CN117788952A - Hot-rolled middle plate warping identification device and method - Google Patents
Hot-rolled middle plate warping identification device and method Download PDFInfo
- Publication number
- CN117788952A CN117788952A CN202410055782.2A CN202410055782A CN117788952A CN 117788952 A CN117788952 A CN 117788952A CN 202410055782 A CN202410055782 A CN 202410055782A CN 117788952 A CN117788952 A CN 117788952A
- Authority
- CN
- China
- Prior art keywords
- image
- contour data
- data
- middle plate
- model
- 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
- 238000000034 method Methods 0.000 title claims abstract description 38
- 238000001931 thermography Methods 0.000 claims abstract description 25
- 238000003708 edge detection Methods 0.000 claims abstract description 14
- 238000012549 training Methods 0.000 claims description 24
- 230000008569 process Effects 0.000 claims description 15
- 238000005096 rolling process Methods 0.000 claims description 8
- 229910000831 Steel Inorganic materials 0.000 claims description 7
- 238000012937 correction Methods 0.000 claims description 7
- 239000010959 steel Substances 0.000 claims description 7
- 238000007781 pre-processing Methods 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 abstract description 19
- 239000003595 mist Substances 0.000 abstract description 17
- 238000005098 hot rolling Methods 0.000 abstract description 15
- 238000010801 machine learning Methods 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 abstract description 2
- 238000011065 in-situ storage Methods 0.000 abstract description 2
- 238000003672 processing method Methods 0.000 abstract description 2
- 238000004519 manufacturing process Methods 0.000 description 7
- 238000013135 deep learning Methods 0.000 description 5
- 238000013527 convolutional neural network Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000008439 repair process Effects 0.000 description 2
- 239000000779 smoke Substances 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000007477 logistic regression Methods 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a hot-rolled middle plate warping identification device and a method, comprising the following steps: s1, acquiring a side image of a middle plate to be detected, and acquiring contour data through an edge detection algorithm; s2, importing the data into a pre-trained recognition model, classifying the contour data by the recognition model, and outputting the result as head warping and normal. The invention discloses an image processing method for hot rolling of a middle plate based on machine learning, which is characterized in that a model for judging whether the middle plate generates warping is trained through supervision learning, then a middle plate side image acquired in real time is identified, and the warping degree can be effectively identified through the side plate image. Furthermore, it is considered that the high in-situ heat during the hot rolling process generates a water mist environment, which results in that the camera cannot acquire a complete profile image. Based on the characteristic of high temperature of the plate in the hot rolling process, the invention further obtains a thermal imaging diagram for auxiliary identification.
Description
Technical Field
The invention discloses a hot-rolled medium plate warping identification device and method, and relates to the technical field of hot-rolled medium plate warping identification.
Background
If excessive warping of the steel sheet occurs during the middle plate rolling process, difficulties may be encountered when attempting to send the warped steel sheet to the next process or machine. This is because the warpage makes the shape of the steel sheet no longer flat, so that it cannot be effectively contacted with the rollers or jigs of the next process or machine, and thus cannot smoothly enter the next process. The seesaw is generally caused by non-uniformity of the material itself, stress accumulation during rolling, temperature variation or machine problems. Therefore, in the middle plate rolling process, the flatness of the steel plate needs to be monitored and controlled to ensure the production efficiency and the product quality.
In the conventional means, whether the head is warped is generally observed in a manual monitoring mode, and if the head is warped, the production line is closed.
Summary of the invention
The invention aims to provide a hot-rolled medium plate warping identification device and a hot-rolled medium plate warping identification method. The problem that manual monitoring is needed for middle plate rolling in the prior art is solved, and automatic detection for the middle plate rolling is achieved.
In order to achieve the technical purpose and the technical effect, the invention is realized by the following technical scheme:
a hot rolling medium plate warping identification device and method comprises the following steps:
s1, acquiring a side image of a middle plate to be detected, and acquiring contour data through an edge detection algorithm;
s2, importing the data into a pre-trained recognition model, classifying the contour data by the recognition model, and outputting the result as head warping and normal.
Further, the method comprises the steps that the obtained image comprises a conventional image obtained by shooting and an infrared image obtained by thermal imaging, and the infrared image and the conventional image are stacked and then are used as multichannel input;
and after inputting, the conventional image and the infrared image are respectively based on the edge algorithm to obtain an edge contour, and the two contours are fused and then input into the recognition model.
