CN113673577A - Industrial quality inspection model deployment method and platform - Google Patents
Industrial quality inspection model deployment method and platform Download PDFInfo
- Publication number
- CN113673577A CN113673577A CN202110848320.2A CN202110848320A CN113673577A CN 113673577 A CN113673577 A CN 113673577A CN 202110848320 A CN202110848320 A CN 202110848320A CN 113673577 A CN113673577 A CN 113673577A
- Authority
- CN
- China
- Prior art keywords
- quality inspection
- industrial quality
- model
- industrial
- 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.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
- G06F18/2178—Validation; Performance evaluation; Active pattern learning techniques based on feedback of a supervisor
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Quality & Reliability (AREA)
- General Factory Administration (AREA)
Abstract
The invention discloses an industrial quality inspection model deployment method and platform, relating to the technical field of industrial quality inspection, and the main technical scheme comprises the following steps: training an industrial quality inspection model to be trained currently by using a training sample set corresponding to an industrial quality inspection project, wherein the training samples in the training sample set are obtained on the basis of industrial product pictures acquired on a production line corresponding to the industrial quality inspection project; testing the trained industrial quality inspection model and displaying a test result; judging whether an adjustment parameter is received, wherein the adjustment parameter is a parameter which is provided by a user according to a test result and is used for adjusting the industrial quality inspection model; if the adjustment parameters are received, adjusting the parameters of the trained industrial quality inspection module by using the adjustment parameters, taking the industrial quality inspection model after the parameters are adjusted as the current industrial quality inspection model to be trained, and continuing training; and if the adjustment parameters are not received, deploying the trained industrial quality inspection model to quality inspection equipment corresponding to the industrial quality inspection project.
Description
Technical Field
The invention relates to the technical field of industrial quality inspection, in particular to an industrial quality inspection model deployment method and platform.
Background
With the rapid development of image sensors and vision algorithms, machine vision becomes an important boosting force for implementing automation strategies in factories, and vision-based automation schemes are widely applied to various links of industrial production. The vision-based automation scheme is accomplished through an industrial quality inspection model based on deep learning techniques.
From the technical point of view, the application form of visual detection based on the deep learning technology in the field of industrial quality inspection can be divided into three types, namely classification, detection and segmentation. Aiming at different forms, more than ten backbone network structures such as MobileNet and ResNet, target detection algorithms such as YOLOv3 and Faster R-CNN, segmentation algorithms such as DeepLabv3 and U-Net, and industrial quality inspection models corresponding to the structures and the algorithms are widely applied to industrial visual detection. But the deep learning model has more hyper-parameters and training parameters and poor performance interpretability due to the black box characteristic. In order to meet specific business requirements, training is generally performed in the form of an algorithm and a combination of various parameter settings, model parameters and training parameters are continuously adjusted, and quick trial and error and iteration are performed until performance requirements are met. In the training process, a user cannot usually intervene, the time is long, the development process is complicated, and the code adjustment is complex, so that the development and deployment efficiency of the industrial quality inspection model is low.
Disclosure of Invention
In view of this, the invention provides an industrial quality inspection model deployment method and platform, and mainly aims to improve the development and deployment efficiency of an industrial quality inspection model.
In a first aspect, the present invention provides a deployment method of an industrial quality inspection model, including:
training an industrial quality inspection model to be trained currently by using a training sample set corresponding to an industrial quality inspection project, wherein the training samples in the training sample set are obtained on the basis of industrial product pictures acquired on a production line corresponding to the industrial quality inspection project;
testing the trained industrial quality inspection model and displaying a test result;
judging whether an adjusting parameter is received, wherein the adjusting parameter is a parameter which is given by a user according to the test result and is used for adjusting the industrial quality inspection model;
if the adjustment parameters are received, the adjustment parameters are used for adjusting the parameters of the trained industrial quality inspection module, and the industrial quality inspection model after parameter adjustment is used as the current industrial quality inspection model to be trained for continuous training;
and if the adjustment parameters are not received, deploying the trained industrial quality inspection model to quality inspection equipment corresponding to the industrial quality inspection project.
In a second aspect, the present invention provides an industrial quality inspection model deployment platform, comprising:
the training unit is used for training an industrial quality inspection model to be trained currently by using a training sample set corresponding to an industrial quality inspection project, wherein the training samples in the training sample set are obtained on the basis of industrial product pictures acquired on a production line corresponding to the industrial quality inspection project;
the test unit is used for testing the trained industrial quality inspection model and displaying a test result;
the judging unit is used for judging whether an adjusting parameter is received or not, wherein the adjusting parameter is a parameter which is given by a user according to the test result and is used for adjusting the industrial quality inspection model;
the adjusting unit is used for adjusting the parameters of the trained industrial quality inspection module by using the adjusting parameters if the judging unit determines that the adjusting parameters are received, and taking the industrial quality inspection model after parameter adjustment as the current industrial quality inspection model to be trained to continue training;
and the deployment unit is used for deploying the trained industrial quality inspection model to the quality inspection equipment corresponding to the industrial quality inspection project if the judgment unit determines that the adjustment parameters are not received.
