Method for constructing orchestration operator generation process based on AI dialogue mode

Abstract

The invention relates to a method for constructing an orchestration operator generation process based on an AI dialogue mode, and the method comprises the steps: describing an image processing process to be constructed to a software platform in a dialogue mode, and submitting a demand description; the software platform analyzes and analyzes the demand description through a natural language processing technology, and key information is extracted; generating a corresponding image processing flow arrangement operator according to the extracted key information; processing and analyzing the generated image processing flow arrangement operator by using a precipitated and trained industry field large model, and quickly drawing and constructing an image flow matched with the image processing flow arrangement operator; and the software platform displays the generated image processing flow in a graphical mode. According to the method, the image processing flow meeting the requirements can be quickly constructed according to the requirements of the user, complex machine vision tasks are supported, the adaptability and flexibility of the system are improved, the construction and debugging efficiency of the image processing flow is improved, the use threshold is reduced, and the time and energy of the user are saved.

Description

Technical Field

[0001] The present invention relates to the field of image processing, and in particular to a method for constructing an orchestration operator generation process based on AI conversational construction. Background Art

[0002] In the existing mainstream machine vision software, the image processing mode is basically completed by dragging and arranging different operators in series, so that users do not need to learn a development language to complete image processing. The demand can be achieved by dragging and dropping, which greatly reduces the learning and use threshold of the product. The current operator arrangement process is: ① Create a new image processing process canvas ② Drag the adaptation operator to the canvas based on image requirements ③ Configure the corresponding parameters to process the image ④ Drag and control its execution order according to image processing requirements ⑤ Debug the image processing process ⑥ Run and publish the image processing process. The existing technology and process have the following problems: First, this method requires users to have certain knowledge and skills of image processing. For users who do not have this expertise, this method is difficult to get started and use, that is, it cannot lower the threshold for use to allow end users to become users. Secondly, this method requires users to manually write code or use low-code tools to build and debug processes such as image recognition, detection, and matching. This method takes a lot of time and effort, and the efficiency is still low. At the same time, there are limitations when dealing with personalized needs. The traditional image processing process construction method often fails to meet user needs when faced with complex machine vision tasks, and cannot quickly build a machine vision process that meets user needs. Summary of the invention

[0003] In order to solve the technical problems existing in the prior art, the present invention provides a method for constructing an orchestration operator generation process based on AI conversational construction.

[0004] To achieve the above object, the technical solution of the present invention is as follows:

[0005] A method for constructing an orchestration operator generation process based on AI conversational construction includes the following steps:

[0006] Step 1: Describe the image processing process to be built to the software platform through dialogue and submit the requirement description;

[0007] Step 2: The software platform uses natural language processing technology to analyze and parse the demand description and extract key information;

[0008] Step 3: The software platform generates the corresponding image processing flow arrangement operator based on the extracted key information;

[0009] Step 4: The software platform uses the industry-specific large model that has been precipitated and trained to process and analyze the generated image processing flow orchestration operator, and quickly draws and builds a matching image flow;

[0010] Step 5: The software platform displays the generated image processing flow in a graphical manner.

[0011] As a preferred technical solution, step one includes: describing to the software platform the construction of the image recognition, detection, matching, and classification processes required by the software platform through text input, file, and image upload.

[0012] As a preferred technical solution, in step three, for image processing processes that require identification, the software platform system generates an image recognition operator to identify the input image; for image processing processes that require detection, the system generates an image detection operator to detect specific objects in the input image; for image processing processes that require matching, the system generates an image matching operator to match the input image with the template image.

[0013] As a preferred technical solution, in step five, the graphical image processing flow supports viewing, editing and adjustment.

[0014] As a preferred technical solution, in step four, the software platform initially focuses on fragment generation in the process. As the degree and frequency of use increase, the model maturity improves and evolves to focus on global generation.

[0015] As a preferred technical solution, let t be a time variable, representing the time process of the large model from the start of training to its use and subsequent training; let M(t) be the maturity of the large model at time t, and the value range of M(t) is from 0 to 1, 0 represents complete immaturity, and 1 represents complete maturity; let K(t) be the amount of industry knowledge accumulated by the large model at time t; let U(t) be the frequency of use of the large model at time t; let α and β be weight coefficients, representing the relative importance of industry knowledge accumulation and use frequency to model maturity, respectively, and α+β=1, 0<α<1, 0<β<1, the formula for model maturity is:

[0016]

[0017] Compared with the prior art, the method of the present invention for constructing an orchestration operator generation process based on AI conversational style has the following beneficial effects:

[0018] (1) This technical solution uses a conversational construction and arrangement of operator generation processes, so that users can quickly build image processing processes that meet their needs without any programming skills, greatly lowering the threshold and allowing more people to easily use machine vision technology. (2) This technical solution uses a conversational construction and arrangement of operator generation processes, so that machine vision processes that meet user needs can be quickly built in a short period of time, greatly improving efficiency. (3) This technical solution uses a conversational construction and arrangement of operator generation processes, so that machine vision processes that meet user needs can be quickly built according to user needs, meeting the needs of complex tasks. (4) This technical solution uses a conversational construction and arrangement of operator generation processes, so that users can easily describe the machine vision process to be built through dialogue, and the system can quickly draw and build an image process that matches it and provide it to users, greatly facilitating user use. (5) This technical solution uses a conversational construction and arrangement of operator generation processes, so that machine vision processes that meet user needs can be quickly built in a short period of time, thereby saving time and labor costs and improving resource utilization efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a flowchart of the method of the present invention for constructing an orchestration operator generation process based on AI conversational style. DETAILED DESCRIPTION

[0020] The technical solution of the present invention is further described below in conjunction with specific implementation methods:

[0021] like Figure 1 As shown, a method for constructing an orchestration operator generation process based on AI conversational construction includes the following steps:

[0022] Step 1: Describe the image processing process to be built to the software platform through dialogue and submit the requirement description; for example, the user can describe the image recognition, detection, matching, classification and other processes to the software platform through text input, file, or image upload.

[0023] Step 2: The software platform uses natural language processing technology to analyze and parse the demand description and extract key information;

[0024] Step three, the software platform generates the corresponding image processing flow orchestration operator based on the extracted key information; for example, for an image processing flow that needs to be identified, the software platform system generates an image recognition operator to identify the input image; for an image processing flow that needs to be detected, the system generates an image detection operator to detect specific objects in the input image; for an image processing flow that needs to be matched, the system generates an image matching operator to match the input image with the template image.

[0025] Step 4: The software platform uses the industry-specific large model that has been precipitated and trained to process and analyze the generated image processing flow orchestration operator, and quickly draws and builds a matching image flow;

[0026] Step 5: The software platform displays the generated image processing flow to the user in a graphical manner, and the user can view, edit and adjust the flow. For example, the software platform can generate a flow chart to show the various steps of the image processing flow and the connection relationship between them, and the user can edit and adjust the flow by dragging and dropping.

[0027] The software platform can quickly build an image processing process that meets the needs of users, support complex machine vision tasks, and improve the adaptability and flexibility of the system. The software platform uses AI conversational construction and orchestration operator generation process technology to quickly draw and build matching image processes, improve the efficiency of image processing process construction and debugging, and save users' time and energy. Users can describe the image processing process to be built through dialogue, without the need for programming skills, which lowers the threshold for use and allows more users to easily use the technology. For example, users can describe the need for image classification tasks through dialogue, and the system can quickly build an image classification process according to user needs and achieve accurate image classification.

[0028] In the early stage of the software platform, the generation of process fragments was mainly carried out. As the degree and frequency of use increased, the model maturity increased, and the global generation gradually became the main focus. The formula for model maturity is as follows:

[0029] Let t be a time variable, which represents the time process of the large model from the beginning of training to its use and subsequent training. The unit can be months, quarters, or years;

[0030] Let M(t) be the maturity of the large model at time t. The value of M(t) ranges from 0 to 1, where 0 represents complete immaturity and 1 represents complete maturity.

[0031] Let K(t) be the amount of industry knowledge accumulated by the large model at time t. K(t) can be a function that grows over time. The amount of knowledge can be measured in appropriate units such as the number of knowledge items or the amount of data (bytes).

[0032] Let U(t) be the usage frequency of the large model at time t. The usage frequency can be an indicator such as the number of times the large model is called per unit time;

[0033] Let α and β be weight coefficients, representing the relative importance of industry knowledge accumulation and usage frequency to model maturity, and α+β=1, 0<α<1, 0<β<1,

[0034]

[0035] Explanation of each part of the formula:

[0036] M(0) is the initial maturity state of the model. For example, for a newly developed large model that has not yet undergone a lot of industry knowledge training and actual use, M(0) may be close to 0, such as

[0037] M(0) = 0.1, which means that the model has only some basic structures and general knowledge, but its maturity in specific areas is very low.

