WO2021068529A1 - Image recognition method and apparatus, computer device and storage medium - Google Patents

Abstract

The present invention relates to the field of artificial intelligence model construction, in particular to an image recognition method and apparatus, a computer device and a storage medium. The method comprises: firstly acquiring an image to undergo recognition; and inputting the image to undergo recognition into a preset image recognition big data model to acquire an image recognition result, wherein the preset image recognition big data model is constructed by executing each execution engine unit on the basis of the execution sequence of each execution engine unit according to the data interaction channel of each execution engine unit. According to the present application, the initial image recognition big data model is split into a plurality of execution engine units, and then the execution engine units are executed on the basis of the execution sequence of the execution engine units to integrate and reconstruct the image recognition big data model, so that the model construction efficiency is effectively improved, and the image recognition efficiency is further improved.

Description

图像识别方法、装置、计算机设备及存储介质Image recognition method, device, computer equipment and storage medium

本申请要求于2019年10月12日提交中国专利局、申请号为CN201910968961.4，发明名称为“图像识别方法、装置、计算机设备及存储介质”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。This application claims the priority of a Chinese patent application filed with the Chinese Patent Office on October 12, 2019, the application number is CN201910968961.4, and the invention title is "Image recognition method, device, computer equipment and storage medium". The entire content of the application is approved The reference is incorporated in this application.

技术领域Technical field

本申请涉及计算机技术领域，特别是涉及一种图像识别方法、装置、计算机设备及存储介质。This application relates to the field of computer technology, in particular to an image recognition method, device, computer equipment and storage medium.

背景技术Background technique

图像识别，是指利用计算机对图像进行处理、分析和理解，以识别各种不同模式的目标和对象的技术。目前可以基于大数据模型来对目标图像进行识别。大数据，指无法在一定时间范围内用常规软件工具进行捕捉、管理和处理的数据集合，是需要新处理模式才能具有更强的决策力、洞察发现力和流程优化能力的海量、高增长率和多样化的信息资产。大数据具有海量的数据规模、快速的数据流转、多样的数据类型和价值密度低四大特征。通过大数据模型可以有效对目标图像进行识别。Image recognition refers to the use of computers to process, analyze, and understand images to identify targets and objects in various patterns. At present, the target image can be recognized based on the big data model. Big data refers to a collection of data that cannot be captured, managed, and processed with conventional software tools within a certain time frame. It is a massive and high growth rate that requires new processing modes to have stronger decision-making power, insight and discovery, and process optimization capabilities. And diversified information assets. Big data has four characteristics: massive data scale, fast data flow, diverse data types, and low value density. The big data model can effectively identify the target image.

然而由于大数据模型开发完成后还需要对模型进行集成、部署、上线等工作。发明人意识到大数据模型中的每个子模型的开发环境可能存在多种组合、不同的模型实现过程使得模型集成方式不固定，这些因素导致模型的集成、部署上线过程比较繁琐，部分模型不符合上线要求甚至需要重新开发，影响大数据模型开发的效率，进而影响图像识别的效率。However, after the development of the big data model is completed, the model needs to be integrated, deployed, and online. The inventor realizes that there may be multiple combinations in the development environment of each sub-model in the big data model, and different model implementation processes make the model integration method not fixed. These factors cause the model integration, deployment and online process to be cumbersome, and some models do not conform to The online requirements even need to be re-developed, which affects the efficiency of big data model development, which in turn affects the efficiency of image recognition.

技术问题technical problem

基于此，有必要针对现有大数据模型构建过程效率低，影响图像识别效率的问题，提供一种效率较高的图像识别方法、装置、计算机设备及存储介质。Based on this, it is necessary to provide a more efficient image recognition method, device, computer equipment, and storage medium in response to the low efficiency of the existing big data model construction process and affect the efficiency of image recognition.

技术解决方案Technical solutions

一种图像识别方法，所述方法包括：An image recognition method, the method includes:

获取待识别图像；Obtain the image to be recognized;

将所述待识别图像输入预设图像识别大数据模型，获取图像识别结果；Input the image to be recognized into a preset image recognition big data model to obtain an image recognition result;

所述预设图像识别大数据模型是根据各执行引擎单元的数据交互通道，并基于各执行引擎单元的执行顺序执行各执行引擎单元构建得到，所述各执行引擎单元是根据模型重构集成配置参数对初始图像识别大数据模型拆分重构获得，所述各执行引擎单元的执行顺序由模型重构集成配置参数确定，所述数据交互通道由所述各执行引擎单元间的数据依赖关系构建。The preset image recognition big data model is constructed based on the data interaction channel of each execution engine unit and executed by each execution engine unit based on the execution order of each execution engine unit, and each execution engine unit is constructed according to the model reconstruction integration configuration The parameters are obtained by splitting and reconstructing the initial image recognition big data model, the execution order of each execution engine unit is determined by the model reconstruction integration configuration parameter, and the data interaction channel is constructed by the data dependency relationship between the execution engine units .

一种图像识别装置，所述装置包括：An image recognition device, the device comprising:

数据获取模块，用于获取待识别图像；The data acquisition module is used to acquire the image to be recognized;

图像识别模块，用于将所述待识别图像输入预设图像识别大数据模型，获取图像识别结果；An image recognition module, configured to input the image to be recognized into a preset image recognition big data model to obtain an image recognition result;

所述预设图像识别大数据模型是根据各执行引擎单元的数据交互通道，并基于各执行引擎单元的执行顺序执行各执行引擎单元构建得到，所述各执行引擎单元是根据模型重构集成配置参数对初始图像识别大数据模型拆分重构获得，所述各执行引擎单元的执行顺序由模型重构集成配置参数确定，所述数据交互通道由所述各执行引擎单元间的数据依赖关系构建。The preset image recognition big data model is constructed based on the data interaction channel of each execution engine unit and executed by each execution engine unit based on the execution order of each execution engine unit, and each execution engine unit is constructed according to the model reconstruction integration configuration The parameters are obtained by splitting and reconstructing the initial image recognition big data model, the execution order of each execution engine unit is determined by the model reconstruction integration configuration parameter, and the data interaction channel is constructed by the data dependency relationship between the execution engine units .

一种计算机设备，包括存储器和处理器，所述存储器存储有计算机程序，所述处理器执行所述计算机程序时实现以下步骤：A computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the following steps when the processor executes the computer program:

获取待识别图像；Obtain the image to be recognized;

将所述待识别图像输入预设图像识别大数据模型，获取图像识别结果；Input the image to be recognized into a preset image recognition big data model to obtain an image recognition result;

所述预设图像识别大数据模型是根据各执行引擎单元的数据交互通道，并基于各执行引擎单元的执行顺序执行各执行引擎单元构建得到，所述各执行引擎单元是根据模型重构集成配置参数对初始图像识别大数据模型拆分重构获得，所述各执行引擎单元的执行顺序由模型重构集成配置参数确定，所述数据交互通道由所述各执行引擎单元间的数据依赖关系构建。The preset image recognition big data model is constructed based on the data interaction channel of each execution engine unit and executed by each execution engine unit based on the execution order of each execution engine unit, and each execution engine unit is constructed according to the model reconstruction integration configuration The parameters are obtained by splitting and reconstructing the initial image recognition big data model, the execution order of each execution engine unit is determined by the model reconstruction integration configuration parameter, and the data interaction channel is constructed by the data dependency relationship between the execution engine units .

一种计算机可读存储介质，其上存储有计算机程序，所述计算机程序被处理器执行时实现以下步骤：A computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the following steps are implemented:

获取待识别图像；Obtain the image to be recognized;

将所述待识别图像输入预设图像识别大数据模型，获取图像识别结果；Input the image to be recognized into a preset image recognition big data model to obtain an image recognition result;

所述预设图像识别大数据模型是根据各执行引擎单元的数据交互通道，并基于各执行引擎单元的执行顺序执行各执行引擎单元构建得到，所述各执行引擎单元是根据模型重构集成配置参数对初始图像识别大数据模型拆分重构获得，所述各执行引擎单元的执行顺序由模型重构集成配置参数确定，所述数据交互通道由所述各执行引擎单元间的数据依赖关系构建。The preset image recognition big data model is constructed based on the data interaction channel of each execution engine unit and executed by each execution engine unit based on the execution order of each execution engine unit, and each execution engine unit is constructed according to the model reconstruction integration configuration The parameters are obtained by splitting and reconstructing the initial image recognition big data model, the execution order of each execution engine unit is determined by the model reconstruction integration configuration parameter, and the data interaction channel is constructed by the data dependency relationship between the execution engine units .

有益效果Beneficial effect

上述图像识别方法、装置、计算机设备以及存储介质，首先获取待识别图像；将待识别图像输入预设图像识别大数据模型，获取图像识别结果；预设图像识别大数据模型是根据各执行引擎单元的数据交互通道，并基于各执行引擎单元的执行顺序执行各执行引擎单元构建得到，各执行引擎单元是根据模型重构集成配置参数对初始图像识别大数据模型拆分重构获得，各执行引擎单元的执行顺序由模型重构集成配置参数确定，数据交互通道由各执行引擎单元间的数据依赖关系构建。本申请的图像识别方法，通过将初始图像识别大数据模型拆分为多个执行引擎单元，而后基于执行引擎单元的执行顺序执行各执行引擎单元来集成重构图像识别大数据模型，从而有效提高模型构建的效率，进而提高图像识别的效率。The above image recognition method, device, computer equipment and storage medium first obtain the image to be recognized; input the image to be recognized into the preset image recognition big data model to obtain the image recognition result; the preset image recognition big data model is based on each execution engine unit The data interaction channel of each execution engine unit is constructed by executing each execution engine unit based on the execution order of each execution engine unit. Each execution engine unit is obtained by splitting and reconstructing the initial image recognition big data model according to the model reconstruction integration configuration parameters. Each execution engine The execution sequence of the units is determined by the model reconstruction integration configuration parameters, and the data interaction channel is constructed by the data dependency relationship between the execution engine units. The image recognition method of the present application integrates the reconstructed image recognition big data model by splitting the initial image recognition big data model into multiple execution engine units, and then executing each execution engine unit based on the execution order of the execution engine units to integrate the reconstructed image recognition big data model. The efficiency of model construction, in turn, improves the efficiency of image recognition.

附图说明Description of the drawings

图1为一个实施例中图像识别方法的应用环境图；Fig. 1 is an application environment diagram of an image recognition method in an embodiment;

图2为一个实施例中图像识别方法的流程示意图；Fig. 2 is a schematic flowchart of an image recognition method in an embodiment;

图3为另一个实施例中图像识别方法的流程示意图；Fig. 3 is a schematic flowchart of an image recognition method in another embodiment;

图4为一个实施例中图3中步骤S340的子流程示意图；FIG. 4 is a schematic diagram of a sub-flow of step S340 in FIG. 3 in an embodiment;

图5为一个实施例中图3中步骤S380的子流程示意图；FIG. 5 is a schematic diagram of a sub-flow of step S380 in FIG. 3 in an embodiment;

图6为一个实施例中图像识别装置的结构框图；Fig. 6 is a structural block diagram of an image recognition device in an embodiment;

图7为一个实施例中计算机设备的内部结构图。Fig. 7 is an internal structure diagram of a computer device in an embodiment.

本发明的最佳实施方式The best mode of the present invention

为了使本申请的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本申请进行进一步详细说明。应当理解，此处描述的具体实施例仅仅用以解释本申请，并不用于限定本申请。In order to make the purpose, technical solutions, and advantages of this application clearer, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present application, and are not used to limit the present application.

本申请提供的图像识别方法，可以应用于如图1所示的应用环境中，其中，图像识别服务器102可以通过网络的方式与终端104进行通信，终端104可以向图像识别服务器102发送待识别图像，图像识别服务器102将获得的待识别图像输入预设图像识别大数据模型，获取对应图像识别结果。其中预设图像识别大数据模型是根据各执行引擎单元的数据交互通道，并基于各执行引擎单元的执行顺序执行各执行引擎单元构建得到，各执行引擎单元是根据模型重构集成配置参数对初始图像识别大数据模型拆分重构获得，各执行引擎单元的执行顺序由模型重构集成配置参数确定，数据交互通道由各执行引擎单元间的数据依赖关系构建。终端102可以但不限于是各种个人计算机、笔记本电脑、智能手机、平板电脑和便携式可穿戴设备，图像识别服务器104可以用独立的服务器或者是多个服务器组成的服务器集群来实现。The image recognition method provided in this application can be applied to the application environment as shown in FIG. 1, where the image recognition server 102 can communicate with the terminal 104 via a network, and the terminal 104 can send the image to be recognized to the image recognition server 102 The image recognition server 102 inputs the obtained image to be recognized into the preset image recognition big data model, and obtains the corresponding image recognition result. The preset image recognition big data model is constructed based on the data interaction channel of each execution engine unit, and executes each execution engine unit based on the execution order of each execution engine unit. Each execution engine unit is constructed according to the model reconstruction integration configuration parameter to the initial The image recognition big data model is split and reconstructed, the execution order of each execution engine unit is determined by the model reconstruction integration configuration parameters, and the data interaction channel is constructed by the data dependency relationship between the execution engine units. The terminal 102 may be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, and portable wearable devices. The image recognition server 104 may be implemented by an independent server or a server cluster composed of multiple servers.

本发明的实施方式Embodiments of the present invention

如图2所示，在其中一个实施例中，本申请的图像识别方法，通过图像识别服务器实现，具体包括以下步骤：As shown in Figure 2, in one of the embodiments, the image recognition method of the present application is implemented by an image recognition server, which specifically includes the following steps:

S200，图像识别服务器获取待识别图像。S200: The image recognition server obtains an image to be recognized.

S400，将待识别图像输入预设图像识别大数据模型，获取图像识别结果。S400: Input the image to be recognized into a preset image recognition big data model, and obtain an image recognition result.

其中预设图像识别大数据模型是根据各执行引擎单元的数据交互通道，并基于各执行引擎单元的执行顺序执行各执行引擎单元构建得到，各执行引擎单元是根据模型重构集成配置参数对初始图像识别大数据模型拆分重构获得，各执行引擎单元的执行顺序由模型重构集成配置参数确定，数据交互通道由各执行引擎单元间的数据依赖关系构建。The preset image recognition big data model is constructed based on the data interaction channel of each execution engine unit, and executes each execution engine unit based on the execution order of each execution engine unit. Each execution engine unit is constructed according to the model reconstruction integration configuration parameter to the initial The image recognition big data model is split and reconstructed, the execution order of each execution engine unit is determined by the model reconstruction integration configuration parameters, and the data interaction channel is constructed by the data dependency relationship between the execution engine units.

模型重构集成配置参数是指模型构建过程中工作人员输入的模型构造参数，其中执行引擎单元是指模型运行的最小部分，其通过将模型拆分为多个部分后，分别配置不同部分对应的执行环境后获得。重构集成配置参数是指符合上线要求对应的集成方式，重构集成配置参数根据预设图像识别大数据模型对应的上线需求来确定，由开发者输入。数据依赖关系具体包括了不同执行引擎单元间的参数依赖关系，确定各个执行引擎单元在新的大数据模型中的执行顺序之后，可以获取各个相连接的执行引擎单元间的数据依赖关系，基于数据依赖关系构建各个相连接的执行引擎单元之间的数据通道。可以将原初始图像识别大数据模型拆分为各个不同的执行引擎单元后，根据开发者输入的重构集成配置参数对模型进行重构，确定各执行引擎单元的执行顺序，并根据执行引擎间参数的交互关系来构建不同执行引擎单元间的数据交互通道，来获得入预设图像识别大数据模型，完成模型构建的工作。Model reconstruction integration configuration parameters refer to the model construction parameters input by the staff during the model building process. The execution engine unit refers to the smallest part of the model operation. After the model is split into multiple parts, the corresponding parts of different parts are configured respectively. Obtained after the execution environment. The reconfiguration integration configuration parameters refer to the integration methods that meet the online requirements. The reconfiguration integration configuration parameters are determined according to the online requirements corresponding to the preset image recognition big data model, and are input by the developer. The data dependence relationship specifically includes the parameter dependence relationship between different execution engine units. After determining the execution order of each execution engine unit in the new big data model, the data dependence relationship between each connected execution engine unit can be obtained, based on the data The dependency relationship constructs a data channel between each connected execution engine unit. After splitting the original initial image recognition big data model into different execution engine units, the model is reconstructed according to the reconstruction integration configuration parameters input by the developer, the execution order of each execution engine unit is determined, and the execution engine The interaction relationship of the parameters is used to construct the data interaction channel between different execution engine units, to obtain the preset image recognition big data model, and complete the work of model construction.

如图3所示，在其中一个实施例中，在步骤S400之前，预设图像识别大数据模型的构建过程具体包括：As shown in FIG. 3, in one of the embodiments, before step S400, the construction process of the preset image recognition big data model specifically includes:

S320，获取初始图像识别大数据模型和模型重构集成配置参数，将初始图像识别大数据模型拆分为最小功能单元。S320: Obtain the initial image recognition big data model and model reconstruction integration configuration parameters, and split the initial image recognition big data model into minimum functional units.

初始图像识别大数据模型是指已经开发完成的大数据模型，模型初始开始时，在定义好模型对接标准后，多位人员同步开发子模块，而后将子模块集成为初始图像识别大数据模型。但是初始图像识别大数据模型中各个子模块的开发环境存在多种组合，不同的模型实现过程使得模型集成方式不固定，这些因素导致模型的集成部署较为繁琐，不符合上线要求。此时可以将该初始图像识别大数据模型拆分开来，通过重构的方式将初始图像识别大数据模型重组为预设图像识别大数据模型，以达到符合上线要求的目的。重构集成配置参数是指符合上线要求对应的集成方式对应的构造参数，重构集成配置参数根据预设图像识别大数据模型对应的上线需求来确定，由开发者输入。最小功能单元是指将模型集成过程中的计算部分抽象出最小的计算单元。初始图像识别大数据模型为通过多种开发环境下的代码实现的对应的执行程序。将初始图像识别大数据模型拆分为各最小功能单元的步骤具体可以通过拆分初始图像识别大数据模型对应的代码来实现。The initial image recognition big data model refers to the big data model that has been developed. At the beginning of the model, after the model docking standard is defined, multiple people develop sub-modules simultaneously, and then integrate the sub-modules into the initial image recognition big data model. However, there are many combinations of the development environment of each sub-module in the initial image recognition big data model. Different model implementation processes make the model integration mode not fixed. These factors cause the integration and deployment of the model to be cumbersome and do not meet the online requirements. At this point, the initial image recognition big data model can be separated, and the initial image recognition big data model can be reorganized into a preset image recognition big data model through reconstruction to achieve the goal of meeting the online requirements. The reconfiguration integration configuration parameters refer to the configuration parameters corresponding to the integration method corresponding to the online requirements. The reconfiguration integration configuration parameters are determined according to the online requirements corresponding to the preset image recognition big data model, and are input by the developer. The smallest functional unit refers to the smallest computing unit that abstracts the computing part of the model integration process. The initial image recognition big data model is a corresponding execution program realized by codes in a variety of development environments. The step of splitting the initial image recognition big data model into the smallest functional units can be specifically implemented by splitting the code corresponding to the initial image recognition big data model.

S340，获取最小功能单元对应的执行过程，根据最小功能单元以及与最小功能单元对应的执行过程，获取执行引擎单元。S340: Obtain an execution process corresponding to the minimum functional unit, and obtain the execution engine unit according to the minimum functional unit and the execution process corresponding to the minimum functional unit.

不同的最小功能单元对应的开发语言环境不同，可以将不同的开发环境下执行代码的过程抽象成标准的执行引擎单元。即将执行各最小功能单元代码的过程抽象为执行引擎单元。Different minimum functional units correspond to different development language environments, and the process of executing code under different development environments can be abstracted into standard execution engine units. The process of executing the code of each minimum functional unit is abstracted as an execution engine unit.

S360，根据模型重构集成配置参数确定执行引擎单元的单元执行顺序。S360: Determine the unit execution sequence of the execution engine unit according to the model reconstruction integration configuration parameter.

模型重构集成配置参数是指符合上线要求对应的输入参数，通过模型重构集成配置参数可以确定重构后各个执行引擎在新的预设图像识别大数据模型中的单元执行顺序。Model reconstruction integrated configuration parameters refer to input parameters that meet the online requirements. The model reconstruction integrated configuration parameters can determine the execution order of units in the new preset image recognition big data model for each execution engine after reconstruction.

S380，获取执行引擎单元间的数据依赖关系，根据数据依赖关系构建执行引擎单元间的数据交互通道，基于数据交互通道、执行顺序以及执行引擎单元构建预设图像识别大数据模型。S380: Obtain the data dependency relationship between the execution engine units, construct a data interaction channel between the execution engine units according to the data dependency relationship, and build a preset image recognition big data model based on the data interaction channel, the execution sequence, and the execution engine unit.

数据依赖关系具体包括了不同执行引擎单元间的参数依赖关系，确定各个执行引擎单元在新的大数据模型中的执行顺序之后，可以获取各个相连接的执行引擎单元间的数据依赖关系，基于数据依赖关系构建各个相连接的执行引擎单元之间的数据通道，并基于数据交互通道、执行顺序以及执行引擎单元构建预设图像识别大数据模型。在其中一个实施例中，可以通过数据库和json方式来构建执行引擎单元间的数据交互通道，完成不同执行引擎单元间的数据交互。The data dependence relationship specifically includes the parameter dependence relationship between different execution engine units. After determining the execution order of each execution engine unit in the new big data model, the data dependence relationship between each connected execution engine unit can be obtained, based on the data The dependency relationship constructs a data channel between each connected execution engine unit, and builds a preset image recognition big data model based on the data interaction channel, execution sequence, and execution engine unit. In one of the embodiments, a data exchange channel between execution engine units can be constructed through a database and json to complete data exchange between different execution engine units.

如图4所示，在其中一个实施例中，步骤S340包括：As shown in FIG. 4, in one of the embodiments, step S340 includes:

S341，识别最小功能单元对应的开发信息数据。S341: Identify the development information data corresponding to the smallest functional unit.

S343，根据最小功能单元的开发信息数据，确定最小功能单元对应执行配置信息。S343: Determine the execution configuration information corresponding to the smallest functional unit according to the development information data of the smallest functional unit.

S345，根据最小功能单元以及执行配置信息，获取执行引擎单元。S345: Acquire the execution engine unit according to the minimum functional unit and the execution configuration information.

服务器可以首先识别最小功能单元对应的开发信息数据，在其中一个实施例中，开发信息数据具体包括了该最小功能单元对应的构成语言以及开发环境等信息。在识别开发信息数据后，可以基于该开发信息数据，获取完成的执行配置信息。执行配置信息包括了执行配置信息具体包括了最小功能单元对应的，执行环境（Python、R、SQL等开发语言）、环境参数配置、执行标准代码（对于代码的中参数用统一的命名格式加以标识）、标准代码中的参数配置，通过执行配置信息来对当前环境下执行引擎单元进行配置，确保当前环境可以无障碍地运行所有拆分后重构获得的执行引擎单元，提高模型重构的可用性。The server may first identify the development information data corresponding to the smallest functional unit. In one embodiment, the development information data specifically includes information such as the constituent language and development environment corresponding to the smallest functional unit. After the development information data is identified, the completed execution configuration information can be obtained based on the development information data. The execution configuration information includes the execution configuration information, which specifically includes the execution environment (Python, R, SQL and other development languages), environment parameter configuration, and execution standard code corresponding to the smallest functional unit (the middle parameters of the code are identified by a unified naming format ), the parameter configuration in the standard code, through the execution configuration information to configure the execution engine unit in the current environment, to ensure that the current environment can run all the execution engine units obtained after the split reconstruction and improve the usability of the model reconstruction .

在其中一个实施例中，步骤S320包括：In one of the embodiments, step S320 includes:

获取初始图像识别大数据模型和模型重构集成配置参数，识别初始图像识别大数据模型中各功能组成部分对应的开发信息数据；Obtain the initial image recognition big data model and model reconstruction integration configuration parameters, and identify the development information data corresponding to each functional component in the initial image recognition big data model;

根据初始图像识别大数据模型各功能组成部分对应的开发信息数据，将初始图像识别大数据模型拆分为各最小功能单元。According to the development information data corresponding to each functional component of the initial image recognition big data model, the initial image recognition big data model is divided into the smallest functional units.

开发信息数据具体包括了该最小功能单元对应的构成语言以及开发环境等信息。由于模型代码的不同部分由不同的开发语言来开发，可以基于初始图像识别大数据模型的程序代码中各部分的开发语言与开发环境的不同，将初始图像识别大数据模型拆分为一段一段的单元，这些单元即为各最小功能单元。如一个待预设图像识别大数据模型的构成语言包括了A、B和C，而获得的初始图像识别大数据模型对应的程序代码按照顺序下来是ACBABAC，这其中每个字母代表了它的构成语言在，则这个模型可以拆分为A1、C1、B1、A2、B2、A3以及C2这七个最小功能单元。The development information data specifically includes information such as the constituent language and development environment corresponding to the smallest functional unit. Since different parts of the model code are developed by different development languages, the initial image recognition big data model can be divided into segments based on the differences in the development language and development environment of each part of the program code of the initial image recognition big data model. Units, these units are the smallest functional units. For example, the composition language of an image recognition big data model to be preset includes A, B, and C, and the program code corresponding to the obtained initial image recognition big data model is ACBABAC in order, where each letter represents its composition If the language is present, this model can be divided into seven minimum functional units: A1, C1, B1, A2, B2, A3, and C2.

在其中一个实施例中，根据初始图像识别大数据模型各功能组成部分对应的开发信息数据，将初始图像识别大数据模型拆分为各最小功能单元包括：In one of the embodiments, splitting the initial image recognition big data model into the smallest functional units according to the development information data corresponding to each functional component of the initial image recognition big data model includes:

获取初始图像识别大数据模型；Obtain the initial image recognition big data model;

根据预设标识符识别初始图像识别大数据模型内的参数数据，在参数数据前添加对应特殊符号，将初始图像识别大数据模型内的参数数据转化为参数值数据；Identify the parameter data in the big data model of the initial image recognition according to the preset identifier, add corresponding special symbols in front of the parameter data, and convert the parameter data in the big data model of the initial image recognition to parameter value data;

执行预设拆分脚本，根据初始图像识别大数据模型对应的开发信息数据，将初始图像识别大数据模型拆分为最小组成单元；Execute the preset split script, and split the initial image recognition big data model into the smallest component units according to the development information data corresponding to the initial image recognition big data model;

删除参数值数据前的特征符号，将最小组成单元内的参数值数据转化为参数数据，获得最小功能单元。Delete the characteristic symbols before the parameter value data, and convert the parameter value data in the smallest component unit into parameter data to obtain the smallest functional unit.

可以通过预设拆分脚本将初始图像识别大数据模型拆分为最小功能单元，为了执行脚本可以现将初始图像识别大数据模型内的参数数据替换为参数值数据，而后在执行完拆分之后，将参数值数据替换回参数数据，获得最小功能单元。可以通过初始图像识别大数据模型内的预设标识符，先识别大数据模型内的参数数据，在参数数据前添加对应特殊符号来实现将参数数据转化为参数值数据，而后拆分完成之后再将其转化会参数数据。在其中一个实施例中，可以在参数数据前面加上一个美元符号，将其替换为参数值数据。通过脚本，可以迅速完成模型拆分的工作，提高模型构建效率。The initial image recognition big data model can be split into the smallest functional units through a preset split script. In order to execute the script, the parameter data in the initial image recognition big data model can be replaced with parameter value data, and then after the split is performed , Replace the parameter value data back to the parameter data to obtain the smallest functional unit. The preset identifiers in the big data model can be identified through the initial image. The parameter data in the big data model can be identified first, and the corresponding special symbols can be added in front of the parameter data to realize the conversion of parameter data into parameter value data, and then after the split is completed Convert it into parameter data. In one of the embodiments, a dollar sign can be added in front of the parameter data to replace it with the parameter value data. Through scripts, you can quickly complete the work of model splitting and improve the efficiency of model construction.

如图5所示，在其中一个实施例中，S380具体包括：As shown in Figure 5, in one of the embodiments, S380 specifically includes:

S381，获取执行引擎单元间的数据依赖关系，根据数据依赖关系构建执行引擎单元间的数据交互通道。S381: Obtain the data dependency relationship between the execution engine units, and construct a data interaction channel between the execution engine units according to the data dependency relationship.

S383，基于数据交互通道以及执行顺序构建模型运行网络，并根据单元执行顺序定位各执行引擎单元在模型运行网络的执行位置。S383: Construct a model running network based on the data interaction channel and the execution sequence, and locate the execution position of each execution engine unit in the model running network according to the unit execution sequence.

S385，将执行引擎单元添加于模型运行网络对应执行位置，获得预设图像识别大数据模型。S385: Add the execution engine unit to the corresponding execution position of the model running network to obtain a preset image recognition big data model.

模型重构需要构建执行引擎单元间的数据交互通道，来实现模块之间的数据互通。由于多个模块之间所使用的参数之类需要通过互通交流才能完成模型的计算任务，所以需要通过构建模块之间的数据通道，并通过多个执行引擎单元间的串联并联关系以及数据交互通道来进一步构建新模型的模型运行网络，同时根据单元执行顺序定位各执行引擎单元在模型运行网络的执行位置，来实现模块之间的数据交流，达成数据互通的效果，这一步骤具体可以通过数据库或者通过json这种轻量级的交换格式来完成不同模块间的数据互通。而后将执行引擎单元设置于模型运行网络对应执行位置，获得预设图像识别大数据模型，达成模型重构的任务。在执行模型运算的任务时，初始的执行引擎单元获取用户输入数据，而后按照执行顺序将当前单元获得的执行结果依次通过数据交互通道输入到下一级的执行引擎单元中，依次执行，直到得出最后的结果。Model reconstruction needs to build a data exchange channel between execution engine units to realize data exchange between modules. Because the parameters used between multiple modules need to communicate with each other to complete the calculation tasks of the model, it is necessary to build the data channel between the modules, and through the series-parallel relationship between the multiple execution engine units and the data exchange channel To further build the model operation network of the new model, and at the same time locate the execution position of each execution engine unit in the model operation network according to the unit execution order, to realize the data exchange between the modules and achieve the effect of data intercommunication. This step can be specifically through the database Or through the lightweight exchange format of json to complete the data intercommunication between different modules. Then, the execution engine unit is set at the corresponding execution position of the model running network, and the preset image recognition big data model is obtained, and the task of model reconstruction is achieved. When performing model calculation tasks, the initial execution engine unit obtains user input data, and then according to the execution order, the execution results obtained by the current unit are sequentially input to the execution engine unit at the next level through the data interaction channel, and executed in sequence until the result is obtained. The final result.

在其中一个实施例中，S400之前还包括：In one of the embodiments, before S400, it further includes:

获取校验数据以及校验数据对应的中间结果数据以及校验输出结果数据。Obtain the verification data, the intermediate result data corresponding to the verification data, and the verification output result data.

将预设校验数据输入预设图像识别大数据模型，获取预设校验数据对应的各个执行引擎单元的输出数据以及预设图像识别大数据模型输出的校验结果数据。The preset verification data is input into the preset image recognition big data model, and the output data of each execution engine unit corresponding to the preset verification data and the verification result data output by the preset image recognition big data model are obtained.

对比预设输出结果数据与校验结果数据，获得验证结果，当验证结果为验证不通过时，对比预设中间结果数据与各个执行引擎单元的输出数据查找执行引擎单元中的问题单元，根据验证结果修正问题单元，返回将预设校验数据输入预设图像识别大数据模型的步骤。Compare the preset output result data with the verification result data to obtain the verification result. When the verification result is that the verification fails, compare the preset intermediate result data with the output data of each execution engine unit to find the problem unit in the execution engine unit, and according to the verification As a result, the problem unit is corrected, and the step of inputting the preset verification data into the preset image recognition big data model is returned.

在生成预设图像识别大数据模型，还可以通过预设的校验数据来对模型效果进行验证，具体可以先查找模型对应的校验数据与校验输出结果数据，通过将校验数据输入新生成的预设图像识别大数据模型，获取结果来对模型进行校验，当模型输出的结果与预估的校验输出结果数据不一致时，通过对比各模块的中间结果数据来查找出问题的执行引擎单元，并根据验证结果修正问题单元，重新返回将预设校验数据输入预设图像识别大数据模型的步骤，再对修正后的模型进行验证。当验证结果为验证通过时，可以直接将待识别图像输入该预设图像识别大数据模型，来获取对应的图像识别结果。When generating a preset image to recognize a big data model, you can also verify the effect of the model through the preset verification data. Specifically, you can first find the verification data corresponding to the model and the verification output result data, and input the verification data into the new model. The generated preset image recognizes the big data model and obtains the result to verify the model. When the model output result is inconsistent with the estimated verification output result data, compare the intermediate result data of each module to find the problematic execution The engine unit, and corrects the problem unit according to the verification result, returns to the step of inputting the preset verification data into the preset image recognition big data model, and then verifies the corrected model. When the verification result is that the verification is passed, the image to be recognized can be directly input into the preset image recognition big data model to obtain the corresponding image recognition result.

应该理解的是，虽然图2-5的流程图中的各个步骤按照箭头的指示依次显示，但是这些步骤并不是必然按照箭头指示的顺序依次执行。除非本文中有明确的说明，这些步骤的执行并没有严格的顺序限制，这些步骤可以以其它的顺序执行。而且，图2-5中的至少一部分步骤可以包括多个子步骤或者多个阶段，这些子步骤或者阶段并不必然是在同一时刻执行完成，而是可以在不同的时刻执行，这些子步骤或者阶段的执行顺序也不必然是依次进行，而是可以与其它步骤或者其它步骤的子步骤或者阶段的至少一部分轮流或者交替地执行。It should be understood that although the various steps in the flowcharts of FIGS. 2-5 are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless there is a clear description in this article, there is no strict order for the execution of these steps, and these steps can be executed in other orders. Moreover, at least part of the steps in Figures 2-5 may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. These sub-steps or stages The execution order of is not necessarily performed sequentially, but may be performed alternately or alternately with at least a part of other steps or sub-steps or stages of other steps.

如图6所示，一种图像识别装置，装置包括：As shown in Fig. 6, an image recognition device includes:

数据获取模块200，用于获取待识别图像；The data acquisition module 200 is used to acquire the image to be recognized;

图像识别模块400，用于将待识别图像输入预设图像识别大数据模型，获取图像识别结果；The image recognition module 400 is used to input the image to be recognized into the preset image recognition big data model to obtain the image recognition result;

预设图像识别大数据模型是根据各执行引擎单元的数据交互通道，并基于各执行引擎单元的执行顺序执行各执行引擎单元构建得到，各执行引擎单元是根据模型重构集成配置参数对初始图像识别大数据模型拆分重构获得，各执行引擎单元的执行顺序由模型重构集成配置参数确定，数据交互通道由各执行引擎单元间的数据依赖关系构建。The preset image recognition big data model is constructed based on the data interaction channel of each execution engine unit, and executes each execution engine unit based on the execution order of each execution engine unit. Each execution engine unit is constructed according to the model reconstruction integration configuration parameter to the initial image The recognition big data model is obtained by splitting and reconstructing, the execution order of each execution engine unit is determined by the model reconstruction integration configuration parameters, and the data interaction channel is constructed by the data dependency relationship between each execution engine unit.

在其中一个实施例中，还包括模型重构模块，用于获取初始图像识别大数据模型和模型重构集成配置参数，将初始图像识别大数据模型拆分为最小功能单元；In one of the embodiments, it further includes a model reconstruction module, which is used to obtain the initial image recognition big data model and model reconstruction integration configuration parameters, and split the initial image recognition big data model into minimum functional units;

获取最小功能单元对应的执行过程，根据最小功能单元以及与最小功能单元对应的执行过程，获取执行引擎单元；Obtain the execution process corresponding to the smallest functional unit, and obtain the execution engine unit according to the smallest functional unit and the execution process corresponding to the smallest functional unit;

根据模型重构集成配置参数确定执行引擎单元的单元执行顺序；Determine the unit execution order of the execution engine unit according to the model reconstruction integration configuration parameters;

获取执行引擎单元间的数据依赖关系，根据数据依赖关系构建执行引擎单元间的数据交互通道，基于数据交互通道、执行顺序以及执行引擎单元构建预设图像识别大数据模型。Obtain the data dependency relationship between the execution engine units, construct a data interaction channel between the execution engine units according to the data dependency relationship, and build a preset image recognition big data model based on the data interaction channel, the execution sequence and the execution engine unit.

在其中一个实施例中，模型重构模块具体用于识别最小功能单元对应的开发信息数据；根据最小功能单元的开发信息数据，确定最小功能单元对应执行配置信息；根据最小功能单元以及执行配置信息，获取执行引擎单元。In one of the embodiments, the model reconstruction module is specifically used to identify the development information data corresponding to the smallest functional unit; determine the execution configuration information corresponding to the smallest functional unit according to the development information data of the smallest functional unit; according to the smallest functional unit and the execution configuration information , Get the execution engine unit.

在其中一个实施例中，模型重构模块具体用于获取初始图像识别大数据模型和模型重构集成配置参数，识别初始图像识别大数据模型中各功能组成部分对应的开发信息数据；根据初始图像识别大数据模型各功能组成部分对应的开发信息数据，将初始图像识别大数据模型拆分为各最小功能单元。In one of the embodiments, the model reconstruction module is specifically used to obtain the initial image recognition big data model and model reconstruction integration configuration parameters, and identify the development information data corresponding to each functional component in the initial image recognition big data model; Identify the development information data corresponding to each functional component of the big data model, and split the initial image recognition big data model into the smallest functional units.

在其中一个实施例中，模型重构模块具体用于：获取初始图像识别大数据模型；根据预设标识符识别初始图像识别大数据模型内的参数数据，在参数数据前添加对应特殊符号，将初始图像识别大数据模型内的参数数据转化为参数值数据；执行预设拆分脚本，根据初始图像识别大数据模型对应的开发信息数据，将初始图像识别大数据模型拆分为最小组成单元；删除参数值数据前的特征符号，将最小组成单元内的参数值数据转化为参数数据，获得最小功能单元。In one of the embodiments, the model reconstruction module is specifically used to: obtain the initial image recognition big data model; recognize the parameter data in the initial image recognition big data model according to the preset identifier, add corresponding special symbols before the parameter data, and change The parameter data in the initial image recognition big data model is converted into parameter value data; the preset split script is executed, and the initial image recognition big data model is split into the smallest component units according to the development information data corresponding to the initial image recognition big data model; Delete the characteristic symbols before the parameter value data, and convert the parameter value data in the smallest component unit into parameter data to obtain the smallest functional unit.

在其中一个实施例中，模型重构模块具体用于获取执行引擎单元间的数据依赖关系，根据数据依赖关系构建执行引擎单元间的数据交互通道；基于数据交互通道以及执行顺序构建模型运行网络，并根据单元执行顺序定位各执行引擎单元在模型运行网络的执行位置；将执行引擎单元添加于模型运行网络对应执行位置，获得预设图像识别大数据模型。In one of the embodiments, the model reconstruction module is specifically used to obtain the data dependency relationship between the execution engine units, build a data interaction channel between the execution engine units according to the data dependency relationship; build a model running network based on the data interaction channel and the execution sequence, And locate the execution position of each execution engine unit in the model running network according to the unit execution sequence; add the execution engine unit to the corresponding execution position of the model running network to obtain the preset image recognition big data model.

在其中一个实施例中，模型重构模块具体用于获取校验数据以及校验数据对应的中间结果数据以及校验输出结果数据；将预设校验数据输入预设图像识别大数据模型，获取预设校验数据对应的各个执行引擎单元的输出数据以及预设图像识别大数据模型输出的校验结果数据；对比预设输出结果数据与校验结果数据，获得验证结果，当验证结果为验证不通过时，对比预设中间结果数据与各个执行引擎单元的输出数据查找执行引擎单元中的问题单元，根据验证结果修正问题单元，返回将预设校验数据输入预设图像识别大数据模型的步骤。In one of the embodiments, the model reconstruction module is specifically used to obtain the verification data, the intermediate result data corresponding to the verification data, and the verification output result data; the preset verification data is input into the preset image recognition big data model to obtain The output data of each execution engine unit corresponding to the preset verification data and the verification result data output by the preset image recognition big data model; compare the preset output result data with the verification result data to obtain the verification result, when the verification result is verification When it fails, compare the preset intermediate result data with the output data of each execution engine unit to find the problem unit in the execution engine unit, correct the problem unit according to the verification result, and return the preset verification data to the preset image recognition big data model step.

关于图像识别装置的具体限定可以参见上文中对于图像识别方法的限定，在此不再赘述。上述图像识别装置中的各个模块可全部或部分通过软件、硬件及其组合来实现。上述各模块可以硬件形式内嵌于或独立于计算机设备中的处理器中，也可以以软件形式存储于计算机设备中的存储器中，以便于处理器调用执行以上各个模块对应的操作。For the specific definition of the image recognition device, please refer to the above definition of the image recognition method, which will not be repeated here. Each module in the above-mentioned image recognition device can be implemented in whole or in part by software, hardware, and a combination thereof. The above-mentioned modules may be embedded in the form of hardware or independent of the processor in the computer equipment, or may be stored in the memory of the computer equipment in the form of software, so that the processor can call and execute the operations corresponding to the above-mentioned modules.

在一个实施例中，提供了一种计算机设备，该计算机设备可以是服务器，其内部结构图可以如图7所示。该计算机设备包括通过***总线连接的处理器、存储器、网络接口以及数据库。其中，该计算机设备的处理器用于提供计算和控制能力。该计算机设备的存储器包括非易失性存储介质、内存储器。该非易失性存储介质存储有操作***和计算机程序。该内存储器为非易失性存储介质中的操作***和计算机程序的运行提供环境。该计算机设备的网络接口用于与外部的终端通过网络连接通信。该计算机设备的数据库用于存储知识图谱相关数据。该计算机程序被处理器执行时以实现一种图像识别方法。In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. 7. The computer equipment includes a processor, a memory, a network interface, and a database connected through a system bus. Among them, the processor of the computer device is used to provide calculation and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. The database of the computer equipment is used to store data related to the knowledge graph. The computer program is executed by the processor to realize an image recognition method.

本领域技术人员可以理解，图7中示出的结构，仅仅是与本申请方案相关的部分结构的框图，并不构成对本申请方案所应用于其上的计算机设备的限定，具体的计算机设备可以包括比图中所示更多或更少的部件，或者组合某些部件，或者具有不同的部件布置。Those skilled in the art can understand that the structure shown in FIG. 7 is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may Including more or fewer parts than shown in the figure, or combining some parts, or having a different arrangement of parts.

在一个实施例中，提供了一种计算机设备，包括存储器和处理器，存储器中存储有计算机程序，该处理器执行计算机程序时实现以下步骤：In one embodiment, a computer device is provided, including a memory and a processor, and a computer program is stored in the memory, and the processor implements the following steps when the processor executes the computer program:

获取待识别图像；Obtain the image to be recognized;

将待识别图像输入预设图像识别大数据模型，获取图像识别结果；Input the image to be recognized into the preset image recognition big data model to obtain the image recognition result;

预设图像识别大数据模型是根据各执行引擎单元的数据交互通道，并基于各执行引擎单元的执行顺序执行各执行引擎单元构建得到，各执行引擎单元是根据模型重构集成配置参数对初始图像识别大数据模型拆分重构获得，各执行引擎单元的执行顺序由模型重构集成配置参数确定，数据交互通道由各执行引擎单元间的数据依赖关系构建。The preset image recognition big data model is constructed based on the data interaction channel of each execution engine unit, and executes each execution engine unit based on the execution order of each execution engine unit. Each execution engine unit is constructed according to the model reconstruction integration configuration parameter to the initial image The recognition big data model is obtained by splitting and reconstructing, the execution order of each execution engine unit is determined by the model reconstruction integration configuration parameters, and the data interaction channel is constructed by the data dependency relationship between each execution engine unit.

在一个实施例中，处理器执行计算机程序时还实现以下步骤：获取初始图像识别大数据模型和模型重构集成配置参数，将初始图像识别大数据模型拆分为最小功能单元；获取最小功能单元对应的执行过程，根据最小功能单元以及与最小功能单元对应的执行过程，获取执行引擎单元；根据模型重构集成配置参数确定执行引擎单元的单元执行顺序；获取执行引擎单元间的数据依赖关系，根据数据依赖关系构建执行引擎单元间的数据交互通道，基于数据交互通道、执行顺序以及执行引擎单元构建预设图像识别大数据模型。In one embodiment, the processor further implements the following steps when executing the computer program: acquiring the initial image recognition big data model and model reconstruction integration configuration parameters, splitting the initial image recognition big data model into the smallest functional units; acquiring the smallest functional unit According to the corresponding execution process, obtain the execution engine unit according to the minimum functional unit and the execution process corresponding to the minimum functional unit; determine the unit execution order of the execution engine unit according to the model reconstruction integration configuration parameter; obtain the data dependency relationship between the execution engine units, The data interaction channel between the execution engine units is constructed according to the data dependency, and the preset image recognition big data model is constructed based on the data interaction channel, the execution sequence and the execution engine unit.

在一个实施例中，处理器执行计算机程序时还实现以下步骤：识别最小功能单元对应的开发信息数据；根据最小功能单元的开发信息数据，确定最小功能单元对应执行配置信息；根据最小功能单元以及执行配置信息，获取执行引擎单元。In one embodiment, the processor further implements the following steps when executing the computer program: identifying the development information data corresponding to the smallest functional unit; determining the execution configuration information corresponding to the smallest functional unit according to the development information data of the smallest functional unit; Execute configuration information and obtain the execution engine unit.

在一个实施例中，处理器执行计算机程序时还实现以下步骤：获取初始图像识别大数据模型和模型重构集成配置参数，识别初始图像识别大数据模型中各功能组成部分对应的开发信息数据；根据初始图像识别大数据模型各功能组成部分对应的开发信息数据，将初始图像识别大数据模型拆分为各最小功能单元。In one embodiment, the processor further implements the following steps when executing the computer program: acquiring the initial image recognition big data model and model reconstruction integration configuration parameters, and identifying the development information data corresponding to each functional component in the initial image recognition big data model; According to the development information data corresponding to each functional component of the initial image recognition big data model, the initial image recognition big data model is divided into the smallest functional units.

在一个实施例中，处理器执行计算机程序时还实现以下步骤：获取初始图像识别大数据模型；根据预设标识符识别初始图像识别大数据模型内的参数数据，在参数数据前添加对应特殊符号，将初始图像识别大数据模型内的参数数据转化为参数值数据；执行预设拆分脚本，根据初始图像识别大数据模型对应的开发信息数据，将初始图像识别大数据模型拆分为最小组成单元；删除参数值数据前的特征符号，将最小组成单元内的参数值数据转化为参数数据，获得最小功能单元。In one embodiment, the processor further implements the following steps when executing the computer program: obtain the initial image to recognize the big data model; recognize the parameter data in the initial image to recognize the big data model according to the preset identifier, and add corresponding special symbols before the parameter data , Convert the parameter data in the initial image recognition big data model into parameter value data; execute the preset split script, and split the initial image recognition big data model into the smallest components according to the development information data corresponding to the initial image recognition big data model Unit: Delete the characteristic symbol before the parameter value data, and convert the parameter value data in the smallest component unit into parameter data to obtain the smallest functional unit.

在一个实施例中，处理器执行计算机程序时还实现以下步骤：获取执行引擎单元间的数据依赖关系，根据数据依赖关系构建执行引擎单元间的数据交互通道；基于数据交互通道以及执行顺序构建模型运行网络，并根据单元执行顺序定位各执行引擎单元在模型运行网络的执行位置；将执行引擎单元添加于模型运行网络对应执行位置，获得预设图像识别大数据模型。In one embodiment, the processor further implements the following steps when executing the computer program: acquiring the data dependency relationship between the execution engine units, constructing a data interaction channel between the execution engine units according to the data dependency relationship; constructing a model based on the data interaction channel and the execution sequence Run the network, and locate the execution position of each execution engine unit in the model running network according to the unit execution sequence; add the execution engine unit to the corresponding execution position of the model running network to obtain the preset image recognition big data model.

在一个实施例中，处理器执行计算机程序时还实现以下步骤：获取校验数据以及校验数据对应的中间结果数据以及校验输出结果数据；将预设校验数据输入预设图像识别大数据模型，获取预设校验数据对应的各个执行引擎单元的输出数据以及预设图像识别大数据模型输出的校验结果数据；对比预设输出结果数据与校验结果数据，获得验证结果，当验证结果为验证不通过时，对比预设中间结果数据与各个执行引擎单元的输出数据查找执行引擎单元中的问题单元，根据验证结果修正问题单元，返回将预设校验数据输入预设图像识别大数据模型的步骤。In one embodiment, the processor further implements the following steps when executing the computer program: acquiring the verification data and the intermediate result data corresponding to the verification data and the verification output result data; inputting the preset verification data into the preset image recognition big data Model, obtain the output data of each execution engine unit corresponding to the preset verification data and the verification result data output by the preset image recognition big data model; compare the preset output result data with the verification result data to obtain the verification result, when the verification When the result is that the verification fails, compare the preset intermediate result data with the output data of each execution engine unit to find the problem unit in the execution engine unit, correct the problem unit according to the verification result, and return the preset verification data to the preset image recognition library. The steps of the data model.

在一个实施例中，提供了一种计算机可读存储介质，所述计算机可读存储介质可以是非易失性，也可以是易失性，其上存储有计算机程序，计算机程序被处理器执行时实现以下步骤：In one embodiment, a computer-readable storage medium is provided. The computer-readable storage medium may be non-volatile or volatile, and a computer program is stored thereon. When the computer program is executed by a processor, Implement the following steps:

获取待识别图像；Obtain the image to be recognized;

将待识别图像输入预设图像识别大数据模型，获取图像识别结果；Input the image to be recognized into the preset image recognition big data model to obtain the image recognition result;

预设图像识别大数据模型是根据各执行引擎单元的数据交互通道，并基于各执行引擎单元的执行顺序执行各执行引擎单元构建得到，各执行引擎单元是根据模型重构集成配置参数对初始图像识别大数据模型拆分重构获得，各执行引擎单元的执行顺序由模型重构集成配置参数确定，数据交互通道由各执行引擎单元间的数据依赖关系构建。The preset image recognition big data model is constructed based on the data interaction channel of each execution engine unit, and executes each execution engine unit based on the execution order of each execution engine unit. Each execution engine unit is constructed according to the model reconstruction integration configuration parameter to the initial image The recognition big data model is obtained by splitting and reconstructing, the execution order of each execution engine unit is determined by the model reconstruction integration configuration parameters, and the data interaction channel is constructed by the data dependency relationship between each execution engine unit.

在一个实施例中，计算机程序被处理器执行时还实现以下步骤：获取初始图像识别大数据模型和模型重构集成配置参数，将初始图像识别大数据模型拆分为最小功能单元；获取最小功能单元对应的执行过程，根据最小功能单元以及与最小功能单元对应的执行过程，获取执行引擎单元；根据模型重构集成配置参数确定执行引擎单元的单元执行顺序；获取执行引擎单元间的数据依赖关系，根据数据依赖关系构建执行引擎单元间的数据交互通道，基于数据交互通道、执行顺序以及执行引擎单元构建预设图像识别大数据模型。In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: acquiring the initial image recognition big data model and model reconstruction integration configuration parameters, splitting the initial image recognition big data model into the smallest functional units; acquiring the smallest function The execution process corresponding to the unit, the execution engine unit is obtained according to the minimum functional unit and the execution process corresponding to the minimum functional unit; the unit execution order of the execution engine unit is determined according to the model reconstruction integration configuration parameter; the data dependency relationship between the execution engine units is obtained Establish a data interaction channel between the execution engine units according to the data dependency, and build a preset image recognition big data model based on the data interaction channel, the execution sequence and the execution engine unit.

在一个实施例中，计算机程序被处理器执行时还实现以下步骤：识别最小功能单元对应的开发信息数据；根据最小功能单元的开发信息数据，确定最小功能单元对应执行配置信息；根据最小功能单元以及执行配置信息，获取执行引擎单元。In one embodiment, when the computer program is executed by the processor, the following steps are further implemented: identifying the development information data corresponding to the smallest functional unit; determining the execution configuration information corresponding to the smallest functional unit according to the development information data of the smallest functional unit; And execute configuration information to obtain the execution engine unit.

在一个实施例中，计算机程序被处理器执行时还实现以下步骤：获取初始图像识别大数据模型和模型重构集成配置参数，识别初始图像识别大数据模型中各功能组成部分对应的开发信息数据；根据初始图像识别大数据模型各功能组成部分对应的开发信息数据，将初始图像识别大数据模型拆分为各最小功能单元。In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: acquiring the initial image recognition big data model and model reconstruction integration configuration parameters, and identifying the development information data corresponding to each functional component in the initial image recognition big data model ; According to the development information data corresponding to each functional component of the initial image recognition big data model, the initial image recognition big data model is split into the smallest functional units.

在一个实施例中，计算机程序被处理器执行时还实现以下步骤：获取初始图像识别大数据模型；根据预设标识符识别初始图像识别大数据模型内的参数数据，在参数数据前添加对应特殊符号，将初始图像识别大数据模型内的参数数据转化为参数值数据；执行预设拆分脚本，根据初始图像识别大数据模型对应的开发信息数据，将初始图像识别大数据模型拆分为最小组成单元；删除参数值数据前的特征符号，将最小组成单元内的参数值数据转化为参数数据，获得最小功能单元。In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: obtain the initial image to recognize the big data model; recognize the parameter data in the initial image to recognize the big data model according to the preset identifier, and add the corresponding special before the parameter data Symbol, convert the parameter data in the initial image recognition big data model into parameter value data; execute the preset split script, according to the development information data corresponding to the initial image recognition big data model, split the initial image recognition big data model into the smallest Component unit; delete the characteristic symbol before the parameter value data, and convert the parameter value data in the smallest component unit into parameter data to obtain the smallest functional unit.

在一个实施例中，计算机程序被处理器执行时还实现以下步骤：获取执行引擎单元间的数据依赖关系，根据数据依赖关系构建执行引擎单元间的数据交互通道；基于数据交互通道以及执行顺序构建模型运行网络，并根据单元执行顺序定位各执行引擎单元在模型运行网络的执行位置；将执行引擎单元添加于模型运行网络对应执行位置，获得预设图像识别大数据模型。In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: obtaining the data dependency relationship between the execution engine units, constructing a data interaction channel between the execution engine units according to the data dependency relationship; constructing based on the data interaction channel and the execution sequence Model running network, and locate the execution position of each execution engine unit in the model running network according to the unit execution sequence; add the execution engine unit to the corresponding execution position of the model running network to obtain the preset image recognition big data model.

在一个实施例中，计算机程序被处理器执行时还实现以下步骤：获取校验数据以及校验数据对应的中间结果数据以及校验输出结果数据；将预设校验数据输入预设图像识别大数据模型，获取预设校验数据对应的各个执行引擎单元的输出数据以及预设图像识别大数据模型输出的校验结果数据；对比预设输出结果数据与校验结果数据，获得验证结果，当验证结果为验证不通过时，对比预设中间结果数据与各个执行引擎单元的输出数据查找执行引擎单元中的问题单元，根据验证结果修正问题单元，返回将预设校验数据输入预设图像识别大数据模型的步骤。In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: obtaining the verification data and the intermediate result data corresponding to the verification data and the verification output result data; and inputting the preset verification data into the preset image recognition large Data model, obtain the output data of each execution engine unit corresponding to the preset verification data and the verification result data output by the preset image recognition big data model; compare the preset output result data with the verification result data to obtain the verification result, when When the verification result is that the verification fails, compare the preset intermediate result data with the output data of each execution engine unit to find the problem unit in the execution engine unit, correct the problem unit according to the verification result, and return the preset verification data to the preset image recognition Steps of the big data model.

本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程，是可以通过计算机程序来指令相关的硬件来完成，的计算机程序可存储于一非易失性计算机可读取存储介质中，该计算机程序在执行时，可包括如上述各方法的实施例的流程。其中，本申请所提供的各实施例中所使用的对存储器、存储、数据库或其它介质的任何引用，均可包括非易失性和/或易失性存储器。非易失性存储器可包括只读存储器（ROM）、可编程ROM（PROM）、电可编程ROM（EPROM）、电可擦除可编程ROM（EEPROM）或闪存。易失性存储器可包括随机存取存储器（RAM）或者外部高速缓冲存储器。作为说明而非局限，RAM以多种形式可得，诸如静态RAM（SRAM）、动态RAM（DRAM）、同步DRAM（SDRAM）、双数据率SDRAM（DDRSDRAM）、增强型SDRAM（ESDRAM）、同步链路（Synchlink） DRAM（SLDRAM）、存储器总线（Rambus）直接RAM（RDRAM）、直接存储器总线动态RAM（DRDRAM）、以及存储器总线动态RAM（RDRAM）等。A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium. When the computer program is executed, it may include the processes of the above-mentioned method embodiments. Wherein, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and/or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not a limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous chain Channel (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

以上实施例的各技术特征可以进行任意的组合，为使描述简洁，未对上述实施例中的各个技术特征所有可能的组合都进行描述，然而，只要这些技术特征的组合不存在矛盾，都应当认为是本说明书记载的范围。The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, all possible combinations of the technical features in the above embodiments are not described. However, as long as there is no contradiction in the combination of these technical features, they should be It is considered as the range described in this specification.

以上实施例仅表达了本申请的几种实施方式，其描述较为具体和详细，但并不能因此而理解为对发明专利范围的限制。应当指出的是，对于本领域的普通技术人员来说，在不脱离本申请构思的前提下，还可以做出若干变形和改进，这些都属于本申请的保护范围。因此，本申请专利的保护范围应以所附权利要求为准。The above embodiments only express several implementation manners of the present application, and the description is relatively specific and detailed, but it should not be understood as a limitation on the scope of the invention patent. It should be pointed out that for those of ordinary skill in the art, without departing from the concept of this application, several modifications and improvements can be made, and these all fall within the protection scope of this application. Therefore, the scope of protection of the patent of this application shall be subject to the appended claims.

Claims (20)

1. 一种图像识别方法，其中，所述方法包括：An image recognition method, wherein the method includes:

   获取待识别图像；Obtain the image to be recognized;

   将所述待识别图像输入预设图像识别大数据模型，获取图像识别结果；Input the image to be recognized into a preset image recognition big data model to obtain an image recognition result;

   所述预设图像识别大数据模型是根据各执行引擎单元的数据交互通道，并基于各执行引擎单元的执行顺序执行各执行引擎单元构建得到，所述各执行引擎单元是根据模型重构集成配置参数对初始图像识别大数据模型拆分重构获得，所述各执行引擎单元的执行顺序由模型重构集成配置参数确定，所述数据交互通道由所述各执行引擎单元间的数据依赖关系构建。The preset image recognition big data model is constructed based on the data interaction channel of each execution engine unit and executed by each execution engine unit based on the execution order of each execution engine unit, and each execution engine unit is constructed according to the model reconstruction integration configuration The parameters are obtained by splitting and reconstructing the initial image recognition big data model, the execution order of each execution engine unit is determined by the model reconstruction integration configuration parameter, and the data interaction channel is constructed by the data dependency relationship between the execution engine units .
2. 根据权利要求1所述的方法，其中，所述将所述待识别图像输入预设图像识别大数据模型，获取图像识别结果之前，还包括：The method according to claim 1, wherein said inputting said to-be-recognized image into a preset image recognition big data model, before obtaining the image recognition result, further comprises:

   获取初始图像识别大数据模型和模型重构集成配置参数，将所述初始图像识别大数据模型拆分为最小功能单元；Acquiring an initial image recognition big data model and model reconstruction integration configuration parameters, and splitting the initial image recognition big data model into minimum functional units;

   获取所述最小功能单元对应的执行过程，根据所述最小功能单元以及与所述最小功能单元对应的执行过程，获取执行引擎单元；Obtaining the execution process corresponding to the minimum functional unit, and obtaining the execution engine unit according to the minimum functional unit and the execution process corresponding to the minimum functional unit;

   根据所述模型重构集成配置参数确定所述执行引擎单元的单元执行顺序；Determining the unit execution sequence of the execution engine unit according to the model reconstruction integration configuration parameter;

   获取所述执行引擎单元间的数据依赖关系，根据所述数据依赖关系构建所述执行引擎单元间的数据交互通道，基于所述数据交互通道、所述执行顺序以及所述执行引擎单元构建预设图像识别大数据模型。Acquire the data dependency relationship between the execution engine units, construct a data interaction channel between the execution engine units according to the data dependency relationship, and construct a preset based on the data interaction channel, the execution sequence, and the execution engine unit Image recognition big data model.
3. 根据权利要求2所述的方法，其中，所述获取所述最小功能单元对应的执行过程，根据所述最小功能单元以及与所述最小功能单元对应的执行过程，获取执行引擎单元包括：3. The method according to claim 2, wherein said obtaining the execution process corresponding to the minimum functional unit, and obtaining the execution engine unit according to the minimum functional unit and the execution process corresponding to the minimum functional unit comprises:

   识别所述最小功能单元对应的开发信息数据；Identifying the development information data corresponding to the smallest functional unit;

   根据所述最小功能单元的开发信息数据，确定所述最小功能单元对应执行配置信息；Determine the execution configuration information corresponding to the smallest functional unit according to the development information data of the smallest functional unit;

   根据所述最小功能单元以及所述执行配置信息，获取执行引擎单元。Obtain an execution engine unit according to the minimum functional unit and the execution configuration information.
4. 根据权利要求2所述的方法，其中，所述获取初始图像识别大数据模型和模型重构集成配置参数，将所述初始图像识别大数据模型拆分为最小功能单元包括：The method according to claim 2, wherein said acquiring the initial image recognition big data model and model reconstruction integrated configuration parameters, and splitting the initial image recognition big data model into minimum functional units comprises:

   获取初始图像识别大数据模型和模型重构集成配置参数，识别所述初始图像识别大数据模型中各功能组成部分对应的开发信息数据；Acquiring the initial image recognition big data model and model reconstruction integration configuration parameters, and identifying the development information data corresponding to each functional component in the initial image recognition big data model;

   根据所述初始图像识别大数据模型各功能组成部分对应的开发信息数据，将所述初始图像识别大数据模型拆分为各最小功能单元。According to the development information data corresponding to each functional component of the initial image recognition big data model, the initial image recognition big data model is split into the smallest functional units.
5. 根据权利要求4所述的方法，其中，所述根据所述初始图像识别大数据模型各功能组成部分对应的开发信息数据，将所述初始图像识别大数据模型拆分为各最小功能单元包括：The method according to claim 4, wherein the splitting the initial image recognition big data model into the smallest functional units according to the development information data corresponding to each functional component of the initial image recognition big data model comprises:

   获取初始图像识别大数据模型；Obtain the initial image recognition big data model;

   根据预设标识符识别所述初始图像识别大数据模型内的参数数据，在所述参数数据前添加对应特殊符号，将初始图像识别大数据模型内的参数数据转化为参数值数据；Recognizing the parameter data in the initial image recognition big data model according to a preset identifier, adding corresponding special symbols before the parameter data, and converting the parameter data in the initial image recognition big data model into parameter value data;

   执行预设拆分脚本，根据所述初始图像识别大数据模型对应的开发信息数据，将所述初始图像识别大数据模型拆分为最小组成单元；Execute a preset split script, and split the initial image recognition big data model into minimum component units according to the development information data corresponding to the initial image recognition big data model;

   删除所述参数值数据前的所述特征符号，将所述最小组成单元内的参数值数据转化为参数数据，获得最小功能单元。The characteristic symbol before the parameter value data is deleted, the parameter value data in the minimum component unit is converted into parameter data, and the minimum functional unit is obtained.
6. 根据权利要求2所述的方法，其中，所述获取所述执行引擎单元间的数据依赖关系，根据所述数据依赖关系构建所述执行引擎单元间的数据交互通道，基于所述数据交互通道、所述执行顺序以及所述执行引擎单元构建预设图像识别大数据模型的步骤包括：3. The method according to claim 2, wherein said acquiring the data dependency relationship between the execution engine units, constructing a data interaction channel between the execution engine units according to the data dependency relationship, and based on the data interaction channel, The execution sequence and the steps of the execution engine unit constructing a preset image recognition big data model include:

   获取所述执行引擎单元间的数据依赖关系，根据所述数据依赖关系构建所述执行引擎单元间的数据交互通道；Acquiring the data dependency relationship between the execution engine units, and constructing a data exchange channel between the execution engine units according to the data dependency relationship;

   基于所述数据交互通道以及所述执行顺序构建模型运行网络，并根据所述单元执行顺序定位各执行引擎单元在所述模型运行网络的执行位置；Construct a model running network based on the data exchange channel and the execution sequence, and locate the execution position of each execution engine unit in the model running network according to the unit execution sequence;

   将所述执行引擎单元添加于所述模型运行网络对应执行位置，获得预设图像识别大数据模型。The execution engine unit is added to the corresponding execution position of the model running network to obtain a preset image recognition big data model.
7. 根据权利要求1所述的方法，其中，所述将所述待识别图像输入预设图像识别大数据模型，获取图像识别结果之前，还包括：The method according to claim 1, wherein said inputting said to-be-recognized image into a preset image recognition big data model, before obtaining the image recognition result, further comprises:

   获取预设校验数据以及所述预设校验数据对应的预设中间结果数据以及预设输出结果数据；Acquiring preset verification data and preset intermediate result data and preset output result data corresponding to the preset verification data;

   将所述预设校验数据输入预设图像识别大数据模型，获取预设校验数据对应的各个执行引擎单元的输出数据以及预设图像识别大数据模型输出的校验结果数据；Input the preset verification data into the preset image recognition big data model, and obtain the output data of each execution engine unit corresponding to the preset verification data and the verification result data output by the preset image recognition big data model;

   对比所述预设输出结果数据与所述校验结果数据，获得验证结果，当所述验证结果为验证不通过时，对比所述预设中间结果数据与各个执行引擎单元的输出数据查找所述执行引擎单元中的问题单元，根据所述验证结果修正所述问题单元，返回将所述预设校验数据输入预设图像识别大数据模型的步骤。Compare the preset output result data with the verification result data to obtain a verification result. When the verification result is that the verification fails, compare the preset intermediate result data with the output data of each execution engine unit to find the The problem unit in the execution engine unit corrects the problem unit according to the verification result, and returns to the step of inputting the preset verification data into the preset image recognition big data model.
8. 一种图像识别装置，其中，所述装置包括：An image recognition device, wherein the device includes:

   数据获取模块，用于获取待识别图像；The data acquisition module is used to acquire the image to be recognized;

   图像识别模块，用于将所述待识别图像输入预设图像识别大数据模型，获取图像识别结果；An image recognition module, configured to input the image to be recognized into a preset image recognition big data model to obtain an image recognition result;

   所述预设图像识别大数据模型是根据各执行引擎单元的数据交互通道，并基于各执行引擎单元的执行顺序执行各执行引擎单元构建得到，所述各执行引擎单元是根据模型重构集成配置参数对初始图像识别大数据模型拆分重构获得，所述各执行引擎单元的执行顺序由模型重构集成配置参数确定，所述数据交互通道由所述各执行引擎单元间的数据依赖关系构建。The preset image recognition big data model is constructed based on the data interaction channel of each execution engine unit and executed by each execution engine unit based on the execution order of each execution engine unit, and each execution engine unit is constructed according to the model reconstruction integration configuration The parameters are obtained by splitting and reconstructing the initial image recognition big data model, the execution order of each execution engine unit is determined by the model reconstruction integration configuration parameter, and the data interaction channel is constructed by the data dependency relationship between the execution engine units .
9. 一种计算机设备，包括存储器和处理器，所述存储器存储有计算机程序，其中，所述处理器执行所述计算机程序时实现如下步骤：A computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the following steps when the processor executes the computer program:

   获取待识别图像；Obtain the image to be recognized;

   将所述待识别图像输入预设图像识别大数据模型，获取图像识别结果；Input the image to be recognized into a preset image recognition big data model to obtain an image recognition result;

   所述预设图像识别大数据模型是根据各执行引擎单元的数据交互通道，并基于各执行引擎单元的执行顺序执行各执行引擎单元构建得到，所述各执行引擎单元是根据模型重构集成配置参数对初始图像识别大数据模型拆分重构获得，所述各执行引擎单元的执行顺序由模型重构集成配置参数确定，所述数据交互通道由所述各执行引擎单元间的数据依赖关系构建。The preset image recognition big data model is constructed based on the data interaction channel of each execution engine unit and executed by each execution engine unit based on the execution order of each execution engine unit, and each execution engine unit is constructed according to the model reconstruction integration configuration The parameters are obtained by splitting and reconstructing the initial image recognition big data model, the execution order of each execution engine unit is determined by the model reconstruction integration configuration parameter, and the data interaction channel is constructed by the data dependency relationship between the execution engine units .
10. 根据权利要求9所述的计算机设备，其中，所述处理器执行所述计算机程序时，在将所述待识别图像输入预设图像识别大数据模型，获取图像识别结果之前，还实现如下步骤：The computer device according to claim 9, wherein when the processor executes the computer program, before inputting the to-be-recognized image into a preset image recognition big data model and obtaining the image recognition result, the following steps are further implemented:

    获取初始图像识别大数据模型和模型重构集成配置参数，将所述初始图像识别大数据模型拆分为最小功能单元；Acquiring an initial image recognition big data model and model reconstruction integration configuration parameters, and splitting the initial image recognition big data model into minimum functional units;

    获取所述最小功能单元对应的执行过程，根据所述最小功能单元以及与所述最小功能单元对应的执行过程，获取执行引擎单元；Obtaining the execution process corresponding to the minimum functional unit, and obtaining the execution engine unit according to the minimum functional unit and the execution process corresponding to the minimum functional unit;

    根据所述模型重构集成配置参数确定所述执行引擎单元的单元执行顺序；Determining the unit execution sequence of the execution engine unit according to the model reconstruction integration configuration parameter;

    获取所述执行引擎单元间的数据依赖关系，根据所述数据依赖关系构建所述执行引擎单元间的数据交互通道，基于所述数据交互通道、所述执行顺序以及所述执行引擎单元构建预设图像识别大数据模型。Acquire the data dependency relationship between the execution engine units, construct a data interaction channel between the execution engine units according to the data dependency relationship, and construct a preset based on the data interaction channel, the execution sequence, and the execution engine unit Image recognition big data model.
11. 根据权利要求10所述的计算机设备，其中，所述获取所述最小功能单元对应的执行过程，根据所述最小功能单元以及与所述最小功能单元对应的执行过程，获取执行引擎单元包括：11. The computer device according to claim 10, wherein said obtaining the execution process corresponding to the minimum functional unit, and obtaining the execution engine unit according to the minimum functional unit and the execution process corresponding to the minimum functional unit comprises:

    识别所述最小功能单元对应的开发信息数据；Identifying the development information data corresponding to the smallest functional unit;

    根据所述最小功能单元的开发信息数据，确定所述最小功能单元对应执行配置信息；Determine the execution configuration information corresponding to the smallest functional unit according to the development information data of the smallest functional unit;

    根据所述最小功能单元以及所述执行配置信息，获取执行引擎单元。Obtain an execution engine unit according to the minimum functional unit and the execution configuration information.
12. 根据权利要求10所述的计算机设备，其中，所述获取初始图像识别大数据模型和模型重构集成配置参数，将所述初始图像识别大数据模型拆分为最小功能单元包括：The computer device according to claim 10, wherein said acquiring the initial image recognition big data model and model reconstruction integrated configuration parameters, and splitting the initial image recognition big data model into minimum functional units comprises:

    获取初始图像识别大数据模型和模型重构集成配置参数，识别所述初始图像识别大数据模型中各功能组成部分对应的开发信息数据；Acquiring the initial image recognition big data model and model reconstruction integration configuration parameters, and identifying the development information data corresponding to each functional component in the initial image recognition big data model;

    根据所述初始图像识别大数据模型各功能组成部分对应的开发信息数据，将所述初始图像识别大数据模型拆分为各最小功能单元。According to the development information data corresponding to each functional component of the initial image recognition big data model, the initial image recognition big data model is split into the smallest functional units.
13. 根据权利要求12所述的计算机设备，其中，所述根据所述初始图像识别大数据模型各功能组成部分对应的开发信息数据，将所述初始图像识别大数据模型拆分为各最小功能单元包括：The computer device according to claim 12, wherein the splitting of the initial image recognition big data model into the smallest functional units according to the development information data corresponding to each functional component of the initial image recognition big data model comprises :

    获取初始图像识别大数据模型；Obtain the initial image recognition big data model;

    根据预设标识符识别所述初始图像识别大数据模型内的参数数据，在所述参数数据前添加对应特殊符号，将初始图像识别大数据模型内的参数数据转化为参数值数据；Recognizing the parameter data in the initial image recognition big data model according to a preset identifier, adding corresponding special symbols before the parameter data, and converting the parameter data in the initial image recognition big data model into parameter value data;

    执行预设拆分脚本，根据所述初始图像识别大数据模型对应的开发信息数据，将所述初始图像识别大数据模型拆分为最小组成单元；Execute a preset split script, and split the initial image recognition big data model into minimum constituent units according to the development information data corresponding to the initial image recognition big data model;

    删除所述参数值数据前的所述特征符号，将所述最小组成单元内的参数值数据转化为参数数据，获得最小功能单元。The characteristic symbol before the parameter value data is deleted, the parameter value data in the minimum component unit is converted into parameter data, and the minimum functional unit is obtained.
14. 根据权利要求10所述的计算机设备，其中，所述获取所述执行引擎单元间的数据依赖关系，根据所述数据依赖关系构建所述执行引擎单元间的数据交互通道，基于所述数据交互通道、所述执行顺序以及所述执行引擎单元构建预设图像识别大数据模型的步骤包括：10. The computer device according to claim 10, wherein said acquiring the data dependency relationship between the execution engine units, and constructing a data interaction channel between the execution engine units according to the data dependency relationship, based on the data interaction channel , The execution sequence and the steps of the execution engine unit constructing a preset image recognition big data model include:

    获取所述执行引擎单元间的数据依赖关系，根据所述数据依赖关系构建所述执行引擎单元间的数据交互通道；Acquiring the data dependency relationship between the execution engine units, and constructing a data exchange channel between the execution engine units according to the data dependency relationship;

    基于所述数据交互通道以及所述执行顺序构建模型运行网络，并根据所述单元执行顺序定位各执行引擎单元在所述模型运行网络的执行位置；Construct a model running network based on the data exchange channel and the execution sequence, and locate the execution position of each execution engine unit in the model running network according to the unit execution sequence;

    将所述执行引擎单元添加于所述模型运行网络对应执行位置，获得预设图像识别大数据模型。The execution engine unit is added to the corresponding execution position of the model running network to obtain a preset image recognition big data model.
15. 根据权利要求9所述的计算机设备，其中，所述处理器执行所述计算机程序时，在所述将所述待识别图像输入预设图像识别大数据模型，获取图像识别结果之前，还实现如下步骤：The computer device according to claim 9, wherein, when the processor executes the computer program, before the image to be recognized is input into a preset image recognition big data model and the image recognition result is obtained, the following is further implemented step:

    获取预设校验数据以及所述预设校验数据对应的预设中间结果数据以及预设输出结果数据；Acquiring preset verification data and preset intermediate result data and preset output result data corresponding to the preset verification data;

    将所述预设校验数据输入预设图像识别大数据模型，获取预设校验数据对应的各个执行引擎单元的输出数据以及预设图像识别大数据模型输出的校验结果数据；Input the preset verification data into the preset image recognition big data model, and obtain the output data of each execution engine unit corresponding to the preset verification data and the verification result data output by the preset image recognition big data model;

    对比所述预设输出结果数据与所述校验结果数据，获得验证结果，当所述验证结果为验证不通过时，对比所述预设中间结果数据与各个执行引擎单元的输出数据查找所述执行引擎单元中的问题单元，根据所述验证结果修正所述问题单元，返回将所述预设校验数据输入预设图像识别大数据模型的步骤。Compare the preset output result data with the verification result data to obtain a verification result. When the verification result is that the verification fails, compare the preset intermediate result data with the output data of each execution engine unit to find the The problem unit in the execution engine unit corrects the problem unit according to the verification result, and returns to the step of inputting the preset verification data into the preset image recognition big data model.
16. 一种计算机可读存储介质，其上存储有计算机程序，其中，所述计算机程序被处理器执行时实现如下步骤：A computer-readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement the following steps:

    获取待识别图像；Obtain the image to be recognized;

    将所述待识别图像输入预设图像识别大数据模型，获取图像识别结果；Input the image to be recognized into a preset image recognition big data model to obtain an image recognition result;

    所述预设图像识别大数据模型是根据各执行引擎单元的数据交互通道，并基于各执行引擎单元的执行顺序执行各执行引擎单元构建得到，所述各执行引擎单元是根据模型重构集成配置参数对初始图像识别大数据模型拆分重构获得，所述各执行引擎单元的执行顺序由模型重构集成配置参数确定，所述数据交互通道由所述各执行引擎单元间的数据依赖关系构建。The preset image recognition big data model is constructed based on the data interaction channel of each execution engine unit and executed by each execution engine unit based on the execution order of each execution engine unit, and each execution engine unit is constructed according to the model reconstruction integration configuration The parameters are obtained by splitting and reconstructing the initial image recognition big data model, the execution order of each execution engine unit is determined by the model reconstruction integration configuration parameter, and the data interaction channel is constructed by the data dependency relationship between the execution engine units .
17. 根据权利要求16所述的计算机可读存储介质，其中，所述计算机程序被处理器执行时，在将所述待识别图像输入预设图像识别大数据模型，获取图像识别结果之前，还实现如下步骤：The computer-readable storage medium according to claim 16, wherein, when the computer program is executed by the processor, before the image to be recognized is input into a preset image recognition big data model, and the image recognition result is obtained, the following is further implemented step:

    获取初始图像识别大数据模型和模型重构集成配置参数，将所述初始图像识别大数据模型拆分为最小功能单元；Acquiring an initial image recognition big data model and model reconstruction integration configuration parameters, and splitting the initial image recognition big data model into minimum functional units;

    获取所述最小功能单元对应的执行过程，根据所述最小功能单元以及与所述最小功能单元对应的执行过程，获取执行引擎单元；Obtaining the execution process corresponding to the minimum functional unit, and obtaining the execution engine unit according to the minimum functional unit and the execution process corresponding to the minimum functional unit;

    根据所述模型重构集成配置参数确定所述执行引擎单元的单元执行顺序；Determining the unit execution sequence of the execution engine unit according to the model reconstruction integration configuration parameter;

    获取所述执行引擎单元间的数据依赖关系，根据所述数据依赖关系构建所述执行引擎单元间的数据交互通道，基于所述数据交互通道、所述执行顺序以及所述执行引擎单元构建预设图像识别大数据模型。Acquire the data dependency relationship between the execution engine units, construct a data interaction channel between the execution engine units according to the data dependency relationship, and construct a preset based on the data interaction channel, the execution sequence, and the execution engine unit Image recognition big data model.
18. 根据权利要求17所述的计算机可读存储介质，其中，所述获取所述最小功能单元对应的执行过程，根据所述最小功能单元以及与所述最小功能单元对应的执行过程，获取执行引擎单元包括：18. The computer-readable storage medium according to claim 17, wherein the acquiring the execution process corresponding to the smallest functional unit, and acquiring the execution engine unit according to the smallest functional unit and the execution process corresponding to the smallest functional unit include:

    识别所述最小功能单元对应的开发信息数据；Identifying the development information data corresponding to the smallest functional unit;

    根据所述最小功能单元的开发信息数据，确定所述最小功能单元对应执行配置信息；Determine the execution configuration information corresponding to the smallest functional unit according to the development information data of the smallest functional unit;

    根据所述最小功能单元以及所述执行配置信息，获取执行引擎单元。Obtain an execution engine unit according to the minimum functional unit and the execution configuration information.
19. 根据权利要求17所述的计算机可读存储介质，其中，所述获取初始图像识别大数据模型和模型重构集成配置参数，将所述初始图像识别大数据模型拆分为最小功能单元包括：The computer-readable storage medium according to claim 17, wherein said acquiring the initial image recognition big data model and model reconstruction integrated configuration parameters, and splitting the initial image recognition big data model into minimum functional units comprises:

    获取初始图像识别大数据模型和模型重构集成配置参数，识别所述初始图像识别大数据模型中各功能组成部分对应的开发信息数据；Acquiring the initial image recognition big data model and model reconstruction integration configuration parameters, and identifying the development information data corresponding to each functional component in the initial image recognition big data model;

    根据所述初始图像识别大数据模型各功能组成部分对应的开发信息数据，将所述初始图像识别大数据模型拆分为各最小功能单元。According to the development information data corresponding to each functional component of the initial image recognition big data model, the initial image recognition big data model is split into the smallest functional units.
20. 根据权利要求19所述的计算机可读存储介质，其中，所述根据所述初始图像识别大数据模型各功能组成部分对应的开发信息数据，将所述初始图像识别大数据模型拆分为各最小功能单元包括：The computer-readable storage medium according to claim 19, wherein, according to the development information data corresponding to each functional component of the initial image recognition big data model, the initial image recognition big data model is split into each smallest Functional units include:

    获取初始图像识别大数据模型；Obtain the initial image recognition big data model;

    根据预设标识符识别所述初始图像识别大数据模型内的参数数据，在所述参数数据前添加对应特殊符号，将初始图像识别大数据模型内的参数数据转化为参数值数据；Recognizing the parameter data in the initial image recognition big data model according to a preset identifier, adding corresponding special symbols before the parameter data, and converting the parameter data in the initial image recognition big data model into parameter value data;

    执行预设拆分脚本，根据所述初始图像识别大数据模型对应的开发信息数据，将所述初始图像识别大数据模型拆分为最小组成单元；Execute a preset split script, and split the initial image recognition big data model into minimum component units according to the development information data corresponding to the initial image recognition big data model;

    删除所述参数值数据前的所述特征符号，将所述最小组成单元内的参数值数据转化为参数数据，获得最小功能单元。The characteristic symbol before the parameter value data is deleted, the parameter value data in the minimum component unit is converted into parameter data, and the minimum functional unit is obtained.