In a third aspect, the present invention provides a computer-readable storage medium, where the storage medium includes a stored program, and when the program runs, the apparatus on which the storage medium is located is controlled to execute the industrial quality inspection model deployment method according to the first aspect.
In a fourth aspect, the present invention provides a storage management apparatus, including:
a memory for storing a program;
a processor, coupled to the memory, for executing the program to perform the industrial quality control model deployment method of the first aspect.
By means of the technical scheme, when the industrial quality inspection model needs to be deployed, the industrial quality inspection model to be trained at present is trained by using the training sample set corresponding to the industrial quality inspection project. And then testing the trained industrial quality inspection model and displaying the test result. And if receiving an adjusting parameter which is given by a user according to the test result and used for adjusting the industrial quality inspection model, using the adjusting parameter to adjust the parameter of the trained industrial quality inspection module, taking the industrial quality inspection model after the parameter adjustment as the current industrial quality inspection model to be trained, and continuing training. And if the adjustment parameters are not received, deploying the trained industrial quality inspection model to quality inspection equipment corresponding to the industrial quality inspection project. Therefore, the scheme provided by the invention meets the full-process development of the industrial quality inspection model from model training to deployment, and the user can adjust the industrial quality inspection model according to the test result of the industrial quality inspection model, so that the development period of the industrial quality inspection model can be shortened, the development difficulty can be reduced, and the development and deployment efficiency of the industrial quality inspection model can be further improved.
The foregoing description is only an overview of the technical solutions of the present invention, and the embodiments of the present invention are described below in order to make the technical means of the present invention more clearly understood and to make the above and other objects, features, and advantages of the present invention more clearly understandable.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a flow chart illustrating a method for deploying an industrial quality inspection model according to an embodiment of the invention;
FIG. 2 is a flow diagram illustrating another industrial quality inspection model deployment method according to another embodiment of the invention;
FIG. 3 is a schematic structural diagram of an industrial quality inspection model deployment platform according to an embodiment of the present invention;
fig. 4 shows a schematic structural diagram of an industrial quality inspection model deployment platform according to another embodiment of the present invention.
Detailed Description
Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
As shown in fig. 1, an embodiment of the present invention provides an industrial quality inspection model deployment method, which mainly includes:
101. the method comprises the steps of training an industrial quality inspection model to be trained currently by using a training sample set corresponding to an industrial quality inspection project, wherein the training samples in the training sample set are obtained on the basis of industrial product pictures acquired on a production line corresponding to the industrial quality inspection project.
In practical applications, any industrial requirement that needs to detect defects based on a visual algorithm is defined as an industrial quality inspection project. Illustratively, the industrial quality inspection items may include, but are not limited to, the following items: the method comprises the steps of defect detection of metallurgical strip steel in a factory, defect detection of automobile parts, defect detection of household appliances, detection of glass panels and the like.
When an industrial quality inspection model needs to be deployed for the determined industrial quality inspection project, a training sample set corresponding to the industrial quality inspection project needs to be acquired so as to train the industrial quality inspection model by using the training sample set. The selection method of the training sample set may be: and displaying data details through a visual interface, and selecting a training sample set corresponding to the type to participate in training based on the type of an industrial quality inspection model required by the industrial quality inspection project input by the user.
The training samples in the training sample set are obtained based on the industrial product pictures acquired on the production line corresponding to the industrial quality inspection project, and the following description is given to a process of how to acquire the training sample set corresponding to the industrial quality inspection project, where the process includes the following steps one to three:
step one, controlling image acquisition equipment arranged on the production line, and acquiring industrial product pictures on the production line.
In practical application, the time for controlling the image acquisition equipment arranged on the production line corresponding to the industrial quality inspection project to acquire the pictures of the industrial products comprises the following two steps: the first is that when a model training instruction is received, the image acquisition equipment arranged on the production line is controlled to acquire the industrial product pictures on the production line, and the mode can acquire the industrial product pictures at any time according to the deployment requirement of the industrial quality inspection model. Secondly, when the specified time is up, the image acquisition equipment arranged on the production line is controlled to acquire the industrial product pictures on the production line, and the mode can acquire the latest industrial product pictures at regular time to form a training sample set.
The image acquisition equipment supports multiple brands and types of industrial cameras, such as Basler, Haikang and the like which are commonly used in the industry, and supports communication interface protocols such as GigE, USB3.0 and the like. The main functions of the image acquisition equipment are to support continuous acquisition, trigger acquisition, video acquisition and other data shooting models, support TCP/IP signal triggering, camera parameter setting, image resolution setting, region of interest (ROI) selection and storage, and image storage and reading.
The specific operation flow of the image acquisition equipment is as follows: firstly, initializing the system, enumerating all online camera devices based on the selected camera brand and displaying the online camera devices in a list, and selecting operation devices by pulling down the list to complete the initialization of the data acquisition system. Then, parameter setting is performed and saved, for example, parameters such as adjusting the resolution, exposure time, gain, and the like of the camera are performed, and a region of interest (ROI) is set and saved. And then, communication debugging is carried out, and based on the set communication IP address and the communication port, the communication debugging module is communicated with an upper computer system to carry out message sending, message reading and error information display. And finally, when the image acquisition equipment is required to acquire the pictures of the industrial products, the image acquisition equipment acquires and stores the pictures on the production line based on the selected data shooting mode and the set parameters.
And step two, providing the collected industrial product pictures for a labeling terminal for labeling by a labeling staff.
After the image acquisition equipment acquires the industrial product pictures on the production line, the acquired industrial product pictures are provided for the labeling terminal for labeling by a labeling person. During labeling, the labeling form generally includes the type of the labeled target object, the selection of the target position with a rectangular frame, the selection of the target position with a curve along an edge circle, and the like according to the different task types. The annotation files correspond to the image data one to one. It should be noted that, in order to meet the training requirements of different types of industrial quality inspection models, multiple forms of labeling may be adopted during labeling, and a corresponding training sample set may be formed for each type of industrial quality inspection model.
And step three, forming a training sample set according to the industrial product pictures marked by the marker.
Acquiring the industrial product picture marked by the marker, and performing at least one operation of the following operations: the image processing method comprises the steps of image frame translation processing, image cropping processing, image brightness changing processing, image adding noise processing, image rotation processing and image mirroring processing of an image. The operation forms a data set.
And taking the obtained data as a data set to carry out operations such as uploading, viewing, segmenting, deleting and the like to form a training sample set. Wherein, uploading: and uploading the data to a special deep learning training server according to the user setting mode and the input information. And (6) viewing: after the data are uploaded, the name, the data type and the data state of the data set are displayed, and the data set can be previewed and selected. Cutting: for model training, the dataset is partitioned proportionally (e.g., 7:2:1) into a training sample set, a validation sample set, and a test sample set. The training set is used for model training, the verification set is used for verifying the performance of the model in the training process, and the test set is used for performance testing after the model training is completed.
If the training samples in the training sample set are not in the format required by training, converting the training samples into the format required by training. It should be noted that one training sample set may be associated with a plurality of industrial quality inspection projects, a plurality of training tasks may be created under each industrial quality inspection project, and each task may obtain a corresponding industrial quality inspection model.
The model between industries to be trained at present is related to whether the current training is the first training, and at least the following situations exist:
first, the non-initial training, the current industrial quality inspection model to be trained is the industrial quality inspection model obtained after the last training and subjected to parameter adjustment according to the adjustment parameters for inputting.
And secondly, training for the first time, wherein the industrial quality inspection model to be trained at present is an industrial quality inspection model which is put into quality inspection equipment corresponding to an industrial quality inspection project and is used. This is to retrain the used industrial quality inspection model, thereby improving the industrial quality inspection effect.
Thirdly, training for the first time, when any parameter module is selected and model parameters and training parameters are input into the parameter module, adjusting the structure of the model corresponding to the parameter module based on the model parameters and adjusting parameters related to training in the model corresponding to the parameter module based on the training parameters, wherein different parameter modules correspond to different types of models; and determining the adjusted model as the current industrial quality inspection model to be trained during the first training.
In order to meet the deployment requirements of different industrial quality inspection models, a plurality of parameter modules are arranged, and different parameter modules correspond to different types of models. When a user needs to deploy a model of a required type, inputting model parameters and training parameters required by the user into a parameter module corresponding to the type of model.
In order to facilitate the user to input parameters, the parameter module is used for the user to input model parameters and training parameters in a visual interface mode, and the model corresponding to the parameter module is adjusted based on the parameters, so that the adjusted model is determined as the current industrial quality inspection model to be trained during the first training. The visual interface can input parameters through point-and-click operation. Some of the key parameters in the parameter module are as follows:
model selection and backbone network: according to the type of the training sample set or the type of the industrial quality control project, optional mainstream deep learning algorithms such as YOLOv3, Faster R-CNN, DeepLabv3, U-Net and the like are preset, and optional common backbone network structures such as MobileNet, ResNet, DarkNet, DenseNet and the like are preset.
Pre-training the model: presetting an optional ImageNet pre-training model or not using the pre-training model;
learning rate: presetting a learning rate value or user autonomous input based on an algorithm selected by a user and a network structure;
batch size: presetting a batch size numerical value or user independent input based on the number of the training display cards selected by the user;
data enhancement strategy: and presetting optional data enhancement strategies such as random brightness, random contrast, random horizontal turning, random vertical turning, random rotation, random saturation, random hue and the like.
And after the industrial quality inspection module to be trained at present is determined, training the industrial quality inspection model by using a training sample set corresponding to the industrial quality inspection project.
It should be noted that, in order to make the user clearly understand the performance condition of the current training model, when the current industrial quality inspection model to be trained is trained, the change condition of the designated index in the training process is output. The specified index change condition can be output by a visual interface. The specified metrics described herein may be user specified based on traffic conditions.
102. And testing the trained industrial quality inspection model and displaying the test result.
In order to know the visual recognition effect of the trained industrial quality inspection model, the trained industrial quality inspection model needs to be tested, and the test method can be two types as follows:
first, a test is performed using a predetermined set of test samples, which are collected in the same lot as the samples in the training sample set.
Secondly, in order to improve the test effect, the image acquisition equipment arranged on the production line is controlled to acquire the pictures of the industrial products on the production line. And providing the collected industrial product pictures for a labeling terminal for labeling by a labeling staff. And forming a test sample set according to the industrial product pictures marked by the marker. And testing the trained industrial quality inspection model by using the test sample set.
And after the test of the trained industrial quality inspection model is finished, displaying the test result in order to enable a user to know the visual identification effect of the trained industrial quality inspection model in time. The test result can be used for evaluating the performance indexes of the model performance for indexes such as precision, recall rate and the like.
The purpose of displaying the test results is to enable a user to judge whether the current industrial quality inspection model can be deployed and used, and if the user determines that the model cannot be deployed and used, the displayed test results provide powerful adjusting basis for the user to adjust the model parameters.
103. And judging whether an adjusting parameter is received, wherein the adjusting parameter is a parameter for adjusting the industrial quality inspection model given by the user according to the test result, and executing 104 or 105.
The purpose of judging whether the adjustment parameters are received is to determine whether the currently trained industrial quality inspection model is trained and can be deployed.
The adjustment parameters may be determined by the user based on the test result, that is, the user may determine the adjustment parameters by combining the test result based on the change of the specified index output by the user during the training process.
104. And if the adjustment parameters are received, performing parameter adjustment on the trained industrial quality inspection model by using the adjustment parameters, taking the industrial quality inspection model after parameter adjustment as the current industrial quality inspection model to be trained, continuing training, and executing the step 101.
If the adjustment parameters are received, it is indicated that the user determines that the performance of the currently trained industrial quality inspection model does not meet the requirement of visual recognition, and the adjustment parameters need to be used for parameter adjustment of the trained industrial quality inspection model, for example, the training parameters are adjusted. And after the parameters are adjusted, taking the industrial quality inspection model after the parameters are adjusted as the current industrial quality inspection model to be trained, and continuing training until the user is satisfied.
In order to meet the deployment requirements of different industrial quality inspection models, a plurality of parameter modules are arranged, and different parameter modules correspond to different types of models. When a user needs to adjust the model parameters of the trained industrial quality inspection module, the adjustment parameters required by the user are input into the parameter module corresponding to the model.
In order to facilitate the user to input the parameters, the parameter module is used for the user to input the adjustment parameters in a visual interface mode. The visual interface can input adjustment parameters through point-and-click operation.
105. And if the adjustment parameters are not received, deploying the trained industrial quality inspection model to quality inspection equipment corresponding to the industrial quality inspection project.
And if the adjustment parameters are not received, the user judges that the performance of the currently trained industrial quality inspection model meets the requirement of visual identification, and deploys the trained industrial quality inspection model to quality inspection equipment corresponding to the industrial quality inspection project so as to perform visual inspection by using the deployed industrial quality inspection model.
After the industrial quality inspection model is deployed in the quality inspection equipment corresponding to the industrial quality inspection project, in order to complete the quality inspection of the industrial products on the production line of the industrial quality inspection project, the following operations are required to be performed:
first, an available image capture device, such as an industrial camera, is selected.
And the communication interface is USB3.0 or GigE for communicating with hardware such as an industrial camera. And setting software/hardware parameters to start hardware equipment, completing initialization of the software and the hardware, and enabling the quality inspection equipment to be in an online state. Meanwhile, based on a Modbus communication protocol, the upper industrial personal computer communicates with the lower PLC, and the signal change condition of the external photoelectric sensor can be monitored.
And secondly, loading an industrial quality inspection model for visual classification, target detection and pixel segmentation according to the project type.
And after the model is loaded successfully, the quality inspection equipment is in a position state and waits for the controller to transmit signals. If the state of the external sensor changes, the signal is output to the PLC controller, and the controller transmits the signal to the industrial personal computer to trigger a detection process.
And thirdly, starting time detection by the quality inspection equipment, executing an industrial quality inspection model reasoning process, storing a detection result to the local for generating a detection report, and transmitting the detection report to a subordinate controller.
And fourthly, based on the detection result, the PLC starts logical operation according to an internal user program and controls the external pneumatic/electric mechanical device to execute corresponding instructions. For example, when the quality inspection equipment detects that a gear has a serious defect such as a crack, a broken tooth, or the like, the industrial robot is driven to move the gear to a reject box with the gripping jaw.
According to the industrial quality inspection model deployment method provided by the invention, when the industrial quality inspection model is required to be deployed, the current industrial quality inspection model to be trained is trained by using the training sample set corresponding to the industrial quality inspection project. And then testing the trained industrial quality inspection model and displaying the test result. And if receiving an adjusting parameter which is given by a user according to the test result and used for adjusting the industrial quality inspection model, using the adjusting parameter to adjust the parameter of the trained industrial quality inspection module, taking the industrial quality inspection model after the parameter adjustment as the current industrial quality inspection model to be trained, and continuing training. And if the adjustment parameters are not received, deploying the trained industrial quality inspection model to quality inspection equipment corresponding to the industrial quality inspection project. Therefore, the scheme provided by the embodiment of the invention meets the full-process development of the industrial quality inspection model from model training to deployment, and the user can adjust the industrial quality inspection model according to the test result of the industrial quality inspection model, so that the development period of the industrial quality inspection model can be shortened, the development difficulty can be reduced, and the development and deployment efficiency of the industrial quality inspection model can be further improved.
Further, according to the method shown in fig. 1, another embodiment of the present invention further provides an industrial quality inspection model deployment method, as shown in fig. 2, the method mainly includes:
201. and when a model training instruction is received or when the specified time is reached, controlling image acquisition equipment arranged on the production line to acquire the pictures of the industrial products on the production line.
202. And providing the collected industrial product pictures for a labeling terminal for labeling by a labeling staff.
203. And forming a training sample set according to the industrial product pictures marked by the marker.
204. When any parameter module is selected and model parameters and training parameters are input into the parameter module, the structure of the model corresponding to the parameter module is adjusted based on the model parameters and parameters related to training in the model corresponding to the parameter module are adjusted based on the training parameters, and different parameter modules correspond to different types of models.
205. And determining the adjusted model as the current industrial quality inspection model to be trained during the first training.
206. The method comprises the steps of training an industrial quality inspection model to be trained currently by using a training sample set corresponding to an industrial quality inspection project, wherein the training samples in the training sample set are obtained on the basis of industrial product pictures acquired on a production line corresponding to the industrial quality inspection project.
207. And when the current industrial quality inspection model to be trained is trained, outputting the designated index change condition in the training process.
208. And controlling image acquisition equipment arranged on the production line to acquire the pictures of the industrial products on the production line.
209. And providing the collected industrial product pictures for a labeling terminal for labeling by a labeling staff.
210. And forming a test sample set according to the industrial product pictures marked by the marker.
211. And testing the trained industrial quality inspection model by using the test sample set.
212. And displaying the test result.
213. And judging whether an adjusting parameter is received, wherein the adjusting parameter is a parameter for adjusting the industrial quality inspection model given by the user according to the test result, and executing 214 or 215.
214. If the adjustment parameters are received, the adjustment parameters are used to adjust the parameters of the trained industrial quality inspection module, the industrial quality inspection model with the adjusted parameters is used as the current industrial quality inspection model to be trained, the training is continued, and step 206 is executed.
215. And if the adjustment parameters are not received, deploying the trained industrial quality inspection model to quality inspection equipment corresponding to the industrial quality inspection project.
The industrial quality inspection model deployment method described by fig. 1 and 2 can be applied to an industrial quality inspection model deployment platform. Due to the fact that problems to be solved by different industrial quality inspection projects have commonality, the industrial quality inspection model deployment platform is established, data acquisition and management, model training and model deployment processes are integrated, the deep learning algorithm can be supported to be on-line and fall to a specific service scene, development efficiency is effectively improved, and development cost is reduced. The industrial quality inspection model deployment method can be used for deploying corresponding industrial quality inspection models no matter metallurgical strip steel detection, automobile part detection, household appliance defect detection or glass panel detection.
It should be noted that the configuration of the industrial quality control model deployment platform for implementing the above-described fig. 1 and 2 can be determined based on business requirements. Illustratively, the industrial quality inspection deployment platform is mainly implemented by languages such as C #, C + +, Python, Java, JavaScript, and HTML, and the supported deep learning framework is PaddlePaddle. The operation recommendation configuration requirements are as follows:
operating the system: windows 1064 bit
A processor: intel Core i7 and above
Memory: minimum 8GB, 16GB or more is recommended
Hard disk: 500GB of available disk space (including space for storing data files, model files, training process files)
A display card: to ensure that the GPU training is running properly, the following configuration is recommended: NVIDIAGeForce GTX 1060 or higher configuration graphics card, GPU power not less than 6.0 (please note: graphics card type meeting the software power requirement, please refer to NVIDIA official network)
And (3) accelerating operation: CUDA 11.2, cuDNN 8.1.0, if multiple GPUs are required for training, NCCL 2.8.3 needs to be configured
Further, according to the above method embodiment, another embodiment of the present invention further provides an industrial quality inspection model deployment platform, as shown in fig. 3, where the platform includes:
the training unit 31 is configured to train an industrial quality inspection model to be currently trained by using a training sample set corresponding to an industrial quality inspection project, where a training sample in the training sample set is obtained based on an industrial product picture acquired on a production line corresponding to the industrial quality inspection project;
the test unit 32 is used for testing the trained industrial quality inspection model and displaying a test result;
a judging unit 33, configured to judge whether an adjustment parameter is received, where the adjustment parameter is a parameter for adjusting the industrial quality inspection model, which is given by a user according to the test result;
the adjusting unit 34 is configured to, if the determining unit determines that the adjustment parameter is received, perform parameter adjustment on the trained industrial quality inspection module by using the adjustment parameter, and continue training by using the industrial quality inspection model after parameter adjustment as the current industrial quality inspection model to be trained;
and the deploying unit 35 is configured to deploy the trained industrial quality inspection model to the quality inspection equipment corresponding to the industrial quality inspection project if the determining unit determines that the adjustment parameter is not received.
According to the industrial quality inspection model deployment platform provided by the invention, when the industrial quality inspection model is required to be deployed, the current industrial quality inspection model to be trained is trained by using the training sample set corresponding to the industrial quality inspection project. And then testing the trained industrial quality inspection model and displaying the test result. And if receiving an adjusting parameter which is given by a user according to the test result and used for adjusting the industrial quality inspection model, using the adjusting parameter to adjust the parameter of the trained industrial quality inspection module, taking the industrial quality inspection model after the parameter adjustment as the current industrial quality inspection model to be trained, and continuing training. And if the adjustment parameters are not received, deploying the trained industrial quality inspection model to quality inspection equipment corresponding to the industrial quality inspection project. Therefore, the scheme provided by the embodiment of the invention meets the full-process development of the industrial quality inspection model from model training to deployment, and the user can adjust the industrial quality inspection model according to the test result of the industrial quality inspection model, so that the development period of the industrial quality inspection model can be shortened, the development difficulty can be reduced, and the development and deployment efficiency of the industrial quality inspection model can be further improved.
Optionally, as shown in fig. 4, the platform further includes:
a first determining unit 36, configured to, when any parameter module is selected and a model parameter and a training parameter are input into the parameter module, adjust a structure of a model corresponding to the parameter module based on the model parameter and adjust a parameter related to training in the model corresponding to the parameter module based on the training parameter, where different parameter modules correspond to different types of models; and determining the adjusted model as the current industrial quality inspection model to be trained during the first training.
Optionally, as shown in fig. 4, the parameter module related to the first determining unit 36 is in the form of a visual interface for the user to input the model parameters and the training parameters.
Optionally, as shown in fig. 4, the platform further includes:
and a second determining unit 37, configured to determine the first-trained industrial quality inspection model to be currently trained as the industrial quality inspection model used by the quality inspection equipment corresponding to the industrial quality inspection item.
Optionally, as shown in fig. 4, the platform further includes:
and the output unit 38 is configured to output the designated index change condition in the training process when the training unit 31 trains the industrial quality inspection model to be trained currently.
Optionally, as shown in fig. 4, the determining unit 33 is configured to determine that the adjustment parameter is received when a parameter module corresponding to the industrial quality inspection model is inputted with a parameter, where different parameter modules correspond to different types of models.
Optionally, as shown in fig. 4, the testing unit 32 is configured to control an image capturing device disposed on the production line to capture an image of an industrial product on the production line; providing the collected industrial product pictures to a labeling terminal for labeling by a labeling person; forming a test sample set according to the industrial product pictures marked by the marker; and testing the trained industrial quality inspection model by using the test sample set.
Optionally, as shown in fig. 4, the platform further includes:
a generating unit 39, configured to control an image acquisition device arranged on the production line to acquire an image of an industrial product on the production line when a model training instruction is received or when a specified time is reached; providing the collected industrial product pictures to a labeling terminal for labeling by a labeling person; and forming a training sample set according to the industrial product pictures marked by the marker.
In the industrial quality inspection model deployment platform provided in the embodiment of the present invention, for details of methods used in the operation process of each functional module, reference may be made to the corresponding methods in the method embodiments of fig. 1 to fig. 2, which are not described herein again.
Further, according to the above embodiment, another embodiment of the present invention further provides a computer-readable storage medium, where the storage medium includes a stored program, and when the program runs, the apparatus on which the storage medium is located is controlled to execute the industrial quality inspection model deployment method described in fig. 1 or fig. 2.
Further, according to the above embodiment, another embodiment of the present invention provides a storage management apparatus, including:
a memory for storing a program;
a processor, coupled to the memory, for executing the program to perform the industrial quality inspection model deployment method of fig. 1 or fig. 2.
In the foregoing embodiments, the descriptions of the respective embodiments have respective emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to related descriptions of other embodiments.
It will be appreciated that the relevant features of the method and apparatus described above are referred to one another. In addition, "first", "second", and the like in the above embodiments are for distinguishing the embodiments, and do not represent merits of the embodiments.
It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
The algorithms and displays presented herein are not inherently related to any particular computer, virtual machine, or other apparatus. Various general purpose systems may also be used with the teachings herein. The required structure for constructing such a system will be apparent from the description above. Moreover, the present invention is not directed to any particular programming language. It is appreciated that a variety of programming languages may be used to implement the teachings of the present invention as described herein, and any descriptions of specific languages are provided above to disclose the best mode of the invention.
In the description provided herein, numerous specific details are set forth. It is understood, however, that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures and techniques have not been shown in detail in order not to obscure an understanding of this description.
Similarly, it should be appreciated that in the foregoing description of exemplary embodiments of the invention, various features of the invention are sometimes grouped together in a single embodiment, figure, or description thereof for the purpose of streamlining the disclosure and aiding in the understanding of one or more of the various inventive aspects. However, the disclosed method should not be interpreted as reflecting an intention that: that the invention as claimed requires more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive aspects lie in less than all features of a single foregoing disclosed embodiment. Thus, the claims following the detailed description are hereby expressly incorporated into this detailed description, with each claim standing on its own as a separate embodiment of this invention.
Those skilled in the art will appreciate that the modules in the device in an embodiment may be adaptively changed and disposed in one or more devices different from the embodiment. The modules or units or components of the embodiments may be combined into one module or unit or component, and furthermore they may be divided into a plurality of sub-modules or sub-units or sub-components. All of the features disclosed in this specification (including any accompanying claims, abstract and drawings), and all of the processes or elements of any method or apparatus so disclosed, may be combined in any combination, except combinations where at least some of such features and/or processes or elements are mutually exclusive. Each feature disclosed in this specification (including any accompanying claims, abstract and drawings) may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise.
Furthermore, those skilled in the art will appreciate that while some embodiments described herein include some features included in other embodiments, rather than other features, combinations of features of different embodiments are meant to be within the scope of the invention and form different embodiments. For example, in the following claims, any of the claimed embodiments may be used in any combination.
The various component embodiments of the invention may be implemented in hardware, or in software modules running on one or more processors, or in a combination thereof. Those skilled in the art will appreciate that a microprocessor or Digital Signal Processor (DSP) may be used in practice to implement some or all of the functions of some or all of the components of the method, apparatus and framework for operation of a deep neural network model in accordance with embodiments of the present invention. The present invention may also be embodied as apparatus or device programs (e.g., computer programs and computer program products) for performing a portion or all of the methods described herein. Such programs implementing the present invention may be stored on computer-readable media or may be in the form of one or more signals. Such a signal may be downloaded from an internet website or provided on a carrier signal or in any other form.
It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and that those skilled in the art will be able to design alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in a claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention may be implemented by means of hardware comprising several distinct elements, and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by one and the same item of hardware. The usage of the words first, second and third, etcetera do not indicate any ordering. These words may be interpreted as names.
Claims (10)
1. A method for deploying an industrial quality inspection model, the method comprising:
training an industrial quality inspection model to be trained currently by using a training sample set corresponding to an industrial quality inspection project, wherein the training samples in the training sample set are obtained on the basis of industrial product pictures acquired on a production line corresponding to the industrial quality inspection project;
testing the trained industrial quality inspection model and displaying a test result;
judging whether an adjusting parameter is received, wherein the adjusting parameter is a parameter which is given by a user according to the test result and is used for adjusting the industrial quality inspection model;
if the adjustment parameters are received, the adjustment parameters are used for adjusting the parameters of the trained industrial quality inspection module, and the industrial quality inspection model after parameter adjustment is used as the current industrial quality inspection model to be trained for continuous training;
and if the adjustment parameters are not received, deploying the trained industrial quality inspection model to quality inspection equipment corresponding to the industrial quality inspection project.
2. The method of claim 1, further comprising:
when any parameter module is selected and model parameters and training parameters are input into the parameter module, adjusting the structure of a model corresponding to the parameter module based on the model parameters and adjusting parameters related to training in the model corresponding to the parameter module based on the training parameters, wherein different parameter modules correspond to different types of models;
and determining the adjusted model as the current industrial quality inspection model to be trained during the first training.
3. The method of claim 2, wherein the parameter module is in the form of a visual interface for a user to input model parameters and training parameters.
4. The method of claim 1, further comprising:
and the first-trained industrial quality inspection model to be trained at present is an industrial quality inspection model which is put into quality inspection equipment corresponding to the industrial quality inspection project and is used.
5. The method according to any one of claims 1-4, further comprising:
and when the current industrial quality inspection model to be trained is trained, outputting the designated index change condition in the training process.
6. The method of any of claims 1-4, wherein determining whether the adjustment parameter is received comprises:
and when the parameter module corresponding to the industrial quality inspection model is input with parameters, determining to receive the adjustment parameters, wherein different parameter modules correspond to different types of models.
7. The method of any one of claims 1-4, wherein testing the trained industrial quality control model comprises:
controlling image acquisition equipment arranged on the production line to acquire industrial product pictures on the production line;
providing the collected industrial product pictures to a labeling terminal for labeling by a labeling person;
forming a test sample set according to the industrial product pictures marked by the marker;
testing the trained industrial quality inspection model by using the test sample set;
and/or the presence of a gas in the gas,
the method further comprises the following steps:
when a model training instruction is received or when the specified time is reached, controlling image acquisition equipment arranged on the production line to acquire industrial product pictures on the production line;
providing the collected industrial product pictures to a labeling terminal for labeling by a labeling person;
and forming a training sample set according to the industrial product pictures marked by the marker.
8. An industrial quality inspection model deployment platform, the platform comprising:
the training unit is used for training an industrial quality inspection model to be trained currently by using a training sample set corresponding to an industrial quality inspection project, wherein the training samples in the training sample set are obtained on the basis of industrial product pictures acquired on a production line corresponding to the industrial quality inspection project;
the test unit is used for testing the trained industrial quality inspection model and displaying a test result;
the judging unit is used for judging whether an adjusting parameter is received or not, wherein the adjusting parameter is a parameter which is given by a user according to the test result and is used for adjusting the industrial quality inspection model;
the adjusting unit is used for adjusting the parameters of the trained industrial quality inspection module by using the adjusting parameters if the judging unit determines that the adjusting parameters are received, and taking the industrial quality inspection model after parameter adjustment as the current industrial quality inspection model to be trained to continue training;
and the deployment unit is used for deploying the trained industrial quality inspection model to the quality inspection equipment corresponding to the industrial quality inspection project if the judgment unit determines that the adjustment parameters are not received.
9. A computer-readable storage medium, wherein the storage medium includes a stored program, and wherein when the program runs, the apparatus on which the storage medium is located is controlled to execute the industrial quality inspection model deployment method according to any one of claims 1 to 7.
10. A storage management apparatus, characterized in that the storage management apparatus comprises:
a memory for storing a program;
a processor, coupled to the memory, for executing the program to perform the industrial quality control model deployment method of any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110848320.2A CN113673577A (en) | 2021-07-27 | 2021-07-27 | Industrial quality inspection model deployment method and platform |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110848320.2A CN113673577A (en) | 2021-07-27 | 2021-07-27 | Industrial quality inspection model deployment method and platform |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113673577A true CN113673577A (en) | 2021-11-19 |
Family
ID=78540289
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110848320.2A Pending CN113673577A (en) | 2021-07-27 | 2021-07-27 | Industrial quality inspection model deployment method and platform |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113673577A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111310934A (en) * | 2020-02-14 | 2020-06-19 | 北京百度网讯科技有限公司 | Model generation method and device, electronic equipment and storage medium |
| CN112700446A (en) * | 2021-03-23 | 2021-04-23 | 常州微亿智造科技有限公司 | Algorithm model training method and device for industrial quality inspection |
| CN112836724A (en) * | 2021-01-08 | 2021-05-25 | 重庆创通联智物联网有限公司 | Object defect recognition model training method and device, electronic equipment and storage medium |
| CN112966772A (en) * | 2021-03-23 | 2021-06-15 | 之江实验室 | Multi-person online image semi-automatic labeling method and system |
-
2021
- 2021-07-27 CN CN202110848320.2A patent/CN113673577A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111310934A (en) * | 2020-02-14 | 2020-06-19 | 北京百度网讯科技有限公司 | Model generation method and device, electronic equipment and storage medium |
| CN112836724A (en) * | 2021-01-08 | 2021-05-25 | 重庆创通联智物联网有限公司 | Object defect recognition model training method and device, electronic equipment and storage medium |
| CN112700446A (en) * | 2021-03-23 | 2021-04-23 | 常州微亿智造科技有限公司 | Algorithm model training method and device for industrial quality inspection |
| CN112966772A (en) * | 2021-03-23 | 2021-06-15 | 之江实验室 | Multi-person online image semi-automatic labeling method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP7004145B2 (en) | Defect inspection equipment, defect inspection methods, and their programs | |
| EP2776896B1 (en) | Method and system for position control based on automated defect detection feedback | |
| US10824906B2 (en) | Image processing device, non-transitory computer readable storage medium, and image processing system | |
| CN110678877A (en) | Visual localization in images using weakly supervised neural networks | |
| CN109060817B (en) | Artificial intelligence reinspection system and method thereof | |
| CN112424827A (en) | Method for creating tagged image data, method for inspecting tagged image data, program for creating tagged image data, device for inspecting tagged image data, and storage medium | |
| JP7435303B2 (en) | Inspection device, unit selection device, inspection method, and inspection program | |
| CN112164048B (en) | Magnetic shoe surface defect automatic detection method and device based on deep learning | |
| CN114026596A (en) | Method, mobile terminal device and system for evaluating laser cutting edge | |
| CN111768398A (en) | Defect detection method and device for photovoltaic module, storage medium and computing equipment | |
| JP2007184723A (en) | Inspection system and inspection method | |
| CN115423755A (en) | Wafer micromachining structure defect detection method and device, equipment and medium thereof | |
| Variz et al. | Machine learning applied to an intelligent and adaptive robotic inspection station | |
| KR102524151B1 (en) | Labeling learning method of artificial intelligence machine for smart products inspection | |
| US20220405586A1 (en) | Model generation apparatus, estimation apparatus, model generation method, and computer-readable storage medium storing a model generation program | |
| CN113673577A (en) | Industrial quality inspection model deployment method and platform | |
| WO2021030322A1 (en) | System and method of object detection using ai deep learning models | |
| CN111028250A (en) | Real-time intelligent cloth inspecting method and system | |
| CN113469991B (en) | Visual online detection method for laser welding spot of lithium battery tab | |
| CN111935480B (en) | Detection method for image acquisition device and related device | |
| CN115100110A (en) | Defect detection method, device and equipment for polarized lens and readable storage medium | |
| CN114972500A (en) | Checking method, marking method, system, device, terminal, equipment and medium | |
| Dighvijay et al. | A Faster R-CNN implementation of presence inspection for parts on industrial produce | |
| CN110252686A (en) | The determination method and device of remote controler quality | |
| CN117409332B (en) | Long wood shaving appearance data detection system and method based on big data processing |
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 |