[0038] This part reflects the contribution of industry knowledge accumulation to model maturity, and α determines the importance of industry knowledge accumulation in the model maturity process. Indicates the relative progress of industry knowledge accumulation. For example, if α = 0.6, as time t passes, the amount of industry knowledge K(t) increases from the initial K(0). When K(t) approaches K max , the contribution of this part to the model maturity will be close to 0.6.

[0039] This part reflects the contribution of usage frequency to model maturity. β determines the importance of usage frequency in the model maturity process. Indicates the relative growth of usage frequency. For example, if β = 0.4, when the usage frequency U(t) increases rapidly from the initial U(0) and approaches U max , the contribution of this part to the model maturity will be close to 0.4.

[0040] Compared with the existing technologies and processes, this technical solution mainly solves the following technical problems: (1) The technology of constructing and arranging the operator generation process based on the AI ​​big model in a conversational way allows users to describe the image processing process to be built to meet their needs through conversations, without the need for professional knowledge of image processing and programming skills, thus lowering the threshold for use. (2) After describing the needs in natural language, the AI ​​big model will understand the needs and automatically arrange and integrate operators based on the trained models, without the need for users to understand the capabilities and related configurations of the operators in the operator library. (3) The problem of low construction and debugging efficiency: The existing image processing process construction technology requires users to manually drag or write code to build and debug processes such as image recognition, detection, and matching, which takes a lot of time and effort. This technical solution uses the technology of constructing and arranging the operator generation process based on the AI ​​conversational way to quickly draw and build an image process that matches it, thereby improving construction and debugging efficiency. (4) This technical solution can realize the fragment / global construction of the image process. In the early stage when the maturity is not enough, the needs can be met through the fragment construction process, so that users have confidence when using this technology. After the maturity gradually increases, the needs can be appropriately met through the global construction, so that users can complete the needs conveniently and easily.

[0041] This embodiment is only a further explanation of the invention rather than a limitation of the invention. After reading this specification, those skilled in the art may make non-creative modifications to this embodiment as needed. However, as long as they are within the scope of the claims of the invention, they are protected by the patent law.

Claims

1. A method for constructing an orchestration operator generation process based on AI conversational style, characterized in that: The following steps are involved: Step 1: Describe the image processing process to be built to the software platform through dialogue and submit the requirement description; Step 2: The software platform analyzes and parses the demand description through natural language processing technology to extract key information; Step 3, the software platform generates a corresponding image processing flow arrangement operator according to the extracted key information; Step 4: The software platform uses the industry-specific large model that has been precipitated and trained to process and analyze the generated image processing flow orchestration operator, and quickly draws and constructs an image flow that matches it; Step 5: The software platform displays the generated image processing flow in a graphical manner.

2. The method for generating a flow of an orchestration operator based on AI conversational construction according to claim 1, characterized in that: The step 1 includes: describing to the software platform the construction of the image recognition, detection, matching and classification process required by the software platform through text input, file and image upload.

3. The method for generating a flow of an orchestration operator based on AI conversational construction according to claim 2, characterized in that: In the step 3, for the image processing flow that needs to be identified, the software platform system generates an image recognition operator to identify the input image; For the image processing flow that needs to be detected, the system generates an image detection operator to detect specific objects in the input image; For the image processing flow that needs to be matched, the system generates an image matching operator to match the input image with the template image.

4. The method for generating a flow of an orchestration operator based on AI conversational construction according to claim 1, characterized in that: In step five, the graphical image processing flow supports viewing, editing and adjustment.

5. The method for generating a flow of an orchestration operator based on AI conversational construction according to claim 1, characterized in that: In the step 4, the software platform is initially based on process segment generation. As the usage level and frequency increase, the model maturity improves and evolves to be based on global generation.

6. The method for generating a flow of an orchestration operator based on AI conversational construction according to claim 5, characterized in that: Let t be the time variable, which represents the time process of the large model from the beginning of training to its use and subsequent training; let M(t) be the maturity of the large model at time t, and the value range of M(t) is from 0 to 1, where 0 represents complete immaturity and 1 represents complete maturity; let K(t) be the amount of industry knowledge accumulated by the large model at time t; let U(t) be the frequency of use of the large model at time t; let α and β be weight coefficients, which represent the relative importance of industry knowledge accumulation and use frequency to model maturity, respectively, and α+β=1, 0<α<1, 0<β<1, and the formula for model maturity is:

Images