluation network branch) > 30% (the first antique coin evaluation network branch) > 20% (the third antique coin evaluation network branch), so the server determines that the second antique coin evaluation network branch is the target antique coin evaluation network branch with the largest standard processing load. Since the target antique coin evaluation network branch (the second antique coin evaluation network branch) is not the first antique coin evaluation network branch, the server starts an adjustment operation. Assume that in the sample evaluation configuration parameters corresponding to the antique coin training direction of the second antique coin evaluation network branch, the evaluation granularity is to divide the pattern according to the main elements for evaluation (for example, dividing the flower pattern on the coin into the flower and branches for separate evaluation), and the sample division dimension is to evaluate the front and back patterns of the coin separately. The server obtains the multiplication operation result between the evaluation granularity and the sample division dimension of the second antique coin evaluation network branch. Assume that the evaluation granularity is 5 (representing that the pattern is divided into 5 main element categories), and the sample division dimension is 2 (front and back), then the multiplication result is 5×2 = 10. For a certain standard antique coin training direction (such as evaluating the style characteristics of coin text) in the evaluation dimension feature domain of the first antique coin evaluation network branch, its original sample division dimension is 3 (dividing the text into three categories: seal script, official script, and regular script according to the font type). The server divides the above multiplication result 10 by the sample division dimension 3 corresponding to this standard antique coin training direction, that is, 10÷3≈3.33.The server can adjust the evaluation granularity corresponding to the training direction of the standard antique coin according to this calculation result. For example, it can refine the originally coarser evaluation granularity to make it more coordinated with the training direction of the target antique coin evaluation network branch in terms of resource utilization and evaluation focus. Thus, similar adjustment operations are performed on each standard antique coin training direction in the evaluation dimension feature domain of the first antique coin evaluation network branch.

[0042] In the embodiment of the present invention, the multiple standard antique coin training directions include the first standard antique coin training direction; the evaluation configuration parameters include sample evaluation configuration parameters, and the sample evaluation configuration parameters include evaluation granularity and sample division dimension.

[0043] Adjusting the first standard antique coin training direction in the evaluation dimension feature domain of the first antique coin evaluation network branch according to the antique coin training direction of the target antique coin evaluation network branch can be implemented through the following examples.

[0044] Obtain the second multiplication operation result between the evaluation granularity and the sample division dimension corresponding to the antique coin training direction of the target antique coin evaluation network branch.

[0045] Use the second division operation result between the second multiplication operation result and the sample division dimension corresponding to the first standard antique coin training direction to adjust the evaluation granularity corresponding to the first standard antique coin training direction; or,

[0046] Use the third division operation result between the second multiplication operation result and the evaluation granularity corresponding to the first standard antique coin training direction to adjust the sample division dimension corresponding to the first standard antique coin training direction.

[0047] In an embodiment of the present invention, exemplarily, the server has determined a target antique coin evaluation network branch (assumed to be a branch focusing on evaluating the complexity of coin patterns), and obtained the sample evaluation configuration parameters corresponding to its antique coin training direction. Among them, the evaluation granularity is set to divide the pattern according to key components. For example, the figure pattern on the coin is divided into 5 parts such as the head, body, and clothing, that is, the evaluation granularity is 5; the sample division dimension is to evaluate the front and back patterns of the coin separately, that is, the sample division dimension is 2. The server multiplies the evaluation granularity and the sample division dimension of the target antique coin evaluation network branch to obtain a second multiplication result, that is, 5×2 = 10. Assume that the first standard antique coin training direction in the evaluation dimension feature domain of the first antique coin evaluation network branch is to evaluate the clarity of coin characters, and its originally corresponding sample division dimension is 3 (divide the characters into three categories according to their positions on the coin: edge characters, middle characters, and corner characters). The server divides the second multiplication result 10 by the sample division dimension 3 corresponding to the first standard antique coin training direction to obtain a second division result, that is, 10÷3≈3.33. The server adjusts the evaluation granularity corresponding to the first standard antique coin training direction according to this second division result. For example, originally the evaluation granularity is to evaluate the characters according to the overall outline, and now according to the calculation result, the evaluation granularity is refined to evaluate the characters according to the stroke details, making the evaluation more precise to adapt to the training direction characteristics of the target antique coin evaluation network branch and achieving better coordination in resource utilization and evaluation focus. Still taking the first standard antique coin training direction of evaluating the clarity of coin characters as an example, assume that its originally corresponding evaluation granularity is 4 (evaluate the character clarity in four grades). The server divides the second multiplication result 10 by the evaluation granularity 4 corresponding to the first standard antique coin training direction to obtain a third division result, that is, 10÷4 = 2.5. The server adjusts the sample division dimension corresponding to the first standard antique coin training direction according to this third division result. For example, originally the characters are divided into three categories according to their positions on the coin, and now according to the calculation result, they are adjusted to be divided into two categories according to the font style (seal script and non-seal script), thereby changing the sample division dimension to make the evaluation method of the first antique coin evaluation network branch in this standard training direction more consistent with the training direction of the target antique coin evaluation network branch, which helps to improve the training efficiency and evaluation accuracy of the entire antique coin evaluation model.

[0048] In an embodiment of the present invention, the evaluation training framework of the antique coin evaluation model can be configured based on the antique coin training directions of the multiple antique coin evaluation network branches, and can be implemented through the following examples.

[0049] Based on the antique coin training directions of each of the antique coin evaluation network branches, construct evaluation units corresponding to each of the antique coin evaluation network branches;

[0050] Establish a multi-modal feature fusion architecture among multiple evaluation units to generate an evaluation training framework for the antique coin evaluation model.

[0051] In an embodiment of the present invention, by way of example, it is assumed that the antique coin evaluation model includes three antique coin evaluation network branches: the first branch is responsible for evaluating the degree of wear of the coin, the second branch is responsible for evaluating the color of the coin pattern, and the third branch is responsible for evaluating the integrity of the coin characters. For the first antique coin evaluation network branch, among the evaluation configuration parameters corresponding to its antique coin training direction, the sample division dimension divides the coin surface into three parts: the front, the back, and the edge. The architecture evaluation configuration parameters include dividing the training process into four stages: data collection, feature extraction, model training, and optimization adjustment. Based on these parameters, the server determines the number of model training environments in the first evaluation unit. For example, according to past experience and computing resource conditions, it is set to 3. The server then obtains the environment configurations in the pre-set model training environment set, which contains various environments with different combinations of hardware performance (such as different models of CPUs, GPUs) and software versions (different versions of deep learning frameworks). The server runs the first antique coin evaluation network branch on the model training environments with each environment configuration respectively, and obtains its standard evaluation period. Suppose the standard evaluation period is 10 seconds on environment A, 12 seconds on environment B, and 8 seconds on environment C. The server then determines the evaluation period on the model training environments with each environment configuration based on the evaluation granularity (such as dividing the degree of wear into three levels: mild, moderate, and severe) corresponding to the antique coin training direction of the first antique coin evaluation network branch, the architecture evaluation configuration parameters, and the above standard evaluation period. After calculation, the evaluation period of environment A is 15 seconds, that of environment B is 18 seconds, and that of environment C is 10 seconds. Thus, the server determines that the target environment configuration corresponding to the shortest evaluation period is environment C. When the number of model training environments with the target environment configuration is greater than or equal to the number of model training environments in the first evaluation unit, the server uses the target environment configuration as the environment configuration of the model training environments in the first evaluation unit. Finally, the server standardizes the first antique coin evaluation network branch in the model training environment set based on the antique coin training direction of the first antique coin evaluation network branch, and combines the number and environment configuration of the model training environments in the first evaluation unit, and successfully constructs the first evaluation unit. In the same way, the server constructs the evaluation units corresponding to the second and third antique coin evaluation network branches respectively. Taking the first evaluation unit (wear degree evaluation) and the second evaluation unit (pattern color evaluation) as examples, the server obtains the common evaluation base between the sample division dimension (front, back, edge) corresponding to the antique coin training direction of the first antique coin evaluation network branch and the sample division dimension (such as dividing the pattern into the main pattern and the background pattern) corresponding to the antique coin training direction of the second antique coin evaluation network branch, which is assumed to be 2. The server uses the common evaluation base as the number of evaluation unit pools, and groups the first evaluation unit and the second evaluation unit based on this, obtaining 2 evaluation unit pools.Each evaluation unit pool contains at least one evaluation subunit from the first evaluation unit (each evaluation subunit contains one or more model training environments) and at least one evaluation subunit from the second evaluation unit. When establishing the multi-modal feature fusion architecture, the server configures dynamic resource allocation ratios for each evaluation unit pool. For example, according to the importance and computational workload of wear degree evaluation and pattern color evaluation, 60% of the computing resources are allocated to the first evaluation unit pool, and 40% of the computing resources are allocated to the second evaluation unit pool. The server determines the feature processing flow sequence between the first evaluation unit and the second evaluation unit according to the front-to-back connection order between the first ancient coin evaluation network branch and the second ancient coin evaluation network branch. Assume that the wear degree evaluation is performed first, and then the pattern color evaluation is performed. The server selects a model training environment as a feature acquisition node among the evaluation subunits obtained from the first evaluation unit; and selects a model training environment as a feature fusion node among the evaluation subunits obtained from the second evaluation unit, and establishes a cross-modal connection path between the two. For the third evaluation unit (character integrity evaluation), the server also establishes a multi-modal feature fusion architecture with the first and second evaluation units in a similar manner, and finally generates an evaluation training framework for the ancient coin evaluation model.

[0052] In the embodiment of the present invention, each of the ancient coin evaluation network branches is represented as the first ancient coin evaluation network branch, and the ancient coin evaluation model further includes a second ancient coin evaluation network branch connected before and after the first ancient coin evaluation network branch. The first ancient coin evaluation network branch corresponds to the first evaluation unit, and the second ancient coin evaluation network branch corresponds to the second evaluation unit;

[0053] The establishment of the multi-modal feature fusion architecture between multiple evaluation units can be implemented through the following examples.

[0054] Group the first evaluation unit and the second evaluation unit to obtain multiple evaluation unit pools; each evaluation unit pool includes at least one evaluation subunit obtained from the first evaluation unit and at least one evaluation subunit obtained from the second evaluation unit;

[0055] When establishing the multi-modal feature fusion architecture between the first evaluation unit and the second evaluation unit, configure dynamic resource allocation ratios for each evaluation unit pool, and establish a cross-modal connection path between the evaluation subunits belonging to different evaluation units in each evaluation unit pool.

[0056] In an embodiment of the present invention, by way of example, assume that the first antique coin evaluation network branch is responsible for evaluating the surface material condition of antique coins, and its corresponding first evaluation unit has been constructed, including multiple evaluation sub-units, and each evaluation sub-unit corresponds to a different model training environment. For example, evaluation sub-unit A runs in a model training environment equipped with a high-end GPU, and evaluation sub-unit B runs in a model training environment with a general CPU. The second antique coin evaluation network branch is responsible for evaluating the pattern style of antique coins, and its corresponding second evaluation unit also includes multiple evaluation sub-units. For example, evaluation sub-unit C runs in a model training environment of a specific version of the deep learning framework, and evaluation sub-unit D runs in a model training environment of another version of the deep learning framework. The server obtains the common evaluation base between the sample division dimension corresponding to the antique coin training direction of the first antique coin evaluation network branch (such as dividing the coin surface into a central area and an edge area) and the sample division dimension corresponding to the antique coin training direction of the second antique coin evaluation network branch (such as dividing the pattern into a traditional pattern and an innovative pattern). Assume that the calculated common evaluation base is 2. The server uses this common evaluation base 2 as the number of evaluation unit pools, and groups the first evaluation unit and the second evaluation unit based on this. The server selects evaluation sub-unit A from evaluation sub-units A and B of the first evaluation unit, and selects evaluation sub-unit C from evaluation sub-units C and D of the second evaluation unit to form the first evaluation unit pool; then selects evaluation sub-unit B of the first evaluation unit and evaluation sub-unit D of the second evaluation unit to form the second evaluation unit pool. In this way, each evaluation unit pool includes at least one evaluation sub-unit obtained by the first evaluation unit and at least one evaluation sub-unit obtained by the second evaluation unit. When establishing a multi-modal feature fusion architecture between the first evaluation unit and the second evaluation unit, the server configures a dynamic resource ratio for each evaluation unit pool according to the requirements of the antique coin evaluation task and the computational complexity of each evaluation unit. For example, since the evaluation of the coin surface material condition is relatively more important and has a larger amount of calculation in the current evaluation task, the server allocates 70% of the system computing resources to the first evaluation unit pool (including evaluation sub-units A and C), and allocates 30% of the system computing resources to the second evaluation unit pool (including evaluation sub-units B and D). The server determines the feature processing flow sequence between the first evaluation unit and the second evaluation unit according to the front-back connection order between the first antique coin evaluation network branch and the second antique coin evaluation network branch (assuming that the surface material condition is evaluated first and then the pattern style is evaluated). In evaluation sub-unit A obtained by the first evaluation unit with the earlier feature processing flow sequence, the server selects this model training environment as the feature acquisition node; in evaluation sub-unit C obtained by the second evaluation unit with the later feature processing flow sequence, the server selects this model training environment as the feature fusion node.Then, the server establishes a cross-modal connection path between the feature acquisition node (the model training environment of the evaluation subunit A) and the feature fusion node (the model training environment of the evaluation subunit C), so that the feature information collected by the first evaluation unit during the evaluation of the surface material condition of the coin can be smoothly transmitted to the second evaluation unit for feature fusion during the pattern style evaluation. For the second evaluation unit pool, the server also establishes a cross-modal connection path between the evaluation subunit B and the evaluation subunit D in a similar manner, thereby completing the establishment of a multi-modal feature fusion architecture between the first evaluation unit and the second evaluation unit to support the more efficient evaluation training of the antique coin evaluation model.

[0057] In the embodiment of the present invention, the antique coin training directions of each of the antique coin evaluation network branches are also used to characterize the evaluation configuration parameters of each of the antique coin evaluation network branches. The evaluation configuration parameters include sample evaluation configuration parameters, and the sample evaluation configuration parameters include sample division dimensions.

[0058] The grouping of the first evaluation unit and the second evaluation unit to obtain multiple evaluation unit pools can be implemented through the following examples.

[0059] Obtain the common evaluation base between the sample division dimension corresponding to the antique coin training direction of the first antique coin evaluation network branch and the sample division dimension corresponding to the antique coin training direction of the second antique coin evaluation network branch.

[0060] Use the common evaluation base as the number of evaluation unit pools.

[0061] Based on the number of evaluation unit pools, group the first evaluation unit and the second evaluation unit to obtain multiple evaluation unit pools.

[0062] In an embodiment of the present invention, exemplarily, the first antique coin evaluation network branch is responsible for evaluating the character features of antique coins. Among the sample evaluation configuration parameters corresponding to its antique coin training direction, the sample division dimension is set to divide the characters on the coin into three categories: seal script, official script, and regular script according to the font type. The second antique coin evaluation network branch is responsible for evaluating the pattern features of the coin. The sample division dimension corresponding to its antique coin training direction is to divide the patterns into three categories: human figure patterns, animal patterns, and plant patterns according to the theme. The server analyzes these two sample division dimensions and determines the common evaluation base between them through specific algorithms or rules. For example, the server can find the greatest common divisor of the number of categories in the two sample division dimensions. In this example, the greatest common divisor of 3 and 3 is 3, so the common evaluation base is 3. The server directly uses the obtained common evaluation base 3 as the number of evaluation unit pools. This is because the common evaluation base reflects a certain degree of association or commonality between the two evaluation network branches in the sample division dimension. Using this as the number of evaluation unit pools helps to better integrate the resources and feature processing of the two evaluation units in the subsequent grouping process. The first evaluation unit contains multiple evaluation sub-units, and these evaluation sub-units respectively correspond to different model training environments. For example, evaluation sub-unit E runs in a model training environment with a high memory capacity and is specifically used to process a large amount of character image data; evaluation sub-unit F runs in a model training environment equipped with a high-speed CPU and can quickly perform preliminary extraction of character features. The second evaluation unit also contains multiple evaluation sub-units. For example, evaluation sub-unit G runs in a model training environment installed with a specific image recognition library, which is beneficial for pattern recognition; evaluation sub-unit H runs in an optimized deep learning framework environment and can efficiently perform pattern feature analysis. Based on the determined number of evaluation unit pools 3, the server groups the first evaluation unit and the second evaluation unit. The server selects evaluation sub-unit E from evaluation sub-units E and F of the first evaluation unit, and selects evaluation sub-unit G from evaluation sub-units G and H of the second evaluation unit to form the first evaluation unit pool; then selects evaluation sub-unit F of the first evaluation unit and evaluation sub-unit H of the second evaluation unit to form the second evaluation unit pool; then combines the remaining part of the resources of the first evaluation unit or an auxiliary evaluation sub-part with another auxiliary evaluation sub-part of the second evaluation unit to form the third evaluation unit pool. Through the above steps, the server successfully groups the first evaluation unit and the second evaluation unit, obtaining multiple evaluation unit pools. Each evaluation unit pool contains evaluation sub-units from two different evaluation units, laying a foundation for the subsequent establishment of a multi-modal feature fusion architecture, enabling better coordination of the evaluation processes related to character features and pattern features during the antique coin evaluation process.

[0063] In an embodiment of the present invention, each of the evaluation sub-units contains one or more model training environments;

[0064] To establish cross-modal connection paths between evaluation sub-units belonging to different evaluation units in each of the evaluation unit pools, the implementation can be carried out through the following examples.

[0065] Determine the feature processing flow sequence between the first evaluation unit and the second evaluation unit according to the front-back connection order between the first ancient coin evaluation network branch and the second ancient coin evaluation network branch;

[0066] Among the evaluation sub-units obtained by the evaluation unit with the earlier feature processing flow sequence, select a model training environment as the feature acquisition node;

[0067] Among the evaluation sub-units obtained by the evaluation unit with the later feature processing flow sequence, select a model training environment as the feature fusion node;

[0068] Establish a cross-modal connection path between the feature acquisition node and the feature fusion node.

[0069] In an embodiment of the present invention, by way of example, assume that the server is performing an operation to establish a cross-modal connection path between the first evaluation unit and the second evaluation unit of the antique coin evaluation model. The first antique coin evaluation network branch is responsible for evaluating the texture features of the antique coin, and the second antique coin evaluation network branch is responsible for evaluating the color features of the antique coin. The server determines the feature processing sequence between the first evaluation unit and the second evaluation unit according to the front-to-back connection order between the first antique coin evaluation network branch and the second antique coin evaluation network branch. During the antique coin evaluation process, the evaluation result of the texture features will affect the analysis of the color features. Therefore, the texture feature evaluation is performed first, and then the color feature evaluation is performed. That is, the feature processing flow corresponding to the first evaluation unit is in the front, and the feature processing flow corresponding to the second evaluation unit is in the back, and such a sequence is determined. The first evaluation unit contains multiple evaluation sub-units, and each evaluation sub-unit corresponds to a different model training environment. Evaluation sub-unit I runs in a model training environment equipped with professional image texture analysis software and can extract the texture details of the antique coin image with high precision; evaluation sub-unit J runs in a hardware environment with powerful parallel computing capabilities and can quickly process a large amount of texture data. Since the first evaluation unit is in the front in the feature processing sequence, the server selects the model training environment of evaluation sub-unit I from the obtained evaluation sub-units as the feature acquisition node. Because this environment has professional texture analysis software and is more suitable for collecting data information related to texture features. The second evaluation unit also contains multiple evaluation sub-units. Evaluation sub-unit K runs in a model training environment installed with an advanced color recognition algorithm library and can accurately identify the color information of the antique coin image; evaluation sub-unit L runs in an optimized deep learning framework environment and can efficiently perform the fusion and analysis of color features. Since the second evaluation unit is in the back in the feature processing sequence, the server selects the model training environment of evaluation sub-unit K from the obtained evaluation sub-units as the feature fusion node. Because this environment has an advanced color recognition algorithm library and can better fuse the texture features collected by the first evaluation unit with its own color features. After the server determines the feature acquisition node (the model training environment of evaluation sub-unit I) and the feature fusion node (the model training environment of evaluation sub-unit K), it establishes a cross-modal connection path between the two. The server configures specific network protocols and data transmission interfaces to ensure that the antique coin texture feature data collected by the first evaluation unit can be smoothly transmitted to the feature fusion node of the second evaluation unit. For example, the server sets up a data transmission channel based on the TCP / IP protocol and performs encryption and compression processing on the data to ensure the security and efficiency of the data during transmission. By establishing this cross-modal connection path, the texture features and color features of the antique coin can be effectively interacted and fused between the two evaluation units, thereby improving the comprehensive evaluation ability of the antique coin evaluation model.

[0070] In an embodiment of the present invention, each of the antique coin evaluation network branches is represented as a first antique coin evaluation network branch, and the first antique coin evaluation network branch corresponds to a first evaluation unit; the antique coin training directions of each of the antique coin evaluation network branches are further used to characterize the evaluation configuration parameters of each of the antique coin evaluation network branches, and the evaluation configuration parameters include sample evaluation configuration parameters and architecture evaluation configuration parameters, and the sample evaluation configuration parameters include sample division dimensions.

[0071] Based on the antique coin training direction of the first antique coin evaluation network branch, the first evaluation unit can be configured by the following examples for implementation.

[0072] Determine the number of model training environments in the first evaluation unit based on the sample division dimension and the architecture evaluation configuration parameter in the evaluation configuration parameter corresponding to the antique coin training direction of the first antique coin evaluation network branch.

[0073] Obtain the environment configuration in the preset model training environment set, and determine the evaluation period of the first antique coin evaluation network branch on the model training environments with each environment configuration based on the antique coin training direction of the first antique coin evaluation network branch.

[0074] Determine the environment configuration of the model training environment in the first evaluation unit based on the evaluation period of the first antique coin evaluation network branch on the model training environments with each environment configuration.

[0075] Normalize the first antique coin evaluation network branch in the model training environment set based on the antique coin training direction of the first antique coin evaluation network branch and based on the number and environment configuration of the model training environments in the first evaluation unit, so as to configure the first evaluation unit.

[0076] In an embodiment of the present invention, by way of example, assume that the server is processing the configuration work of the first antique coin evaluation network branch and its corresponding first evaluation unit in the antique coin evaluation model. The first antique coin evaluation network branch is responsible for evaluating the rust condition of antique coins. Among the sample evaluation configuration parameters corresponding to the training direction of antique coins of the first antique coin evaluation network branch, the sample division dimension divides the surface of the antique coin into three categories according to the rust distribution area, namely the central rust area, the edge rust area, and the mixed rust area; the architecture evaluation configuration parameters include dividing the training process into four stages: data preprocessing, feature extraction, model training, and accuracy verification. The server determines the number of model training environments in the first evaluation unit based on these evaluation configuration parameters, taking into account the complexity of the evaluation task and the expected processing efficiency. Since the evaluation of the rust condition requires processing image data in different regions and the training process is relatively complex, the server determines that the number of model training environments is 4. The server obtains a preset set of model training environments, which contains various different environment configurations. For example, environment M is configured with a high-performance GPU and a large amount of memory, suitable for processing complex image data; environment N is configured with an ordinary CPU and a medium amount of memory; environment O installs a specific version of the deep learning framework, which is optimized for image feature extraction; environment P has a special setting in network transmission. Based on the training direction of antique coins of the first antique coin evaluation network branch, the server respectively runs test tasks of this branch on the model training environments with the above various environment configurations. The server selects a set of antique coin image samples with different rust conditions, allows the first antique coin evaluation network branch to process them, records the time from the start to the completion of the evaluation, and obtains the evaluation cycle on the model training environments with various environment configurations. Assume that the evaluation cycle on environment M is 12 seconds, on environment N is 20 seconds, on environment O is 15 seconds, and on environment P is 18 seconds. The server compares the evaluation cycles of the first antique coin evaluation network branch on the model training environments with various environment configurations. Since the evaluation cycle of environment M is the shortest, which is 12 seconds and can complete the rust condition evaluation task more efficiently, the server determines to use the configuration of environment M as the environment configuration of the model training environment in the first evaluation unit. Based on the training direction of antique coins of the first antique coin evaluation network branch, as well as the determined number (4) and environment configuration (environment M) of the model training environment in the first evaluation unit, the server performs a standardization operation on the first antique coin evaluation network branch in the set of model training environments. The server deploys the same hardware and software conditions for the 4 model training environments in the first evaluation unit according to the configuration requirements of environment M, including installing the corresponding GPU driver, deep learning framework version, etc.Meanwhile, according to the training direction of the first antique coin evaluation network branch, set the parameters in the model training environment, such as adjusting the algorithm parameters related to the extraction of rust features, etc., so as to complete the configuration work of the first evaluation unit and enable it to efficiently perform the evaluation task of the rust condition of antique coins.

[0077] In the embodiment of the present invention, determining the environment configuration of the model training environment in the first evaluation unit based on the evaluation period of the first antique coin evaluation network branch on the model training environments configured in each environment can be implemented through the following examples.

[0078] Based on the evaluation period of the first antique coin evaluation network branch on the model training environments configured in each environment, determine the target environment configuration corresponding to the shortest evaluation period;

[0079] Obtain the number of model training environments with the target environment configuration from the set of model training environments;

[0080] When the number of model training environments with the target environment configuration is greater than or equal to the number of model training environments in the first evaluation unit, use the target environment configuration as the environment configuration of the model training environment in the first evaluation unit.

[0081] In an embodiment of the present invention, by way of example, assume that the server is determining the environmental configuration of the model training environment for the first evaluation unit in the antique coin evaluation model, and the first antique coin evaluation network branch is responsible for evaluating the pattern fineness on the antique coin. The server has obtained the evaluation cycles of the first antique coin evaluation network branch on the model training environments with different environmental configurations. There are three different environmental configurations in the model training environment set: Environment X is equipped with high-end GPUs and optimized deep learning software, and is specifically used for fast operations for processing complex images; Environment Y uses ordinary CPUs and common versions of deep learning frameworks; Environment Z has special optimizations in terms of storage and data reading. The first antique coin evaluation network branch evaluates a set of antique coin image samples with various pattern finenesses on Environment X, and the time taken to complete the evaluation is 8 seconds; when evaluating the same samples on Environment Y, the time taken is 15 seconds; when evaluating on Environment Z, the time taken is 12 seconds. The server compares these three evaluation cycles, 8 seconds < 12 seconds < 15 seconds, and thus determines that Environment X is the target environmental configuration corresponding to the shortest evaluation cycle because it can complete the evaluation task of the pattern fineness of antique coins at the fastest speed. The server queries the number of model training environments owned by the target environmental configuration (Environment X) from the database or resource management system storing the model training environment information. Assume that through the query, it is known that there are currently 6 available model training environments of Environment X within the resource range of the server. The server previously determined the number of model training environments in the first evaluation unit to be 5 based on factors such as the evaluation task requirements of the first antique coin evaluation network branch. Since the number of model training environments of the target environmental configuration (Environment X) is 6, which is greater than the number of model training environments in the first evaluation unit, i.e., 5, meeting the condition of "greater than or equal to". Therefore, the server uses the target environmental configuration (Environment X) as the environmental configuration of the model training environment in the first evaluation unit. Next, the server will deploy and set up the 5 model training environments in the first evaluation unit according to the hardware and software configuration requirements of Environment X, such as installing the same high-end GPU driver as Environment X, configuring the corresponding deep learning software parameters, etc., to ensure that the first evaluation unit can operate in an efficient environment and accurately evaluate the pattern fineness of antique coins.

[0082] In an embodiment of the present invention, the sample evaluation configuration parameter further includes an evaluation granularity;

[0083] Determining the evaluation cycles of the first antique coin evaluation network branch on the model training environments with different environmental configurations based on the antique coin training direction of the first antique coin evaluation network branch includes:

[0084] Obtaining the standard evaluation cycles of the first antique coin evaluation network branch on the model training environments with different environmental configurations;

[0085] Determine the evaluation period of the first antique coin evaluation network branch in the model training environments with various environmental configurations based on the evaluation granularity, architecture evaluation configuration parameters corresponding to the antique coin training direction of the first antique coin evaluation network branch, and the standard evaluation period of the first antique coin evaluation network branch in the model training environments with various environmental configurations.

[0086] In an embodiment of the present invention, by way of example, assume that the server is handling the determination of the evaluation cycle of the first antique coin evaluation network branch in the antique coin evaluation model under different model training environments, and the first antique coin evaluation network branch is responsible for evaluating the wear condition of antique coins. There are three different environmental configurations in the set of model training environments managed by the server: Environment A has a combination of high-performance CPUs and GPUs and a relatively large memory capacity; Environment B uses a CPU with ordinary performance and a relatively small memory; Environment C has an advantage in data transmission bandwidth. The server standardizes the first antique coin evaluation network branch on the model training environments with these three environmental configurations respectively to make it a runnable network entity. Then, a set of representative samples is selected from the database storing antique coin image samples, and these samples cover antique coin images with different degrees of wear. The server allows the network entity of the first antique coin evaluation network branch to process this set of samples on the model training environment of Environment A, records the time from start to end, and obtains the standard evaluation cycle on Environment A as 10 seconds. Similarly, the standard evaluation cycle on Environment B is 18 seconds, and the standard evaluation cycle on Environment C is 12 seconds. Among the sample evaluation configuration parameters corresponding to the antique coin training direction of the first antique coin evaluation network branch, the evaluation granularity is set to divide the wear condition of antique coins into five grades according to the proportion of the wear area (slight: proportion of wear area < 10%; relatively light: 10% ≤ proportion of wear area < 25%; moderate: 25% ≤ proportion of wear area < 50%; relatively heavy: 50% ≤ proportion of wear area < 75%; severe: proportion of wear area ≥ 75%). The architecture evaluation configuration parameters include dividing the training process into four stages: data screening, feature extraction, model training, and result verification. For Environment A, based on the evaluation granularity, considering that dividing the wear condition into five grades requires more calculations and judgments. At the same time, combining the resource requirements of the four training stages in the architecture evaluation configuration parameters, the previously obtained standard evaluation cycle of 10 seconds is adjusted. Assume that according to a specific calculation rule (such as considering the increased calculation amount due to the evaluation granularity and the resource occupancy of the architecture evaluation configuration parameters), the evaluation cycle on Environment A is finally determined to be 13 seconds. For Environment B, due to its relatively weak hardware performance, it is less efficient in processing the evaluation tasks of the subdivided wear grades and meeting the requirements of the architecture evaluation configuration parameters. Based on the evaluation granularity and the architecture evaluation configuration parameters, the standard evaluation cycle of 18 seconds is adjusted to obtain the evaluation cycle on Environment B as 22 seconds. For Environment C, although it has an advantage in data transmission, when processing the evaluation tasks of the subdivided wear grades, combined with the architecture evaluation configuration parameters, some additional calculation time is still required. Based on the evaluation granularity and the architecture evaluation configuration parameters, the standard evaluation cycle of 12 seconds is adjusted to obtain the evaluation cycle on Environment C as 15 seconds.Through the above steps, the server comprehensively considers the evaluation granularity, the architecture evaluation configuration parameters, and the standard evaluation period, and determines the evaluation period of the first antique coin evaluation network branch on the model training environments with various environmental configurations, providing an accurate basis for subsequently selecting a suitable model training environment to configure the first evaluation unit.

[0087] In an embodiment of the present invention, when performing a training process on the antique coin evaluation model, multiple evaluation units in the evaluation training framework are called to respectively execute the evaluation dimensions of the corresponding antique coin evaluation network branch based on the training processes corresponding to the antique coin training directions of the corresponding antique coin evaluation network branches, and the implementation can be carried out through the following examples.

[0088] When performing a training process on the antique coin evaluation model, the training stage of the antique coin evaluation model is divided into multiple training steps in the training session, and one training step corresponds to the evaluation dimension of one antique coin evaluation network branch;

[0089] Execute the training session, and when executing the training session, determine the target evaluation unit for executing the current training step, and call the target evaluation unit to execute the evaluation dimension training process of the corresponding antique coin evaluation network branch based on the training process corresponding to the antique coin training direction of the antique coin evaluation network branch corresponding to the target evaluation unit.

[0090] In an embodiment of the present invention, by way of example, assume that the server is performing a training process on an antique coin evaluation model, which includes three antique coin evaluation network branches: Branch A is responsible for evaluating the material of the antique coin, Branch B is responsible for evaluating the age characteristics of the patterns on the coin, and Branch C is responsible for evaluating the clarity of the text. When the server performs the training process on the antique coin evaluation model, the training phase is divided into multiple training sessions, and each training session is further subdivided into multiple training steps. For example, the training session includes a data preprocessing session, a feature extraction session, and a model training session. In the data preprocessing session, it is further divided into training steps such as sample screening and data annotation. Among them, the sample screening step corresponds to the evaluation dimension of Branch A, that is, screening out samples suitable for evaluating the material from a large number of antique coin image samples; the data annotation step corresponds to the evaluation dimension of Branch B, and annotating information such as the age to which the coin pattern belongs. In the feature extraction session, it is divided into a material feature extraction step, a pattern age feature extraction step, and a text clarity feature extraction step, corresponding to the evaluation dimensions of Branch A, Branch B, and Branch C respectively. In the model training session, training steps for each branch evaluation dimension are also set, such as a material evaluation model training step, a pattern age evaluation model training step, and a text clarity evaluation model training step. In this way, one training step corresponds to the evaluation dimension of one antique coin evaluation network branch. When the server executes the material feature extraction step in the feature extraction session, it determines that the target evaluation unit for executing the current training step is the first evaluation unit corresponding to Branch A. The training process corresponding to the antique coin training direction of the first evaluation unit includes: First, divide the samples according to a specific evaluation granularity (classifying the material into metal materials, non-metal materials, etc.); then, according to the architecture evaluation configuration parameters (the training session process segmentation dimension is data import, preliminary feature extraction, fine feature extraction, and feature verification, and the feature analysis dimension includes color, density, etc.), process the samples in the selected model training environment. The server calls the first evaluation unit, and this unit extracts the material features of the antique coin image samples based on the training process corresponding to the antique coin training direction of Branch A. Specifically, first classify the samples according to the evaluation granularity, and then, according to the training session process segmentation dimension, sequentially perform data import, use the preliminary feature extraction algorithm to extract preliminary material features such as color and density, then perform fine feature extraction to obtain more accurate material information, and finally ensure the accuracy and reliability of the extracted features through feature verification, completing the training process corresponding to the evaluation dimension of Branch A. Similarly, when executing other training steps, the server will also determine the corresponding target evaluation unit and call it to execute the evaluation dimension training process according to the training direction of the corresponding antique coin evaluation network branch, gradually completing the training of the entire antique coin evaluation model.

[0091] An embodiment of the present invention provides a computer device 100. The computer device 100 includes a processor and a non-volatile memory storing computer instructions. When the computer instructions are executed by the processor, the computer device 100 executes the foregoing comprehensive evaluation method for the appearance state of antique coins based on joint modeling of image recognition and deep learning. As Figure 2 shown, Figure 2 is a structural block diagram of the computer device 100 provided by an embodiment of the present invention. The computer device 100 includes a memory 111, a processor 112, and a communication unit 113. To achieve data transmission or interaction, the elements of the memory 111, the processor 112, and the communication unit 113 are electrically connected to each other directly or indirectly. For example, these elements can be electrically connected to each other through one or more communication buses or signal lines.

[0092] For illustrative purposes, the foregoing description has been made with reference to specific embodiments. However, the above illustrative discussion is not intended to be exhaustive or to limit the disclosure to the precise forms disclosed. Numerous modifications and variations are possible in light of the above teachings. The embodiments were chosen and described in order to best illustrate the principles of the disclosure and its practical applications, to thereby enable those skilled in the art to best utilize the disclosure and to utilize various embodiments with various modifications as are suited to the particular application contemplated.

Claims

1. A comprehensive evaluation method for the appearance of antique coins based on joint modeling of image recognition and deep learning, characterized in that: include: Acquire an untrained antique coin evaluation model, wherein the antique coin evaluation model includes a plurality of antique coin evaluation network branches; The evaluation dimension allocation process is performed on each antique coin evaluation network branch respectively to obtain the antique coin training direction of each antique coin evaluation network branch; the antique coin training direction of each antique coin evaluation network branch is used to characterize the evaluation dimension of each antique coin evaluation network branch in the training stage and the training process of the evaluation dimension; Based on the antique coin training directions of the multiple antique coin evaluation network branches, an evaluation training framework of the antique coin evaluation model is configured, wherein the evaluation training framework includes multiple evaluation units; one evaluation unit corresponds to one antique coin evaluation network branch, and each evaluation unit is configured based on the antique coin training direction of the corresponding antique coin evaluation network branch; When executing the training process for the antique coin evaluation model, calling multiple evaluation units in the evaluation training framework to respectively execute the evaluation dimension training process of the corresponding antique coin evaluation network branch based on the training process corresponding to the antique coin training direction of the corresponding antique coin evaluation network branch, so as to obtain the antique coin evaluation model that has completed the training; Importing the antique coin image to be evaluated into the trained antique coin evaluation model to obtain a comprehensive evaluation result of the appearance state of the antique coin image to be evaluated; The antique coin training direction based on the multiple antique coin evaluation network branches configures the evaluation training framework of the antique coin evaluation model, including: Based on the antique coin training direction of each antique coin evaluation network branch, construct an evaluation unit corresponding to each antique coin evaluation network branch; Establishing a multimodal feature fusion architecture between multiple evaluation units to generate an evaluation training framework for the antique coin evaluation model; Each of the antique coin evaluation network branches is represented as a first antique coin evaluation network branch, and the antique coin evaluation model also includes a second antique coin evaluation network branch connected to the first antique coin evaluation network branch, the first antique coin evaluation network branch corresponds to a first evaluation unit, and the second antique coin evaluation network branch corresponds to a second evaluation unit; The multimodal feature fusion architecture between multiple evaluation units is established, including: The first evaluation unit and the second evaluation unit are grouped to obtain a plurality of evaluation unit pools; each of the evaluation unit pools includes at least one evaluation sub-unit obtained by the first evaluation unit and at least one evaluation sub-unit obtained by the second evaluation unit; When establishing a multimodal feature fusion architecture between the first evaluation unit and the second evaluation unit, dynamic resource allocation is configured for each evaluation unit pool, and a cross-modal connection path between evaluation sub-units belonging to different evaluation units is established in each evaluation unit pool; The antique coin training direction of each antique coin evaluation network branch is also used to characterize the evaluation configuration parameters of each antique coin evaluation network branch, wherein the evaluation configuration parameters include sample evaluation configuration parameters, and the sample evaluation configuration parameters include sample division dimensions; The first evaluation unit and the second evaluation unit are grouped to obtain a plurality of evaluation unit pools, including: Obtaining a common evaluation cardinality between a sample division dimension corresponding to the antique coin training direction of the first antique coin evaluation network branch and a sample division dimension corresponding to the antique coin training direction of the second antique coin evaluation network branch; Taking the public assessment base as the number of assessment unit pools; Based on the number of the evaluation unit pools, grouping the first evaluation unit and the second evaluation unit to obtain a plurality of evaluation unit pools; Each of the evaluation subunits includes one or more model training environments; The step of establishing a cross-modal connection path between evaluation sub-units belonging to different evaluation units in each evaluation unit pool includes: Determine a feature processing flow sequence between the first evaluation unit and the second evaluation unit according to a front-to-back connection order between the first antique coin evaluation network branch and the second antique coin evaluation network branch; Selecting a model training environment as a feature collection node from among the evaluation sub-units acquired by the previous evaluation unit in the feature processing flow sequence; In each of the evaluation sub-units acquired by the evaluation unit following the feature processing flow sequence, a model training environment is selected as a feature fusion node; A cross-modal connection path is established between the feature collection node and the feature fusion node.

2. The method according to claim 1, characterized in that The antique coin training direction of each antique coin evaluation network branch is also used to characterize the evaluation configuration parameters of each antique coin evaluation network branch, and each antique coin evaluation network branch is represented as a first antique coin evaluation network branch; Performing evaluation dimension allocation processing on the first antique coin evaluation network branch to obtain the antique coin training direction of the first antique coin evaluation network branch includes: Acquire an evaluation dimension feature domain of the first antique coin evaluation network branch, wherein the evaluation dimension feature domain of the first antique coin evaluation network branch includes a plurality of standard antique coin training directions predetermined for the first antique coin evaluation network branch; Determining the evaluation cycle of the first antique coin evaluation network branch under each of the standard antique coin training directions based on the standard evaluation configuration parameters corresponding to each of the standard antique coin training directions in the evaluation dimension feature domain of the first antique coin evaluation network branch; The standard antique coin training direction corresponding to the shortest evaluation period in the evaluation dimension feature domain of the first antique coin evaluation network branch is selected as the antique coin training direction of the first antique coin evaluation network branch; Among them, the evaluation configuration parameters include sample evaluation configuration parameters and architecture evaluation configuration parameters; the sample evaluation configuration parameters include evaluation granularity and sample division dimension; the architecture evaluation configuration parameters include training phase process segmentation dimension and feature analysis dimension.

3. The method according to claim 2, characterized in that The plurality of standard antique coin training directions include a first standard antique coin training direction; the evaluation configuration parameters include sample evaluation configuration parameters and architecture evaluation configuration parameters, and the sample evaluation configuration parameters include evaluation granularity; Determining an evaluation cycle of the first antique coin evaluation network branch under the first standard antique coin training direction based on the standard evaluation configuration parameters corresponding to the first standard antique coin training direction includes: In a first model training environment, the first antique coin evaluation network branch is standardized into a network entity of the first antique coin evaluation network branch, and the network entity of the first antique coin evaluation network branch is called to process a sample antique coin image to obtain a standard evaluation cycle of the first antique coin evaluation network branch in the first model training environment; Obtaining a first multiplication result between an evaluation granularity in a standard evaluation configuration parameter corresponding to the first standard antique coin training direction and a standard evaluation period of the first antique coin evaluation network branch in the first model training environment; A first division result between the first multiplication result and the architecture evaluation configuration parameter corresponding to the first standard antique coin training direction is used as an evaluation period of the first antique coin evaluation network branch under the first standard antique coin training direction.

4. The method according to claim 2, characterized in that Before determining the evaluation cycle of the first antique coin evaluation network branch under each of the standard antique coin training directions based on the standard evaluation configuration parameters corresponding to each of the standard antique coin training directions in the evaluation dimension feature domain of the first antique coin evaluation network branch, the method further includes: Obtaining a standard processing load of the plurality of antique coin evaluation network branches on the same model training environment; Determining a target antique coin evaluation network branch with the largest standard processing load from among the plurality of antique coin evaluation network branches; If the target antique coin evaluation network branch is not the first antique coin evaluation network branch, then the various standard antique coin training directions in the evaluation dimension feature domain of the first antique coin evaluation network branch are adjusted according to the antique coin training direction of the target antique coin evaluation network branch.

5. The method according to claim 4, characterized in that The plurality of standard antique coin training directions include a first standard antique coin training direction; the evaluation configuration parameters include sample evaluation configuration parameters, and the sample evaluation configuration parameters include evaluation granularity and sample division dimension; Adjusting the first standard antique coin training direction in the evaluation dimension feature domain of the first antique coin evaluation network branch according to the antique coin training direction of the target antique coin evaluation network branch includes: Obtaining a second multiplication result between the evaluation granularity corresponding to the antique coin training direction of the target antique coin evaluation network branch and the sample division dimension; Using the second multiplication result and the second division result between the sample division dimension corresponding to the first standard antique coin training direction, the evaluation granularity corresponding to the first standard antique coin training direction is adjusted; or, The sample division dimension corresponding to the first standard antique coin training direction is adjusted by using the third division result between the second multiplication result and the evaluation granularity corresponding to the first standard antique coin training direction.

6. The method according to claim 1, characterized in that Each of the antique coin evaluation network branches is represented as a first antique coin evaluation network branch, and the first antique coin evaluation network branch corresponds to a first evaluation unit; the antique coin training direction of each of the antique coin evaluation network branches is also used to characterize the evaluation configuration parameters of each of the antique coin evaluation network branches, and the evaluation configuration parameters include sample evaluation configuration parameters and architecture evaluation configuration parameters, and the sample evaluation configuration parameters include sample division dimensions and evaluation granularity; Based on the antique coin training direction of the first antique coin evaluation network branch, configuring the first evaluation unit includes: Determine the number of model training environments in the first evaluation unit based on the sample partitioning dimension and the architecture evaluation configuration parameter in the evaluation configuration parameter corresponding to the antique coin training direction of the first antique coin evaluation network branch; Obtaining the environment configuration in the preset model training environment set, and obtaining the standard evaluation cycle of the first antique coin evaluation network branch in the model training environment of each environment configuration; Determine the evaluation cycle of the first antique coin evaluation network branch in the model training environment of each environment configuration based on the evaluation granularity and architecture evaluation configuration parameters corresponding to the antique coin training direction of the first antique coin evaluation network branch and the standard evaluation cycle of the first antique coin evaluation network branch in the model training environment of each environment configuration; Determine the target environment configuration corresponding to the shortest evaluation cycle based on the evaluation cycle of the first antique coin evaluation network branch in the model training environment of each environment configuration; Acquire the number of model training environments configured in the target environment from the set of model training environments; When the number of model training environments of the target environment configuration is greater than or equal to the number of model training environments in the first evaluation unit, using the target environment configuration as the environment configuration of the model training environment in the first evaluation unit; Based on the antique coin training direction of the first antique coin evaluation network branch and based on the number and environment configuration of the model training environments in the first evaluation unit, the first antique coin evaluation network branch is standardized in the model training environment set to configure the first evaluation unit.

7. The method according to claim 1, characterized in that When executing the training process for the antique coin evaluation model, calling multiple evaluation units in the evaluation training framework to respectively perform the evaluation dimensions of the corresponding antique coin evaluation network branches based on the training processes corresponding to the antique coin training directions of the corresponding antique coin evaluation network branches, including: When executing the training process for the antique coin evaluation model, the training phase of the antique coin evaluation model is divided into a plurality of training steps in the training link, and one training step corresponds to an evaluation dimension of one antique coin evaluation network branch; Execute the training link, and when executing the training link, determine the target evaluation unit used to execute the current training step, call the training process corresponding to the antique coin training direction of the target evaluation unit based on the antique coin evaluation network branch corresponding to the target evaluation unit, and execute the evaluation dimension training process of the antique coin evaluation network branch corresponding to the target evaluation unit.

8. A server system, characterized in that: The method comprises a server, wherein the server is used to execute the method described in any one of claims 1 to 7.

Images