Further, the side image of the middle plate to be measured is input into a pre-trained image restoration model, and the image restoration model comprises: a generator and a real sample acquirer;
the generator predicts the contour data of the middle plate to be tested based on the conventional image processing obtained by the pre-training model;
the real sample acquirer acquires the contour data of the infrared image, identifies whether the predicted contour data accords with expectations, outputs a predicted image if the predicted contour data accords with expectations, and corrects the predicted contour data based on the infrared image contour data if the predicted contour data does not accord with expectations.
Further, the real samples in the real sample acquirer are replaced by empty sample sets, and the thermal imaging profile data acquired in real time are replaced by single empty samples one by one to be used as real samples for judging the corresponding predicted profile data.
Further, the correction is: and calculating the difference between the contour data predicted by the generator and the real sample, and correcting the contour data predicted by the generator based on the calculated difference.
Further, the recognition model training process includes:
(1) The image acquisition and preprocessing, namely inputting a sample set, wherein the sample set comprises side images in the middle plate rolling process, and the images have clear outlines which are not blocked;
(2) Edge detection, namely detecting the edge of the steel plate by using an edge detection algorithm to obtain a binary image, wherein white pixels represent the detected edge;
(3) The label data is that for each image, whether the image has excessive tilting is manually judged, and the judgment result is used as a label;
(4) Feature extraction, namely extracting features of the binary image obtained by edge detection;
(5) Training the model, namely performing supervision training by using the features and the corresponding labels to obtain the identification model.
Further, the image restoration model is a model obtained based on countermeasure network training.
Another object of the present invention is to disclose a hot rolled middle plate warping recognition device, for implementing the foregoing method, including an image acquisition device, where the image acquisition device acquires a conventional image of a side surface of the middle plate and a thermal imaging image of the side surface of the middle plate;
an image recognition apparatus, the image recognition apparatus comprising:
the contour acquisition module is used for acquiring contour data of the conventional image and the thermal imaging image;
the generation module is loaded with the image restoration model and generates complete predicted contour data based on the contour of the conventional image;
the identification module is used for identifying whether the predicted contour data is matched with the contour data of the thermal imaging graph or not, and if the predicted contour data is not matched with the contour data of the thermal imaging graph, the contour data is corrected;
the identification module is loaded with the identification module, and the processed profile data is input into the identification model to judge whether the profile data is in a head-warping state or not;
the image recognition device inputs the profile data of the conventional image and the thermal imaging image acquired in real time into the generation module and the identification module respectively, and the identification module outputs the result to the identification module to finish the head-warping identification.
The beneficial effects are that:
the invention discloses an image processing method for hot rolling of a middle plate based on machine learning, which is characterized in that a model for judging whether the middle plate generates warping is trained through supervision learning, then a middle plate side image acquired in real time is identified, and the warping degree can be effectively identified through the side plate image. Furthermore, it is considered that the high in-situ heat during the hot rolling process generates a water mist environment, which results in that the camera cannot acquire a complete profile image. Based on the characteristic of high temperature of the plate in the hot rolling process, the invention further obtains a thermal imaging diagram for auxiliary identification.
Specifically, based on the repair characteristics of the countermeasure network, the model is trained as well, the incomplete conventional image contour is predicted and completed, and then the real sample is replaced by the acquired thermally imaged image for verification to improve the accuracy of the test (the thermally imaged image is the actual real sample in the present case).
Of course, it is not necessary for any one product to practice the invention to achieve all of the advantages set forth above at the same time.
Drawings
FIG. 1 is a flow chart of a method implementation in an embodiment of the invention.
FIG. 2 is a flow chart of a method implementation in an embodiment of the invention.
Detailed Description
In order to more clearly illustrate the technical scheme of the present invention, the present invention will be described in detail with reference to examples.
Examples
The embodiment discloses a hot-rolled middle plate warping identification method, which comprises the following steps:
s1, acquiring a side image of the middle plate to be detected, and acquiring contour data through an edge detection algorithm.
It is considered that the phenomenon of warping of the middle plate during hot rolling is mainly represented on the side profile of the middle plate. The edge detection algorithm can accurately acquire the contour information of the side surface of the middle plate, and basic data is provided for subsequent head-warping identification.
S2, importing the data into a pre-trained recognition model, classifying the contour data by the recognition model, and outputting the result as head warping and normal.
In step S2, the acquired contour data is imported into a pre-trained recognition model for classification. The automatic identification is carried out on the warping phenomenon by utilizing a machine learning technology, so that the problems of errors and low efficiency of manual detection are avoided. The pre-trained recognition model can effectively recognize the warped head and the normal middle plate by learning a large amount of training data, thereby ensuring the quality and the efficiency of the production process.
In the production and manufacturing process, the image acquisition device acquires side images of the middle plate in real time in the hot rolling process to judge. After the image is obtained, the image is processed through an edge detection algorithm, and contour data of the side face of the middle plate are obtained. This step is performed on a high performance computing device and a large amount of image data can be processed in real time. The acquired profile data is then imported into a pre-trained recognition model for classification. The recognition model is obtained by training on a large amount of training data, and can accurately recognize the warped head and the normal middle plate. The identification result is fed back to the production line in real time, and if the head is identified, the production line can be immediately adjusted to ensure the quality of the product.
In fact, during hot rolling of the middle plate, a large amount of mist is generated in the hot rolling area, and a good result cannot be obtained by a conventional image acquisition mode, so that the system is difficult to obtain correct middle plate warping data. In view of this problem, in the present embodiment, acquiring an image includes capturing an acquired regular image and an infrared image acquired by thermal imaging, and stacking the infrared image and the regular image as a multichannel input. By stacking the infrared image and the conventional image, multichannel input data can be obtained, and the identification accuracy is further improved.
And after inputting, the conventional image and the infrared image are respectively based on the edge algorithm to obtain an edge contour, and the two contours are fused and then input into the recognition model. In this embodiment, a mode of acquiring both a normal image and an infrared image is adopted. During hot rolling, a large amount of water mist is generated due to the high temperature and the presence of water. The influence of the water mist on the conventional optical image is large, so that the image is blurred, and the contour information of the middle plate is difficult to obtain clearly. However, the infrared image is not affected by the water mist, and clear contour information can be obtained in the water mist environment. Therefore, by stacking the infrared image and the regular image, an accurate profile image can be acquired in a water mist environment.
Based on the foregoing, in some embodiments, in order to obtain the infrared image and the conventional image, and further obtain a more real image result, an image restoration model is further adopted in this embodiment. The model is obtained based on the countermeasure network training, and can repair the conventional image affected by the water mist. And taking the contour data obtained by thermal imaging as a real sample, and correcting the predicted contour data. Thus, even in a water mist environment, an accurate contour image can be obtained through the image restoration model. The method not only improves the accuracy of the warped head identification, but also enhances the stability and reliability of the system in a complex environment.
Specifically, the side image of the middle plate to be tested is input into a pre-trained image restoration model, and the image restoration model comprises: the generator and the real sample acquirer.
And the generator predicts the contour data of the middle plate to be tested based on the conventional image processing obtained by the pre-training model.
In this embodiment, the main task of the generator is to receive the input side image of the middle plate to be measured, and process the obtained conventional image through the pre-training model to predict the contour data of the middle plate to be measured. First, the generator will receive side images of the middle plate under test, which images may be affected by the water mist environment during hot rolling, and the profile information may not be clear. The generator will then take these image data as inputs into a pre-trained image restoration model. The pre-training model may include various deep learning structures, such as a Convolutional Neural Network (CNN) structure, a Long Short Term Memory (LSTM) structure, etc., which can perform deep learning on input image data to predict contour data of the middle plate to be measured.
In the prediction process, the generator simulates the generation process of the outline of the middle plate to be detected, extracts deep features in image data through a multi-layer neural network structure of a pre-training model, and then generates predicted outline data of the middle plate to be detected through operations such as deconvolution. These prediction data are intended to simulate the actual midplane contour, but may be different from the actual situation due to the prediction error of the model and the noise effect of the input image. Thus, the predicted profile data generated by the generator will be used as part of the countermeasure network for authentication and correction by subsequent real sample acquisitors.
The real sample acquirer acquires the contour data of the infrared image, identifies whether the predicted contour data accords with expectations, outputs a predicted image if the predicted contour data accords with expectations, and corrects the predicted contour data based on the infrared image contour data if the predicted contour data does not accord with expectations.
In this embodiment, the main task of the real sample acquirer is to acquire the contour data of the infrared image and, based on these data, to discriminate whether the contour data predicted by the generator meets the expectations. Outputting a predicted image if it meets the expectation; if not, the predicted contour data is modified based on the infrared image contour data.
First, the real sample acquirer acquires contour data of the infrared image, which can provide clear contour information in a water mist environment, and is regarded as a real sample. The true sample acquirer then uses these true samples to identify the contour data predicted by the generator. This discrimination process is typically achieved by calculating a loss function between the predicted contour data and the real samples, such as mean square error loss, cross entropy loss, etc. And if the value of the loss function is smaller than a preset threshold value, the predicted contour data is considered to be in line with the expectations, and a predicted image is output. If the value of the loss function is greater than a preset threshold, the predicted profile data is considered to be not in line with expectations and needs to be corrected.
In the correction process, the real sample acquirer corrects the predicted contour data based on the infrared image contour data. This correction is typically accomplished by optimization algorithms such as back propagation and gradient descent, and the parameters of the generator will be updated in the direction of gradient descent to minimize the value of the loss function, thereby making the predicted profile data more nearly real. After a plurality of iterations, the predicted contour data gradually trend towards the real contour data, so that the conventional image affected by the water mist is repaired.
By the method, even in a water mist environment, an accurate contour image can be obtained through the image restoration model. The identification accuracy of the middle plate to be detected in a complex environment is greatly improved, and meanwhile, the stability and reliability of the system are enhanced. In a specific implementation process, the generator and the real sample acquirer participate in the training process at the same time, and the performance of the image restoration model is gradually improved through continuous countermeasure and learning, so that the conventional image influenced by water mist can be accurately identified and restored in various complex environments.
In an implementation case, the real samples in the real sample acquirer are replaced by empty sample sets, and the thermal imaging profile data acquired in real time are replaced by single empty samples one by one to be used as real samples for evaluating the corresponding predicted profile data.
Specifically, the correction is: and calculating the difference between the contour data predicted by the generator and the real sample, and correcting the contour data predicted by the generator based on the calculated difference.
As described in the previous embodiments, identifying the sheath head requires a pre-training model. In a specific embodiment, the recognition model training process includes:
(1) And (3) image acquisition and preprocessing, namely acquiring images in the middle plate rolling process by using a camera. Some pre-processing of the image may be required, such as resizing the image, contrast enhancement, denoising, etc., as desired.
(2) Edge detection-edge detection algorithms (e.g., sobel, canny, etc.) are used to detect edges of the steel sheet. This will result in a binary image in which white pixels represent the detected edges.
(3) And (3) label data, namely manually judging whether each image has excessive tilting or not, and taking the judgment result as a label. This will be used for subsequent supervised learning.
(4) And extracting the characteristics of the binary image obtained by edge detection. These features may be global features of the image (e.g., histograms of the image) or local features (e.g., texture, shape, etc. of the image). It is noted that which features are selected is critical to the final classification result.
(5) Training models-training a supervised learning model using the features and corresponding labels. This model may be a logistic regression, support vector machine, decision tree, neural network, etc. The optimal model and parameters are selected using a cross-validation approach.
(6) Model testing and evaluation the performance of the model was tested on a separate test set. According to the requirement, the indexes such as the accuracy, the precision, the recall rate and the like of the model can be calculated.
The image restoration model is a model obtained based on countermeasure network training.
Based on the method, the invention also discloses a hot-rolled medium plate warping identification device for realizing the method, which comprises an image acquisition device and an image identification device, wherein the image identification device further comprises a contour acquisition module, a generation module, an identification module and an identification module. Through this hot rolling medium plate wander recognition device, can realize under water smoke environment, discern and restore the conventional image that receives the water smoke influence through acquireing infrared image and conventional image to whether the medium plate is in the wander state effectively.
Specifically, the image acquisition device is mainly responsible for acquiring conventional images and thermal imaging images of the side of the middle plate. The conventional image is mainly used for acquiring basic contour information of the middle plate, but may be affected to some extent due to a water mist environment possibly existing in the hot rolling process, so that the contour information is not clear enough. The thermal imaging image can provide clear contour information in a water mist environment, and is regarded as a real sample for subsequent contour data identification and correction.
The image recognition device is a core part for realizing the head-warping recognition and comprises four modules: the device comprises a contour acquisition module, a generation module, an identification module and an identification module. The contour acquisition module is used for acquiring contour data of the conventional image and the thermal imaging image. The generating module is responsible for receiving an input image, performing deep learning through a pre-trained image restoration model, and predicting outline data of the middle plate to be detected. These predicted profile data are then fed into an authentication module for authentication. The discrimination module discriminates whether the contour data predicted by the generation module matches the expectation or not based on the real sample obtained from the thermal imaging map. If not, the discrimination module corrects the predicted contour data based on the infrared image contour data. The corrected contour data is sent to the identification module for judging the warping state.
The recognition module mainly comprises a pre-trained recognition model which receives the processed contour data and then judges whether the middle plate is in a head-tilting state or not according to the data. This recognition process may be based on a deep learning algorithm, such as Convolutional Neural Network (CNN), long-term memory (LSTM), etc., which may perform deep learning on the input profile data to achieve accurate determination of the warp state.
In the actual operation process, the image recognition device can acquire outline data of the conventional image and the thermal imaging image in real time, and input the data into the generation module and the identification module respectively. The generation module generates predicted contour data based on contour data of the regular image, which is fed to the authentication module for authentication. The identification module corrects the predicted contour data according to the contour data of the thermal imaging image, and then sends the corrected contour data into the identification module to judge the head-warping state. The identification module judges whether the middle plate is in a head-tilting state according to the data, and outputs a judging result.
The above is only an example portion of the application and is not intended to limit the application in any way. Any simple modification, equivalent variation and modification of the above embodiments still fall within the scope of the protection of the technical solution of this application.
Claims (8)
1. The hot-rolled medium plate warping identification method is characterized by comprising the following steps of:
s1, acquiring a side image of a middle plate to be detected, and acquiring contour data through an edge detection algorithm;
s2, importing the data into a pre-trained recognition model, classifying the contour data by the recognition model, and outputting the result as head warping and normal.
2. The hot-rolled medium plate warping recognition method according to claim 1, wherein the acquiring the image includes capturing an acquired regular image and an infrared image acquired by thermal imaging, and stacking the infrared image and the regular image to be used as a multichannel input;
and after inputting, the conventional image and the infrared image are respectively based on the edge algorithm to obtain an edge contour, and the two contours are fused and then input into the recognition model.
3. The hot rolled medium plate warp recognition method according to claim 2, wherein the side image of the medium plate to be measured is input into a pre-trained image restoration model, the image restoration model comprising: a generator and a real sample acquirer;
the generator predicts the contour data of the middle plate to be tested based on the conventional image processing obtained by the pre-training model;
the real sample acquirer acquires the contour data of the infrared image, identifies whether the predicted contour data accords with expectations, outputs a predicted image if the predicted contour data accords with expectations, and corrects the predicted contour data based on the infrared image contour data if the predicted contour data does not accord with expectations.
4. The hot-rolled medium plate warping recognition method according to claim 3, wherein the real samples in the real sample acquirer are replaced by empty sample sets, and the thermal imaging profile data acquired in real time are replaced by single empty samples one by one to be used as real samples for judging the corresponding predicted profile data.
5. The hot rolled medium plate warp recognition method according to claim 3, wherein the correction is: and calculating the difference between the contour data predicted by the generator and the real sample, and correcting the contour data predicted by the generator based on the calculated difference.
6. The hot rolled medium plate warp recognition method of claim 1, wherein the recognition model training process comprises:
(1) The image acquisition and preprocessing, namely inputting a sample set, wherein the sample set comprises side images in the middle plate rolling process, and the images have clear outlines which are not blocked;
(2) Edge detection, namely detecting the edge of the steel plate by using an edge detection algorithm to obtain a binary image, wherein white pixels represent the detected edge;
(3) The label data is that for each image, whether the image has excessive tilting is manually judged, and the judgment result is used as a label;
(4) Feature extraction, namely extracting features of the binary image obtained by edge detection;
(5) Training the model, namely performing supervision training by using the features and the corresponding labels to obtain the identification model.
7. A hot rolled mid-plate warp recognition method as claimed in claim 3, wherein the image restoration model is a model obtained based on countermeasure network training.
8. A hot-rolled middle plate warping recognition device for realizing the method of claim 3, which is characterized by comprising an image acquisition device, wherein the image acquisition device acquires a conventional image of the side surface of the middle plate and a thermal imaging image of the side surface of the middle plate;
an image recognition apparatus, the image recognition apparatus comprising:
the contour acquisition module is used for acquiring contour data of the conventional image and the thermal imaging image;
the generation module is loaded with the image restoration model and generates complete predicted contour data based on the contour of the conventional image;
the identification module is used for identifying whether the predicted contour data is matched with the contour data of the thermal imaging graph or not, and if the predicted contour data is not matched with the contour data of the thermal imaging graph, the contour data is corrected;
the identification module is loaded with the identification module, and the processed profile data is input into the identification model to judge whether the profile data is in a head-warping state or not;
the image recognition device inputs the profile data of the conventional image and the thermal imaging image acquired in real time into the generation module and the identification module respectively, and the identification module outputs the result to the identification module to finish the head-warping identification.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410055782.2A CN117788952A (en) | 2024-01-15 | 2024-01-15 | Hot-rolled middle plate warping identification device and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410055782.2A CN117788952A (en) | 2024-01-15 | 2024-01-15 | Hot-rolled middle plate warping identification device and method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117788952A true CN117788952A (en) | 2024-03-29 |
Family
ID=90391019
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410055782.2A Pending CN117788952A (en) | 2024-01-15 | 2024-01-15 | Hot-rolled middle plate warping identification device and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117788952A (en) |
-
2024
- 2024-01-15 CN CN202410055782.2A patent/CN117788952A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111179251B (en) | Defect detection system and method based on twin neural network and by utilizing template comparison | |
| US10964004B2 (en) | Automated optical inspection method using deep learning and apparatus, computer program for performing the method, computer-readable storage medium storing the computer program, and deep learning system thereof | |
| CN107437245B (en) | High-speed railway contact net fault diagnosis method based on deep convolutional neural network | |
| CN110648305B (en) | Industrial image detection method, system and computer readable recording medium | |
| KR100598381B1 (en) | in-line typed apparatus for auto wafer-defect classification and control method thereof | |
| US20200193219A1 (en) | Discrimination device and machine learning method | |
| CN115953373B (en) | Glass defect detection method, device, electronic equipment and storage medium | |
| CN114494780A (en) | Semi-supervised industrial defect detection method and system based on feature comparison | |
| CN117333467B (en) | Image processing-based glass bottle body flaw identification and detection method and system | |
| CN112529109A (en) | Unsupervised multi-model-based anomaly detection method and system | |
| KR102666787B1 (en) | Method, apparatus and program for inspecting defect | |
| CN116485779A (en) | Adaptive wafer defect detection method and device, electronic equipment and storage medium | |
| CN113554645B (en) | Industrial anomaly detection method and device based on WGAN | |
| CN117455917B (en) | Establishment of false alarm library of etched lead frame and false alarm on-line judging and screening method | |
| CN114549414A (en) | Abnormal change detection method and system for track data | |
| JP2022015575A (en) | Anomaly detection system, learning apparatus, anomaly detection program, learning program, anomaly detection method, and learning method | |
| CN117788952A (en) | Hot-rolled middle plate warping identification device and method | |
| US20240282098A1 (en) | Inspection method, classification method, management method, steel material production method, learning model generation method, learning model, inspection device, and steel material production equipment | |
| TWI647658B (en) | Device, system and method for automatically identifying image features | |
| KR20230036650A (en) | Defect detection method and system based on image patch | |
| CN113267506A (en) | Wood board AI visual defect detection device, method, equipment and medium | |
| CN107123105A (en) | Images match defect inspection method based on FAST algorithms | |
| CN112508946A (en) | Cable tunnel abnormity detection method based on antagonistic neural network | |
| CN117474916B (en) | Image detection method, electronic equipment and storage medium | |
| CN118096747B (en) | Automatic PCBA (printed circuit board assembly) board detection method and system based on deep learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication |