CN114237635B - Method, system and storage medium for rapid deployment, operation and maintenance of semiconductor visual inspection - Google Patents
Method, system and storage medium for rapid deployment, operation and maintenance of semiconductor visual inspection Download PDFInfo
- Publication number
- CN114237635B CN114237635B CN202210168459.7A CN202210168459A CN114237635B CN 114237635 B CN114237635 B CN 114237635B CN 202210168459 A CN202210168459 A CN 202210168459A CN 114237635 B CN114237635 B CN 114237635B
- Authority
- CN
- China
- Prior art keywords
- algorithm
- debugging
- data
- algorithm function
- training
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F8/00—Arrangements for software engineering
- G06F8/60—Software deployment
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/36—Preventing errors by testing or debugging software
- G06F11/362—Software debugging
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F8/00—Arrangements for software engineering
- G06F8/70—Software maintenance or management
- G06F8/71—Version control; Configuration management
Abstract
The application provides a method, a system and a storage medium for rapid deployment, operation and maintenance of semiconductor visual inspection. The method adopts the idea of low code, rapidly realizes the debugging visualization and the logic flow of the machine vision algorithm in a configuration mode, and rapidly deploys in the form of a solution package; a large number of manually operated functional scenes in the field of semiconductor packaging and wafer detection are integrated in a set of software platform, the configuration parameters, the training data and the training model of a machine table are temporarily stored and the version is managed, the model training process is monitored in real time through a visual chart, the model data are accurate and quick, and the problems of low development response speed, high maintenance cost in the field production process, high debugging iteration complexity and the like caused by complex and variable requirements are solved.
Description
Technical Field
The application belongs to the technical field of semiconductor detection, and particularly relates to a method, a system and a storage medium for rapid deployment, operation and maintenance of semiconductor visual detection.
Background
As the foundation of modern information technology industry, the semiconductor industry has become the fundamental, strategic and precedent industry of social development and national economy, and is an essential important component of modern daily life and future technological progress. With the continuous expansion of the processing difficulty caused by the doubling reduction of the processing area of the semiconductor product, the manufacturing equipment for producing the semiconductor product in the future becomes more and more refined. Some new technologies in recent years have been more and more developing a wider variety of semiconductor chips, such as: AI chip, 5G chip, thing networking chip, memory chip, optical chip and MEMS chip etc.. With the increase of the types of chips and the stricter requirements of chips of the same type, the requirements for quality inspection become stricter.
Semiconductor chip detection methods are diverse, ranging from traditional image algorithms, machine learning algorithms, to developing AI neural network algorithms. In addition, different chip manufacturing stages and different chip types have different requirements on detection, and a large number of detection algorithms need to be developed to meet the complex requirements. In order to deal with these changes, a set of relatively standard software flow is also urgently needed to accelerate the development, deployment, operation and maintenance and other stages in the chip quality inspection process.
Algorithms developed for the detection requirements of different chips are usually adapted by programming changes in the process of deploying the algorithms to field detection software. After that, parameter debugging is needed to obtain better results, and visual result display is often implemented with the aid of software programming. The whole process is repeated with increasing demand.
With the increasing refinement of chip size, there are multiple chip instances under one camera view, so the whole image needs to be cut before detecting a single chip. If AI algorithm is used for detection, the cut small materials are required to be marked and classified. And then returning the labeled data set to perform model training. The whole process needs a lot of human intervention, such as whether the cutting result is correct, naming and specification in the classification process, verification in the data copying process, management in the model iteration process and the like. Problems in one link can cause final result errors and difficult troubleshooting, so that the problems of low development response speed, high maintenance cost in the field production process, high debugging iteration complexity and the like exist.
Disclosure of Invention
In view of this, the application provides a method, a system and a storage medium for rapid deployment and operation and maintenance of semiconductor visual inspection, a set of standard flow and management mode is standardized by combining industry experience, and deployment and operation and maintenance are performed in a software automation mode, so that the problems of low development response speed, high maintenance cost in a field production process, high complexity of debugging iteration and the like caused by complex and variable requirements are solved.
The specific technical scheme of the application is as follows:
the application provides a method for rapid deployment and operation and maintenance of semiconductor visual inspection, which comprises the following steps:
analyzing the input parameters and the output results of the algorithm function according to the preset data type and format, and carrying out visualization processing on the input parameters and the output results of the algorithm function;
generalizing the logical relation of the algorithm functions according to the deployment result of the visualization processing, and reconfiguring the logical relation of a plurality of algorithm functions in the target solution to generate a debugging algorithm library;
carrying out visual positioning and cutting manufacturing on the image coordinate and the physical coordinate of the semiconductor chip by using a debugging algorithm library, and outputting debugging data in real time;
and labeling, classifying and generating a training set which can be analyzed by an algorithm function on the cut small graph, and generating and iteratively debugging a model by using the training set.
Preferably, analyzing the input parameters and the output results of the algorithm function, and performing visualization processing on the input parameters and the output results of the algorithm function specifically comprises:
respectively registering and defining input parameters and output results of the algorithm function according to the data type and format;
analyzing the data type of the algorithm function input parameter to generate a corresponding UI input control, and loading an output result of the algorithm function in an image interaction mode according to the UI input control;
and analyzing the data type of the algorithm function output result to generate a corresponding drawing element, and drawing on the image by using the drawing element.
Preferably, inducing the logical relationship of the algorithm function according to the deployment result of the visualization processing, and reconfiguring the logical relationship of the plurality of algorithm functions in the target solution specifically include:
acquiring the logic relationship between the algorithm functions at the boundary according to the region division condition in the deployment result, and judging whether the logic relationship of each algorithm function needs to be corrected;
if the correction is needed, identifying the relation parameters among algorithm functions with problems in the target solution;
and modifying the relation format and the node attribute among the logic nodes in the identified relation parameters according to the area division condition in the deployment result.
Preferably, reconfiguring the logical relationship of the plurality of algorithm functions in the target solution, and generating the debugging algorithm library specifically includes:
reconfiguring the logic relationship between the algorithm functions at the boundary, and generating the logic relationship of each algorithm function in the debugging algorithm library, wherein the logic relationship comprises dependence, sharing and parallelism;
when the logic relation is dependence, generating an output parameter of a next algorithm function according to the output parameter of the previous algorithm function;
when the logic relation is sharing, synchronizing the input parameter of the previous algorithm function with the input parameter of the next algorithm function;
and when the logical relation is in common, synchronizing the input parameters, operators and output parameters of the previous algorithm function with the next algorithm function.
Preferably, the method further comprises the following steps:
monitoring the output debugging data in real time, and comparing the debugging data with preset debugging data, wherein the preset debugging data is generated according to a debugging algorithm library under a visual positioning and cutting manufacturing scheme;
and judging whether the debugging algorithm library needs to be adjusted or not according to the comparison result of the debugging data, and if the debugging algorithm library needs to be adjusted again, reconfiguring the logic relation of the algorithm function in the solution.
Preferably, after labeling and classifying the cut small graphs, before generating a training set that can be analyzed by an algorithm function, the method further includes:
acquiring the cut picture label data and the cut picture data, and performing forced format check on the label data and the cut picture data according to the byte attributes of the data;
and if the format verification result does not pass, correcting and aligning the label data and the graphic data by using a preset format template.
Preferably, the generating and iterative debugging model by using the training set specifically comprises:
uploading the generated training set to a cloud platform through tool software, and calling a debugging algorithm library in the cutting process;
training the debugging algorithm library by using the training set generated in the previous step, and sequentially generating debugging models in each stage;
and (3) carrying out updating iteration in the generation process by downloading a suitable debugging model training in the cutting and manufacturing process.
Preferably, the method further comprises the following steps:
monitoring a training process in real time, and acquiring a progress parameter and an effect parameter in the training process;
and generating a training task monitoring visual chart in real time according to the progress parameters and the effect parameters, wherein feedback indexes of the monitoring visual chart comprise label detection, accuracy, progress and loss.
In a second aspect of the present application, a system for rapid deployment and operation and maintenance of semiconductor visual inspection includes a memory and a processor, where the memory includes a program of the method for rapid deployment and operation and maintenance of semiconductor visual inspection, and when the program is executed by the processor, the following steps are implemented:
analyzing the input parameters and the output results of the algorithm function according to the preset data type and format, and carrying out visualization processing on the input parameters and the output results of the algorithm function;
inducing the logical relationship of the algorithm functions according to the deployment result of the visualization processing, and reconfiguring the logical relationship of a plurality of algorithm functions in the target solution to generate a debugging algorithm library;
carrying out visual positioning and cutting manufacturing on the image coordinate and the physical coordinate of the semiconductor chip by using a debugging algorithm library, and outputting debugging data in real time;
and labeling the cut small pictures, classifying the small pictures to generate a training set which can be analyzed by an algorithm function, and generating and iteratively debugging a model by using the training set.
Preferably, analyzing the input parameters and the output results of the algorithm function, and performing visualization processing on the input parameters and the output results of the algorithm function specifically comprises:
respectively registering and defining input parameters and output results of the algorithm function according to the data type and format;
analyzing the data type of the algorithm function input parameter to generate a corresponding UI input control, and loading an output result of the algorithm function in an image interaction mode according to the UI input control;
and analyzing the data type of the algorithm function output result to generate a corresponding drawing element, and drawing on the image by using the drawing element.
Preferably, inducing the logical relationship of the algorithm functions according to the deployment result of the visualization processing, and reconfiguring the logical relationship of the plurality of algorithm functions in the target solution specifically include:
acquiring the logic relationship between the algorithm functions at the boundary according to the region division condition in the deployment result, and judging whether the logic relationship of each algorithm function needs to be corrected;
if the correction is needed, identifying the relation parameters among algorithm functions with problems in the target solution;
and modifying the relation format and the node attribute among the logic nodes in the identified relation parameters according to the region division condition in the deployment result.
Preferably, reconfiguring a logical relationship of a plurality of algorithm functions in the target solution, and generating the debugging algorithm library specifically includes:
reconfiguring the logic relationship between the algorithm functions at the boundary, and generating the logic relationship of each algorithm function in a debugging algorithm library, wherein the logic relationship comprises dependence, sharing and parallelism;
when the logic relation is dependence, generating an output parameter of a next algorithm function according to the output parameter of the previous algorithm function;
when the logic relation is sharing, synchronizing the input parameter of the previous algorithm function with the input parameter of the next algorithm function;
and when the logic relation is in common, synchronizing the input parameters, the operators and the output parameters of the previous algorithm function with the next algorithm function.
Preferably, the method further comprises the following steps:
monitoring the output debugging data in real time, and comparing the debugging data with preset debugging data, wherein the preset debugging data is generated according to a debugging algorithm library under a visual positioning and cutting manufacturing scheme;
and judging whether the debugging algorithm library needs to be adjusted or not according to the comparison result of the debugging data, and if the debugging algorithm library needs to be adjusted again, reconfiguring the logic relation of the algorithm function in the solution.
Preferably, after labeling and classifying the cut small graphs, before generating a training set that can be analyzed by an algorithm function, the method further includes:
acquiring the cut picture label data and the cut picture data, and performing forced format check on the label data and the cut picture data according to the byte attributes of the data;
and if the format verification result does not pass, correcting and aligning the label data and the graphic data by using a preset format template.
Preferably, the generating and iteratively debugging model by using the training set specifically comprises:
uploading the generated training set to a cloud platform through tool software, and calling a debugging algorithm library in the cutting process;
training the debugging algorithm library by using the training set generated in the previous step, and sequentially generating debugging models in each stage;
and (3) carrying out updating iteration in the generation process by downloading a suitable debugging model training in the cutting and manufacturing process.
Preferably, the method further comprises the following steps:
monitoring a training process in real time, and acquiring a progress parameter and an effect parameter in the training process;
and generating a training task monitoring visual chart in real time according to the progress parameters and the effect parameters, wherein feedback indexes of the monitoring visual chart comprise label detection, accuracy, progress and loss.
A third aspect of the present application provides a computer-readable storage medium, where the computer-readable storage medium includes a program for a rapid deployment and operation and maintenance method for semiconductor visual inspection, and when the program is executed by a processor, the program implements the steps of the rapid deployment and operation and maintenance method for semiconductor visual inspection.
Compared with the prior art, the beneficial effects of this application are as follows:
1. the idea of low codes is adopted, debugging visualization and logic flow of a machine vision algorithm are rapidly realized in a configuration mode, rapid deployment is carried out in a solution package mode, and development period and maintenance cost of vision detection application are greatly reduced.
2. A large number of manually operated functional scenes in the field of semiconductor packaging and wafer detection are integrated in a set of software platform, so that the operation flow in the industry is standardized and forced verification is carried out, errors caused by manual operation are reduced, the problems caused by a plurality of misoperation are avoided, and the stability of the whole system in the production stage is enhanced.
3. The cloud computing is introduced into the field of semiconductor detection, temporary storage and version management are carried out on configuration parameters, training data and training models of a machine table, and safety and accuracy of data are guaranteed. The process of model training is monitored in real time by developing visual charts through a Web technology, so that the iteration of the model is more accurate and faster.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the description below are only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained according to the drawings without inventive labor.
Fig. 1 is a flowchart of a method for rapid deployment and operation and maintenance of semiconductor visual inspection according to the present application;
fig. 2 is a block diagram of a system for rapid deployment and operation and maintenance of semiconductor visual inspection according to the present application.
Detailed Description
In order to make the objects, features and advantages of the present application more obvious and understandable, the technical solutions in the embodiments of the present application are clearly and completely described, and it is obvious that the embodiments described below are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
Referring to fig. 1, fig. 1 is a flowchart of a method for rapid deployment and operation and maintenance of semiconductor visual inspection according to the present application.
The first aspect of the embodiments of the present application provides a method for rapid deployment and operation and maintenance of semiconductor visual inspection, including the following steps:
s102, analyzing the input parameters and the output results of the algorithm function according to the preset data type and format, and carrying out visualization processing on the input parameters and the output results of the algorithm function;
s104, inducing the logic relation of the algorithm function according to the deployment result of the visualization processing, and reconfiguring the logic relation of a plurality of algorithm functions in the target solution to generate a debugging algorithm library;
s106, carrying out visual positioning and cutting manufacturing on the image coordinate and the physical coordinate of the semiconductor chip by using a debugging algorithm library, and outputting debugging data in real time;
and S108, labeling, classifying and generating a training set which can be analyzed by an algorithm function on the cut small graph, and generating and iteratively debugging a model by using the training set.
It should be noted that, a tool for implementing a "solution configurator" in S102 adopts the idea of low code, and can quickly visualize input parameters and output results of an algorithm through configuration, so as to perform functions such as quick verification of the algorithm and construction of a solution in a production process.
In S104, one solution is composed of a plurality of standard algorithms, common logic relations among the algorithms are generalized, and the algorithms are conveniently and logically organized in a configurable mode. After a set of detection algorithm scheme and visual debugging scheme aiming at specific detection service are configured by the solution configurator, a series of parameters are generated and stored as Json format files. When the field is generated for deployment, debugging and running can be performed only by importing the solution package and the corresponding algorithm library through field software.
In addition to the need for visual debugging of machine vision algorithms in the industrial inspection field, there are many other steps in S106 and S108 that require human intervention to assist in achieving the final function. The embodiment of the application is based on the application of semiconductor packaging and wafer detection services, a set of processes is specified and integrated in a software platform. As the size of semiconductor chips is refined, there are often multiple dies in a camera field of view, and the whole wafer is diced to separate individual die segments, which are then processed based on the die segments. Both steps are debugged based on visualization, and the debugging result can be seen in real time.
According to the embodiment of the application, the analyzing the input parameters and the output results of the algorithm function, and the visualizing the input parameters and the output results of the algorithm function specifically comprises the following steps:
respectively registering and defining input parameters and output results of the algorithm function according to the data type and format;
analyzing the data type of the algorithm function input parameter to generate a corresponding UI input control, and loading the output result of the algorithm function by adopting an image interaction mode according to the UI input control;
and analyzing the data type of the algorithm function output result to generate a corresponding drawing element, and drawing on the image by using the drawing element.
It should be noted that the implementation here is through factory-plus-reflection mode in software programming. The custom data type is reflected to the UI control object and the drawing element object at runtime. The data types of the input parameters and the output results are respectively floating point numbers, arrays, coordinate points, image templates and the like; point sets, rectangle sets, contours, etc.
According to the embodiment of the application, inducing the logical relationship of the algorithm functions according to the deployment result of the visualization processing, and reconfiguring the logical relationship of the plurality of algorithm functions in the target solution specifically include:
acquiring the logic relationship between the algorithm functions at the boundary according to the region division condition in the deployment result, and judging whether the logic relationship of each algorithm function needs to be corrected;
if correction is needed, identifying the relation parameters among algorithm functions with problems in the target solution;
and modifying the relation format and the node attribute among the logic nodes in the identified relation parameters according to the region division condition in the deployment result.
According to the embodiment of the application, reconfiguring the logical relationship of a plurality of algorithm functions in the target solution, and generating the debugging algorithm library specifically comprises:
reconfiguring the logic relationship between the algorithm functions at the boundary, and generating the logic relationship of each algorithm function in the debugging algorithm library, wherein the logic relationship comprises dependence, sharing and parallelism;
when the logic relation is dependence, generating an output parameter of a next algorithm function according to the output parameter of the previous algorithm function;
when the logic relation is sharing, synchronizing the input parameter of the previous algorithm function with the input parameter of the next algorithm function;
and when the logical relation is in common, synchronizing the input parameters, operators and output parameters of the previous algorithm function with the next algorithm function.
It should be noted that the dependency relationship is that the input of one algorithm depends on the output of another algorithm. The sharing relationship is that two algorithms share the same parameter. A parallel relationship is the execution of one algorithm in parallel with the execution of another algorithm.
According to the embodiment of the application, the method further comprises the following steps:
monitoring the output debugging data in real time, and comparing the debugging data with preset debugging data, wherein the preset debugging data is generated according to a debugging algorithm library under a visual positioning and cutting manufacturing scheme;
and judging whether the debugging algorithm library needs to be adjusted or not according to the comparison result of the debugging data, and if the debugging algorithm library needs to be adjusted again, reconfiguring the logic relation of the algorithm function in the solution.
According to the embodiment of the application, after labeling and classifying the cut small graphs, generating a training set which can be analyzed by an algorithm function further comprises:
acquiring the cut picture label data and the cut picture data, and performing forced format check on the label data and the cut picture data according to the byte attributes of the data;
and if the format verification result does not pass, correcting and aligning the label data and the graphic data by using a preset format template.
According to the embodiment of the application, the generation and iterative debugging model by using the training set specifically comprises the following steps:
uploading the generated training set to a cloud platform through tool software, and calling a debugging algorithm library in the cutting process;
training the debugging algorithm library by using the training set generated in the previous step, and sequentially generating debugging models in each stage;
and (3) carrying out updating iteration in the generation process by downloading a suitable debugging model training in the cutting and manufacturing process.
According to the embodiment of the application, the method further comprises the following steps:
monitoring a training process in real time, and acquiring a progress parameter and an effect parameter in the training process;
and generating a training task monitoring visual chart in real time according to the progress parameters and the effect parameters, wherein feedback indexes of the monitoring visual chart comprise label detection, accuracy, progress and loss.
In another embodiment of the present application, generalizing the logical relationship of the algorithm function according to the deployment result of the visualization processing, and reconfiguring the logical relationship of the plurality of algorithm functions in the target solution specifically includes:
drawing a visualization deployment result obtained after a plurality of algorithm functions are fused in the target solution, and evaluating and grading the region division bias degree in the deployment result according to a preset threshold;
setting different weight coefficients for relation parameters of corresponding algorithm functions under different gears according to the gear grading condition;
and adjusting the relation format and the node attribute among the logic nodes in the relation parameters by using the weight coefficient.
In another embodiment of the present application, the method further comprises:
monitoring the iterative debugging models in real time, acquiring progress parameters and effect parameters in the training process, and calculating the accuracy and loss data of each debugging model;
and (4) calling the debugging model with the optimal accuracy and loss data, sending the debugging model to a terminal and automatically completing deployment.
It should be noted that, in the embodiment of the present application, the application of the machine vision algorithm may also be developed rapidly by using commercially available software, such as a general machine vision algorithm library: halcon, EVision, VisionMaster, etc. The image classification labeling function in the embodiment of the present application may also use labeling tool software, such as labelimg, labelme, and the like. In the embodiment of the application, management modes such as parameters, training sets and models of the machine through cloud services can be managed based on a file mode.
Referring to fig. 2, fig. 2 is a block diagram of a system for rapid deployment and operation and maintenance of semiconductor vision inspection according to the present invention.
A second aspect of the embodiment of the present application provides a semiconductor visual inspection rapid deployment and maintenance system, including a memory 21 and a processor 22, where the memory 21 includes a program of a method for semiconductor visual inspection rapid deployment and maintenance, and when the program is executed by the processor 22, the following steps are implemented:
analyzing the input parameters and the output results of the algorithm function according to the preset data type and format, and carrying out visualization processing on the input parameters and the output results of the algorithm function;
inducing the logical relationship of the algorithm functions according to the deployment result of the visualization processing, and reconfiguring the logical relationship of a plurality of algorithm functions in the target solution to generate a debugging algorithm library;
carrying out visual positioning and cutting manufacturing on the image coordinate and the physical coordinate of the semiconductor chip by using a debugging algorithm library, and outputting debugging data in real time;
and labeling the cut small pictures, classifying the small pictures to generate a training set which can be analyzed by an algorithm function, and generating and iteratively debugging a model by using the training set.
It should be noted that, the semiconductor visual inspection rapid deployment and operation and maintenance system according to the embodiment of the present application may be implemented by the following modules:
cloud platform functional module: the information of all the machines is uniformly managed by the cloud platform, and the operation condition of each machine can be monitored in real time. And the machine is taken as a basic node, and corresponding data is stored under the node. And information such as the solution, configuration parameters, operation logs and the like of each machine can be uploaded to the machine nodes corresponding to the cloud platform. When one deployed machine is copied to other machines, the artificial data copying step is avoided, and the safety and the correctness of the data are ensured. When multiple sets of configuration parameters are caused by multiple materials, the cloud platform is responsible for temporarily storing corresponding model data, and data loss and damage caused by abnormity are avoided.
A model training module: the classified picture data set can be uploaded to a cloud platform, training is started, breakpoint continuous transmission, data verification and data set version management are supported. And monitoring data such as training progress, training labels, training precision and training loss in real time in a visual chart mode. And downloading the trained model and deploying the model locally by one key.
According to the embodiment of the application, the analyzing the input parameters and the output results of the algorithm function, and the visualizing the input parameters and the output results of the algorithm function specifically comprises the following steps:
respectively registering and defining input parameters and output results of the algorithm function according to the data type and format;
analyzing the data type of the algorithm function input parameter to generate a corresponding UI input control, and loading the output result of the algorithm function by adopting an image interaction mode according to the UI input control;
and analyzing the data type of the algorithm function output result to generate a corresponding drawing element, and drawing on the image by using the drawing element.
According to the embodiment of the application, inducing the logical relationship of the algorithm functions according to the deployment result of the visualization processing, and reconfiguring the logical relationship of the plurality of algorithm functions in the target solution specifically include:
acquiring the logic relationship between the algorithm functions at the boundary according to the area division condition in the deployment result, and judging whether the logic relationship of each algorithm function needs to be corrected;
if correction is needed, identifying the relation parameters among algorithm functions with problems in the target solution;
and modifying the relation format and the node attribute among the logic nodes in the identified relation parameters according to the region division condition in the deployment result.
According to the embodiment of the application, reconfiguring the logical relationship of a plurality of algorithm functions in the target solution, and generating the debugging algorithm library specifically comprises:
reconfiguring the logic relationship between the algorithm functions at the boundary, and generating the logic relationship of each algorithm function in the debugging algorithm library, wherein the logic relationship comprises dependence, sharing and parallelism;
when the logic relation is dependence, generating an output parameter of a next algorithm function according to the output parameter of the previous algorithm function;
when the logic relation is sharing, synchronizing the input parameter of the previous algorithm function with the input parameter of the next algorithm function;
and when the logic relation is in common, synchronizing the input parameters, the operators and the output parameters of the previous algorithm function with the next algorithm function.
According to the embodiment of the application, the method further comprises the following steps:
monitoring the output debugging data in real time, and comparing the debugging data with preset debugging data, wherein the preset debugging data is generated according to a debugging algorithm library under a visual positioning and cutting manufacturing scheme;
and judging whether the debugging algorithm library needs to be adjusted or not according to the comparison result of the debugging data, and if the debugging algorithm library needs to be readjusted, reconfiguring the logic relation of the algorithm function in the solution.
According to the embodiment of the application, after labeling and classifying the cut small graphs, generating a training set which can be analyzed by an algorithm function further comprises:
acquiring the cut picture label data and the cut picture data, and performing forced format check on the label data and the cut picture data according to the byte attributes of the data;
and if the format verification result does not pass, correcting and aligning the label data and the graphic data by using a preset format template.
According to the embodiment of the application, the generation and iterative debugging model by using the training set specifically comprises the following steps:
uploading the generated training set to a cloud platform through tool software, and calling a debugging algorithm library in the cutting process;
training the debugging algorithm library by using the training set generated in the previous step, and sequentially generating debugging models in each stage;
and (3) carrying out updating iteration in the generation process by downloading a suitable debugging model training in the cutting and manufacturing process.
According to the embodiment of the application, the method further comprises the following steps:
monitoring a training process in real time, and acquiring a progress parameter and an effect parameter in the training process;
and generating a training task monitoring visual chart in real time according to the progress parameters and the effect parameters, wherein feedback indexes of the monitoring visual chart comprise label detection, accuracy, progress and loss.
In another embodiment of the present application, generalizing the logical relationship of the algorithm function according to the deployment result of the visualization processing, and reconfiguring the logical relationship of the plurality of algorithm functions in the target solution specifically includes:
drawing a visualization deployment result obtained after a plurality of algorithm functions are fused in the target solution, and evaluating and grading the region division bias degree in the deployment result according to a preset threshold;
setting different weight coefficients for relation parameters of corresponding algorithm functions under different gears according to the gear grading condition;
and adjusting the relation format and the node attribute among the logic nodes in the relation parameters by using the weight coefficient.
In another embodiment of the present application, the method further comprises:
monitoring the iterative debugging models in real time, acquiring progress parameters and effect parameters in the training process, and calculating the accuracy and loss data of each debugging model;
and (4) calling the debugging model with the optimal accuracy and loss data, sending the debugging model to the terminal and automatically finishing deployment.
A third aspect of the embodiments of the present application provides a computer-readable storage medium, where the computer-readable storage medium includes a program of a method for rapid deployment and operation and maintenance of semiconductor visual inspection, and when the program is executed by a processor, the steps of the method for rapid deployment and operation and maintenance of semiconductor visual inspection are implemented, which are specifically described in fig. 1 for description of the method steps, and are not repeated here.
In the several embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other manners. The above-described device embodiments are merely illustrative, for example, the division of the unit is only a logical functional division, and there may be other division ways in actual implementation, such as: multiple units or components may be combined, or may be integrated into another system, or some features may be omitted, or not implemented. In addition, the coupling, direct coupling or communication connection between the components shown or discussed may be through some interfaces, and the indirect coupling or communication connection between the devices or units may be electrical, mechanical or other forms.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units; can be located in one place or distributed on a plurality of network units; some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, all functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately used as one unit, or two or more units may be integrated into one unit; the integrated unit may be implemented in the form of hardware, or in the form of hardware plus a software functional unit.
Those of ordinary skill in the art will understand that: all or part of the steps for realizing the method embodiments can be completed by hardware related to program instructions, the program can be stored in a computer readable storage medium, and the program executes the steps comprising the method embodiments when executed; and the aforementioned storage medium includes: a mobile storage device, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
Alternatively, the integrated unit of the present invention may be stored in a computer-readable storage medium if it is implemented in the form of a software functional module and sold or used as a separate product. Based on such understanding, the technical solutions of the embodiments of the present invention may be essentially implemented or a part contributing to the prior art may be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the methods described in the embodiments of the present invention. And the aforementioned storage medium includes: a removable storage device, a ROM, a RAM, a magnetic or optical disk, or various other media capable of storing program code.
The above embodiments are only used to illustrate the technical solutions of the present application, and not to limit the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present application.
Claims (7)
1. A method for rapid deployment and operation and maintenance of semiconductor visual inspection is characterized by comprising the following steps:
analyzing the input parameters and the output results of the algorithm function according to the preset data type and format, and carrying out visualization processing on the input parameters and the output results of the algorithm function;
inducing the logical relationship of the algorithm functions according to the deployment result of the visualization processing, and reconfiguring the logical relationship of a plurality of algorithm functions in the target solution to generate a debugging algorithm library;
carrying out visual positioning and cutting manufacturing on the image coordinate and the physical coordinate of the semiconductor chip by using a debugging algorithm library, and outputting debugging data in real time;
labeling, classifying and generating a training set which can be analyzed by an algorithm function for the cut small pictures, and generating and iteratively debugging a model by using the training set;
analyzing the input parameters and the output results of the algorithm function, and performing visualization processing on the input parameters and the output results of the algorithm function specifically comprises the following steps:
respectively registering and defining input parameters and output results of the algorithm function according to the data type and format;
analyzing the data type of the algorithm function input parameter to generate a corresponding UI input control, and loading the output result of the algorithm function by adopting an image interaction mode according to the UI input control;
analyzing the data type of the algorithm function output result to generate a corresponding drawing element, and drawing on the image by using the drawing element;
the inducing the logical relationship of the algorithm functions according to the deployment result of the visualization processing, and the reconfiguring the logical relationship of the plurality of algorithm functions in the target solution are specifically as follows:
acquiring the logic relationship between the algorithm functions at the boundary according to the region division condition in the deployment result, and judging whether the logic relationship of each algorithm function needs to be corrected;
if the correction is needed, identifying the relation parameters among algorithm functions with problems in the target solution;
modifying the relation format and node attribute among the logic nodes in the identified relation parameters according to the region division condition in the deployment result;
the reconfiguring the logical relationship of the multiple algorithm functions in the target solution, and the generating the debugging algorithm library specifically comprises:
reconfiguring the logic relationship between the algorithm functions at the boundary, and generating the logic relationship of each algorithm function in the debugging algorithm library, wherein the logic relationship comprises dependence, sharing and parallelism;
when the logic relation is dependence, generating the output parameter of the next algorithm function according to the output parameter of the previous algorithm function;
when the logic relation is sharing, synchronizing the input parameter of the previous algorithm function with the input parameter of the next algorithm function;
and when the logic relation is in common, synchronizing the input parameters, the operators and the output parameters of the previous algorithm function with the next algorithm function.
2. The method for rapid deployment and operation and maintenance of semiconductor visual inspection of claim 1, further comprising:
monitoring the output debugging data in real time, and comparing the debugging data with preset debugging data, wherein the preset debugging data is generated according to a debugging algorithm library under a visual positioning and cutting manufacturing scheme;
and judging whether the debugging algorithm library needs to be adjusted or not according to the comparison result of the debugging data, and if the debugging algorithm library needs to be adjusted again, reconfiguring the logic relation of the algorithm function in the solution.
3. The method for rapid deployment and operation and maintenance of semiconductor visual inspection according to claim 1, wherein after the labeling and classification of the cut small graphs, the method further comprises before the generation of a training set that can be analyzed by an algorithm function:
acquiring the cut picture label data and the cut picture data, and performing forced format check on the label data and the cut picture data according to the byte attributes of the data;
and if the format verification result does not pass, correcting and aligning the label data and the graphic data by using a preset format template.
4. The method for rapid deployment and operation and maintenance of semiconductor visual inspection according to claim 1, wherein the generation and iterative debugging model using the training set specifically comprises:
uploading the generated training set to a cloud platform through tool software, and calling a debugging algorithm library in the cutting process;
training the debugging algorithm library by using the training set generated in the previous step, and sequentially generating debugging models in each stage;
and (3) carrying out updating iteration in the generation process by downloading a suitable debugging model training in the cutting and manufacturing process.
5. The method for rapid deployment and operation and maintenance of semiconductor visual inspection according to claim 1, further comprising:
monitoring a training process in real time, and acquiring a progress parameter and an effect parameter in the training process;
and generating a training task monitoring visual chart in real time according to the progress parameters and the effect parameters, wherein feedback indexes of the monitoring visual chart comprise label detection, accuracy, progress and loss.
6. A semiconductor visual inspection rapid deployment and operation and maintenance system is characterized by comprising a memory and a processor, wherein the memory comprises a semiconductor visual inspection rapid deployment and operation and maintenance method program, and when the program is executed by the processor, the steps of the semiconductor visual inspection rapid deployment and operation and maintenance method are realized according to any one of claims 1 to 5.
7. A computer-readable storage medium, wherein the computer-readable storage medium includes a program for a rapid deployment and operation method for semiconductor visual inspection, and when the program is executed by a processor, the program implements the steps of the rapid deployment and operation method for semiconductor visual inspection according to any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210168459.7A CN114237635B (en) | 2022-02-24 | 2022-02-24 | Method, system and storage medium for rapid deployment, operation and maintenance of semiconductor visual inspection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210168459.7A CN114237635B (en) | 2022-02-24 | 2022-02-24 | Method, system and storage medium for rapid deployment, operation and maintenance of semiconductor visual inspection |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114237635A CN114237635A (en) | 2022-03-25 |
| CN114237635B true CN114237635B (en) | 2022-07-15 |
Family
ID=80747927
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210168459.7A Active CN114237635B (en) | 2022-02-24 | 2022-02-24 | Method, system and storage medium for rapid deployment, operation and maintenance of semiconductor visual inspection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114237635B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019191810A1 (en) * | 2018-04-04 | 2019-10-10 | Basyrov Marat | Methods and systems for resolving user interface features, and related applications |
| CN110928528A (en) * | 2019-10-23 | 2020-03-27 | 深圳市华讯方舟太赫兹科技有限公司 | Development method of algorithm model, terminal device and computer storage medium |
| CN112463301A (en) * | 2020-11-30 | 2021-03-09 | 常州微亿智造科技有限公司 | Container-based model training test tuning and deployment method and device |
| CN112734703A (en) * | 2020-12-28 | 2021-04-30 | 佛山市南海区广工大数控装备协同创新研究院 | PCB defect optimization method by utilizing AI cloud collaborative detection |
| CN113537496A (en) * | 2021-06-07 | 2021-10-22 | 贵州优联博睿科技有限公司 | Deep learning model visual construction system and application and design method thereof |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11029359B2 (en) * | 2018-03-09 | 2021-06-08 | Pdf Solutions, Inc. | Failure detection and classsification using sensor data and/or measurement data |
| CN113903081A (en) * | 2021-09-29 | 2022-01-07 | 北京许继电气有限公司 | Visual identification artificial intelligence alarm method and device for images of hydraulic power plant |
-
2022
- 2022-02-24 CN CN202210168459.7A patent/CN114237635B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019191810A1 (en) * | 2018-04-04 | 2019-10-10 | Basyrov Marat | Methods and systems for resolving user interface features, and related applications |
| CN110928528A (en) * | 2019-10-23 | 2020-03-27 | 深圳市华讯方舟太赫兹科技有限公司 | Development method of algorithm model, terminal device and computer storage medium |
| CN112463301A (en) * | 2020-11-30 | 2021-03-09 | 常州微亿智造科技有限公司 | Container-based model training test tuning and deployment method and device |
| CN112734703A (en) * | 2020-12-28 | 2021-04-30 | 佛山市南海区广工大数控装备协同创新研究院 | PCB defect optimization method by utilizing AI cloud collaborative detection |
| CN113537496A (en) * | 2021-06-07 | 2021-10-22 | 贵州优联博睿科技有限公司 | Deep learning model visual construction system and application and design method thereof |
Non-Patent Citations (1)
| Title |
|---|
| 自成开发模式与统宜(VOPES)仿真可视平台;符长明等;《系统仿真学报》;20150808(第08期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114237635A (en) | 2022-03-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20240087104A1 (en) | Method for monitoring manufacture of assembly units | |
| JP2018206362A (en) | Process analyzer, process analysis method, and process analysis program | |
| WO2020127895A1 (en) | A method and system for generating an artificial intelligence model | |
| US20220414066A1 (en) | Data management system, management method, and storage medium | |
| CN113656864B (en) | Building construction method, system, computer device and storage medium based on BIM | |
| CN110069404A (en) | Code debugging method, apparatus, equipment and medium | |
| CN112527676A (en) | Model automation test method, device and storage medium | |
| CN114237635B (en) | Method, system and storage medium for rapid deployment, operation and maintenance of semiconductor visual inspection | |
| CN114819553A (en) | Dynamic configuration method, device, equipment and storage medium based on workflow engine | |
| CN106529070B (en) | Circuit board CAM data automatic processing method and system | |
| KR20200012664A (en) | Methods and systems for generating a wafer image from a computer-aided design file | |
| CN112633501A (en) | Development method and system of machine learning model framework based on containerization technology | |
| WO2021183382A1 (en) | Graph-based method for inductive bug localization | |
| CN111767218A (en) | Automatic testing method, equipment and storage medium for continuous integration | |
| CN116778036A (en) | Method and device for checking animation data, electronic equipment and medium | |
| CN108021353B (en) | Device, method, equipment and storage medium for generating description document of software product | |
| CN116089739A (en) | Message pushing method, device, electronic equipment and storage medium | |
| CN115829925A (en) | Appearance defect detection method and device, computer equipment and storage medium | |
| TWI801820B (en) | Systems and methods for manufacturing processes | |
| Gors et al. | An adaptable framework to provide AR-based work instructions and assembly state tracking using an ISA-95 ontology | |
| CN112748950A (en) | Software code examination method and device | |
| US11797646B2 (en) | Method, electronic device, and computer program product for standardizing image annotation | |
| CN112527631A (en) | bug positioning method, system, electronic equipment and storage medium | |
| EP2573694A1 (en) | Conversion method and system | |
| CN115293466B (en) | Cloud manufacturing method and system for manufacturing nodes through virtual intelligent configuration based on cloud |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |