CN111860421A - C-Mn steel structure identification method based on Mask R-CNN network - Google Patents
C-Mn steel structure identification method based on Mask R-CNN network Download PDFInfo
- Publication number
- CN111860421A CN111860421A CN202010749365.XA CN202010749365A CN111860421A CN 111860421 A CN111860421 A CN 111860421A CN 202010749365 A CN202010749365 A CN 202010749365A CN 111860421 A CN111860421 A CN 111860421A
- Authority
- CN
- China
- Prior art keywords
- mask
- image data
- cnn network
- steel
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/69—Microscopic objects, e.g. biological cells or cellular parts
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10056—Microscopic image
- G06T2207/10061—Microscopic image from scanning electron microscope
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30136—Metal
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30204—Marker
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Multimedia (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a C-Mn steel structure identification method based on a Mask R-CNN network, and relates to the technical field of steel. The C-Mn steel structure identification method based on the Mask R-CNN network separates the training process from the identification process, does not depend on the high performance of a processor, and is suitable for more software and hardware platforms. The C-Mn steel structure identification application program based on the Mask R-CNN network can still quickly identify the microstructure of the C-Mn steel on the premise of ensuring high identification rate; by applying the trained Mask R-CNN network model, the speed and the accuracy of the C-Mn steel microstructure identification are greatly improved while artificial subjective factors are reduced.
Description
Technical Field
The invention relates to the technical field of steel, in particular to a C-Mn steel structure identification method based on a Mask R-CNN network.
Background
With the continuous development of artificial intelligence, deep learning is widely applied in various fields. In the industrial field, the application of intelligent robots, mechanical arms and other equipment developed by an artificial intelligence technology greatly improves the efficiency of industrial production; in the medical field, the artificial intelligence technology is used for assisting clinical diagnosis, so that misjudgment caused by subjective factors is effectively avoided; in the field of security protection, technologies such as face recognition and fingerprint recognition also bring convenience to the life of people while ensuring safety.
At present, artificial intelligence technology and application are in high-speed development, but the research results of automatic identification of C-Mn steel microstructure based on artificial intelligence method are few. The traditional identification method mainly depends on manual identification by people, large judgment errors are caused by artificial subjective factors, and compared with the artificial intelligence method, the method has the advantages of rapidness, accuracy and the like in the identification field.
Disclosure of Invention
Aiming at the problems of the existing identification method in the aspects of speed and accuracy, the invention provides a C-Mn steel microstructure identification method based on a Mask R-CNN network.
The technical scheme adopted by the invention is as follows:
a C-Mn steel structure identification method based on Mask R-CNN network comprises the following steps:
step 1, acquiring microstructure image data of C-Mn steel by using microscope equipment, and establishing a microstructure image data set of the C-Mn steel;
the C-Mn steel microstructure image data are divided into training set image data, verification set image data and test set image data; and carrying out secondary arrangement on the image data, and dividing the arranged image data into a training set, a verification set and a test set according to the ratio of 4:1: 1.
Step 2, marking the acquired C-Mn steel microstructure training set image data and the acquired verification set image data by using a data marking tool, and performing format conversion on the marked image data;
step 3, building a Mask R-CNN network model under a computer platform, and training the Mask R-CNN network model by using an image data set to obtain training parameters of the model;
step 3.1, normalizing the image data acquired by the network model into a set size through size normalization processing;
step 3.2, establishing a back bone network of a Mask R-CNN network model under a deep learning framework;
step 3.3, training a Mask R-CNN network model: and (3) inputting the image training set and the verification set processed in the step (2) into the constructed Mask R-CNN network model, and continuously optimizing the structural parameters of the model in the training process to finally obtain the trained model weight.
And 4, establishing an application program of the C-Mn steel structure identification method based on the Mask R-CNN network, designing a graphic user interaction interface, realizing that a user selects a C-Mn steel microstructure image to be identified, and automatically displaying an image category segmentation chart, a phase fraction and a grain size.
The application program packs the program runtime library by using a dynamic compiling method so as to realize cross-platform operation.
Adopt the produced beneficial effect of above-mentioned technical scheme to lie in:
the invention provides a C-Mn steel structure identification method based on a Mask R-CNN network, which separates training and identification processes, does not depend on the high performance of a processor, and is suitable for more software and hardware platforms. The C-Mn steel structure identification application program based on the Mask R-CNN network can still quickly identify the microstructure of the C-Mn steel on the premise of ensuring high identification rate.
The C-Mn steel structure identification method based on the Mask R-CNN network has expandability, and a Mask R-CNN network model can be quickly obtained by modifying a network structure, training parameters and retraining.
Drawings
FIG. 1 is a flow chart of a C-Mn steel structure identification method based on a Mask R-CNN network;
FIG. 2 is a flowchart illustrating an application program identifying an image to be detected and outputting an identification result according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of an image to be detected according to an embodiment of the present invention;
FIG. 4 is a ferrite, pearlite and boundary segmentation diagram of C-Mn steel based on Mask R-CNN network according to an embodiment of the present invention.
Detailed Description
The following detailed description of embodiments of the invention refers to the accompanying drawings.
A C-Mn steel structure identification method based on Mask R-CNN network is shown in figure 1 and comprises the following steps:
step 1, collecting microstructure image data of C-Mn steel by utilizing equipment such as a metallographic microscope, a scanning electron microscope, a transmission electron microscope and the like, and establishing a microstructure image data set of the C-Mn steel;
the C-Mn steel microstructure image data are divided into training set image data and testing set image data; and performing secondary sorting on large-scale training data, screening 1000 pieces of training set image data in the embodiment, and selecting one fifth of the training set image data as verification set image data.
Step 2, labeling the acquired C-Mn steel microstructure image data in the training set and the verification set by using a labelme data labeling tool, and converting json files generated after labeling into tfrechrd file formats;
step 3, building a Mask R-CNN network model under a computer platform, and training the Mask R-CNN network model by using an image data set to obtain training parameters of the model;
the experimental platform used by the Mask R-CNN network model in the embodiment is as follows:
hardware environment: the system takes a desktop computer which is loaded with an Intel (R) core (TM) i 5-75003.40 GHz processor and an internal memory 8G as a basic operating platform, and is configured with an NVIDIA GTX1060 video card with a video memory of 6G as an accelerating device of CUDA.
Software environment: the operating system used by the experimental platform was Windows 10. The installed support libraries mainly comprise Python3.6, OpenCV-Python and Pythrch-1.
Step 3.1, the acquired image data are all normalized to a size of 512 × 512 by size normalization processing.
And 3.2, establishing a back bone network of a Mask R-CNN network model under a deep learning framework, namely ResNet-FPN-50, wherein the network model is provided with 50 convolutional layers which are marked as conv1, conv2_ x, conv3_ x, conv4_ x and conv5_ x. The convolution kernel size of conv1 is 7 × 7, and the number is 64; the conv2_ x convolution layer has 3 modes with convolution kernel sizes of 1 × 1, 3 × 3 and 1 × 1, and the number of each convolution kernel is 64, 64 and 256 respectively; the conv3_ x convolution layer has 4 modes with convolution kernel sizes of 1 × 1, 3 × 3 and 1 × 1, and the number of each convolution kernel is 128, 128 and 512; the conv4_ x convolution layer has 6 modes with convolution kernel sizes of 1 × 1, 3 × 3 and 1 × 1, and the number of each convolution kernel is 256, 256 and 1024 respectively; the conv5_ x convolutional layer has 3 modulo of convolutional kernel size 1 × 1, 3 × 3, 1 × 1, and the number of each convolutional kernel is 512, 2048 respectively.
Step 3.3, training a Mask R-CNN network model: inputting the processed image training set and the verification set into the constructed Mask R-CNN network model, and continuously optimizing the structural parameters of the model in the training process to finally obtain the trained model weight. The training parameters are set as: iteration is carried out 50k times, mini-batch is 6, and the initial learning rate is set to be 0.01.
And 4, establishing an application program of the C-Mn steel structure identification method based on the Mask R-CNN network, building a graphical user interface by utilizing Qt, and realizing that a user selects a C-Mn steel microstructure image to be identified and automatically displays an image category segmentation graph, a phase proportion score and a grain size as shown in figure 2.
The application program packs the program runtime library by using a dynamic compiling method so as to realize cross-platform operation.
An image of the microstructure to be examined in the test image dataset is shown in fig. 3. Fig. 4 is a phase-segmented image output from the network model, in which a light gray region represents a pearlite structure, a white region represents a ferrite structure, and black represents grain boundaries. It can be seen that the segmentation effect is significant and there is no phenomenon of erroneous segmentation.
Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; such modifications and substitutions do not depart from the spirit of the corresponding technical solutions and scope of the present invention as defined in the appended claims.
Claims (3)
1. A C-Mn steel structure identification method based on Mask R-CNN network is characterized by comprising the following steps:
step 1, acquiring microstructure image data of C-Mn steel by using microscope equipment, and establishing a microstructure image data set of the C-Mn steel;
the C-Mn steel microstructure image data are divided into training set image data, verification set image data and test set image data; performing secondary arrangement on the image data, and dividing the arranged image data into a training set, a verification set and a test set according to the ratio of 4:1: 1;
step 2, marking the acquired C-Mn steel microstructure training set image data and the acquired verification set image data by using a data marking tool, and performing format conversion on the marked image data;
step 3, building a Mask R-CNN network model under a computer platform, and training the Mask R-CNN network model by using an image data set to obtain training parameters of the model;
and 4, establishing an application program of the C-Mn steel structure identification method based on the Mask R-CNN network, designing a graphic user interaction interface, realizing that a user selects a C-Mn steel microstructure image to be identified, and automatically displaying an image category segmentation chart, a phase fraction and a grain size.
2. The C-Mn steel structure identification method based on the Mask R-CNN network according to claim 1, wherein the step 3 specifically comprises:
step 3.1, normalizing the image data acquired by the network model into a set size through size normalization processing;
step 3.2, establishing a back bone network of a Mask R-CNN network model under a deep learning framework;
step 3.3, training a Mask R-CNN network model: and (3) inputting the image training set and the verification set processed in the step (2) into the constructed Mask R-CNN network model, and continuously optimizing the structural parameters of the model in the training process to finally obtain the trained model weight.
3. The C-Mn steel organization identification method based on the Mask R-CNN network as claimed in claim 1, wherein the application program in step 4 uses a dynamic compiling method to pack a program runtime library to realize cross-platform operation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010749365.XA CN111860421A (en) | 2020-07-30 | 2020-07-30 | C-Mn steel structure identification method based on Mask R-CNN network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010749365.XA CN111860421A (en) | 2020-07-30 | 2020-07-30 | C-Mn steel structure identification method based on Mask R-CNN network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111860421A true CN111860421A (en) | 2020-10-30 |
Family
ID=72945426
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010749365.XA Pending CN111860421A (en) | 2020-07-30 | 2020-07-30 | C-Mn steel structure identification method based on Mask R-CNN network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111860421A (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109948712A (en) * | 2019-03-20 | 2019-06-28 | 天津工业大学 | A kind of nanoparticle size measurement method based on improved Mask R-CNN |
| CN110197170A (en) * | 2019-06-05 | 2019-09-03 | 北京科技大学 | Coil of strip scroll defects detection recognition methods based on target detection |
| CN110619355A (en) * | 2019-08-28 | 2019-12-27 | 武汉科技大学 | Automatic steel material microstructure identification method based on deep learning |
-
2020
- 2020-07-30 CN CN202010749365.XA patent/CN111860421A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109948712A (en) * | 2019-03-20 | 2019-06-28 | 天津工业大学 | A kind of nanoparticle size measurement method based on improved Mask R-CNN |
| CN110197170A (en) * | 2019-06-05 | 2019-09-03 | 北京科技大学 | Coil of strip scroll defects detection recognition methods based on target detection |
| CN110619355A (en) * | 2019-08-28 | 2019-12-27 | 武汉科技大学 | Automatic steel material microstructure identification method based on deep learning |
Non-Patent Citations (2)
| Title |
|---|
| ADACHI等: "Mircrostructure Recognition by Deep Learning", 《TETSU TO HAGANE-JOURNAL OF THE IRON AND STEEL INSTITUTE OF JAPAN》 * |
| 班晓娟等: "深度学习在材料显微图像分析中的应用与挑战", 《材料科学与工艺》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112286824B (en) | Test case generation method and system based on binary search iteration and electronic equipment | |
| CN116361801B (en) | Malicious software detection method and system based on semantic information of application program interface | |
| CN111694937A (en) | Interviewing method and device based on artificial intelligence, computer equipment and storage medium | |
| CN113448843B (en) | Image recognition software test data enhancement method and device based on defect analysis | |
| CN111401156A (en) | Image identification method based on Gabor convolution neural network | |
| CN114065199B (en) | Cross-platform malicious code detection method and system | |
| CN112364893B (en) | Semi-supervised zero-sample image classification method based on data enhancement | |
| EP4105872A3 (en) | Data processing method and apparatus | |
| CN115659175A (en) | Multi-mode data analysis method, device and medium for micro-service resources | |
| CN113591093A (en) | Industrial software vulnerability detection method based on self-attention mechanism | |
| JP2022541199A (en) | A system and method for inserting data into a structured database based on image representations of data tables. | |
| Zhang | Application of artificial intelligence recognition technology in digital image processing | |
| CN114639152A (en) | Multi-modal voice interaction method, device, equipment and medium based on face recognition | |
| CN117521063A (en) | Malicious software detection method and device based on residual neural network and combined with transfer learning | |
| Susilo et al. | Accuracy enhancement in early detection of breast cancer on mammogram images with convolutional neural network (CNN) methods using data augmentation and transfer learning | |
| CN110347579B (en) | Deep learning test case selection method based on neuron output behavior pattern | |
| CN116778164A (en) | Semantic segmentation method for improving deep V < 3+ > network based on multi-scale structure | |
| CN111860421A (en) | C-Mn steel structure identification method based on Mask R-CNN network | |
| CN115373982A (en) | Test report analysis method, device, equipment and medium based on artificial intelligence | |
| Feng et al. | A knowledge-integrated deep learning framework for cellular image analysis in parasite microbiology | |
| CN109146058B (en) | Convolutional neural network with transform invariant capability and consistent expression | |
| CN113673476A (en) | Face recognition model training method and device, storage medium and electronic equipment | |
| Yan et al. | Yolo-pd: Abnormal signal detection in gas pipelines based on improved yolov7 | |
| Yu et al. | Automatic bug inference via deep image understanding | |
| Kazangirler et al. | UIBee: An improved deep instance segmentation and classification of UI elements in wireframes |
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 |