CN114372955A - Casting defect X-ray diagram automatic identification method based on improved neural network - Google Patents
Casting defect X-ray diagram automatic identification method based on improved neural network Download PDFInfo
- Publication number
- CN114372955A CN114372955A CN202111474014.3A CN202111474014A CN114372955A CN 114372955 A CN114372955 A CN 114372955A CN 202111474014 A CN202111474014 A CN 202111474014A CN 114372955 A CN114372955 A CN 114372955A
- Authority
- CN
- China
- Prior art keywords
- improved
- model
- defect
- neural network
- yolov5
- 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
- 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
- G06T7/0008—Industrial image inspection checking presence/absence
-
- 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
- G06F18/24—Classification techniques
-
- 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
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- 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
- G06T2207/10116—X-ray image
-
- 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]
-
- 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
- G06T2207/30164—Workpiece; Machine component
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Quality & Reliability (AREA)
- Analysing Materials By The Use Of Radiation (AREA)
Abstract
The invention provides a casting defect X-ray diagram automatic identification method based on an improved neural network, and belongs to the field of material casting and defect image target detection. The technical scheme adopts convolution to generate a confrontation network model to complete the image characteristic enhancement of the data set. And completing defect marking on the casting defect data set by adopting an image marking tool. And improving the network structure of the target detection YOLOv5 model, adding an attention mechanism, and improving a loss function to train the model. And training the model by using a training set, finally comparing various improved network models, obtaining a YOLOv5-MNv2-SE4 model by taking factors such as accuracy, recall rate and the like which influence the quality of the neural network model as evaluation indexes, and detecting the casting image and identifying the defect type and position coordinates by using the model. The method can effectively enhance the defect characteristics of the sample image and improve the accuracy of identifying the small target object.
Description
Technical Field
The invention relates to the field of material casting and defect image target detection, in particular to a casting defect X-ray diagram automatic identification method based on an improved neural network.
Background
With the development of automobile lightweight technology, the demand of aluminum alloy castings is rapidly increased, and the quality of the aluminum alloy castings directly influences the safety performance of automobile parts. Therefore, quality assurance of the casting is extremely important. In the production of the casting, various defects such as holes, cracks, etc. are inevitably generated due to the limitation of different manufacturing processes. The presence of these defect areas would seriously affect the performance of the casting and, once the production and use continues, would have unimaginable negative consequences for the entire mechanical component. In order to ensure the quality of castings, the detection method based on the traditional machine vision and deep learning is widely applied at present, the problems of low efficiency, poor accuracy, high false detection rate and the like existing in the manual detection in the past are greatly solved, the identification precision of defects of the castings is improved, and the detection cost is greatly saved.
The traditional machine vision detection method usually collects pictures first, then needs to design features and a model based on a machine learning algorithm to complete image processing, and finally obtains a detection result. With the application development of deep learning technology, a Convolutional Neural Network (CNN) is used for solving a difficult image-driven recognition task, can effectively extract key features and learn autonomously, and greatly saves the time and cost for manually designing the features. The detection method based on deep learning is used as a data-driven detection technology, and can obtain a good training effect under the condition of sufficient samples. In practical research, small sample data is often required to be processed, and the use of an X-ray detector to obtain an X-ray image is an important sample data source. Identifying casting defects in X-ray images remains a challenging task. Meanwhile, for defect images with complex backgrounds and unobvious contrasts, the traditional image processing means such as cutting, contrast changing, noise reduction and the like can realize data enhancement but have limited effect and repeatability, and the generated countermeasure network model (GAN) learns the distribution of real samples through a group of random noises, so that the data characteristics of the images are greatly enhanced, and the quality of the generated samples and the stability of model training are improved.
Disclosure of Invention
The invention aims to provide a casting defect X-ray diagram automatic identification method based on an improved neural network, which can effectively improve the accuracy and the accuracy of identification of a small target object.
In order to achieve the purpose, the invention adopts the technical scheme that:
an automatic identification method of casting defect X-ray diagrams based on an improved neural network is characterized in that a convolution countermeasure network and an improved YOLOv5 network model are based on, and the method mainly comprises the following steps:
(1) inputting a plurality of acquired X-ray images of the castings;
(2) performing image preprocessing on the image;
(3) marking the defects in each preprocessed image, and randomly selecting and forming a training set and a verification set according to a ratio of 9: 1;
(4) constructing an improved convolutional neural network model based on YOLOv5, wherein the improved convolutional neural network model comprises adding a channel attention mechanism module SEnet and optimizing a loss function; training the improved convolutional neural network model based on Yolov5 using the training set;
(5) after iteration is carried out for N times, the trained model is used for completing defect identification of the X-ray images in the verification set, identification results of various comparison models are compared, if all indexes of the evaluation model are more comprehensive, the step (7) is carried out, and if not, the step (6) is carried out;
(6) judging the convergence condition of model parameters by using the training set, continuously improving the parameters of the network model, and returning to the step (4);
(7) obtaining a weight file of an improved YOLOv5 network model based on a training process;
(8) and inputting the X-ray image of the casting into a weight file of the improved YOLOv5 network model, and detecting and identifying the defect type and the position coordinates of the defect type of the casting.
The image preprocessing is to utilize convolution to generate a countermeasure network, learn the distribution of real samples through a group of random noises, enhance the characteristics of defective images and generate new sample data.
And the defect labeling is to label the defect in each preprocessed image by using a graphical image annotation tool label.
The casting is an aluminum alloy casting commonly used in the automobile industry.
The defect types are six types.
The improved convolutional neural network model based on the YOLOv5 adopts YOLOv5s with small volume and weight as a model for detecting and identifying the casting defects, and is improved on the basis of the model.
The method for training the improved convolutional neural network model based on the YOLOv5 mainly comprises the following steps:
(4.1) the input layer adaptively zooms the pictures, randomly calls the pictures by adopting a Mosaic data enhancement mode, and automatically calculates the optimal anchor frame value of the data set;
(4.2) adding a Focus structure at the Backbone stage, constructing a CSP structure, and obtaining a downsampling characteristic diagram with complete information through 4 slicing operations and 1 convolution operation of 32 convolution kernels;
(4.3) the output layer adopts CIoU _ Loss, namely:
in the formula, ρ2(b，bgt) The Euclidean distance of the central points of the prediction frame and the real frame; c is the diagonal distance of the minimum closure area which simultaneously contains the prediction frame and the real frame; alpha is a balance parameter; β measures the uniformity of aspect ratio, where α, β can be expressed as follows:
the corresponding loss function is:
replacing the GIoU as a loss function, increasing the measure of the intersection scale;
(4.4) adopting a ShuffleNetv2 structure or a MobileNetv2 structure to replace a convolutional neural network layer structure except for a Focus structure in the original model backbone at the convolutional layer, and simultaneously selecting to introduce the channel attention mechanism module SENet (Squeeze-and-Excitation Networks) at the tail end part of the backbone to construct improved YOLOv5-SNv2-SE and YOLOv5-MNv2-SE network models.
The indexes of the evaluation model comprise GIoU Loss, Objectness Loss, Classification Loss, Precision, Recall, mAP @0.5 (the average Precision value of multiple classes under the threshold value of 0.5 generalized cross-over ratio), mAP @0.5:0.95 (the average Precision value of multiple classes under the threshold value of variation cross-over ratio).
The various comparison models include a Yolov5 prototype model, the improved Yolov5-SNv2-SE model, and a Yolov5-MNv2-SE network model.
The improved YOLOv5 network model is the YOLOv5-MNv2-SE model.
The invention has the beneficial effects that:
the invention provides a casting defect X-ray diagram automatic identification method based on an improved neural network, which does not adopt a traditional method in an image preprocessing stage, utilizes a convolution to generate a confrontation network to complete image feature enhancement of a data set, improves a neural network structure, adds a channel attention mechanism and improves a loss function, compares the improved data set with a Yolov5 original model and other improved models, and selects an optimal network model. Compared with the traditional machine learning method, the method has stronger detection accuracy, the optimized model method has obviously improved recognition accuracy and average accuracy, and has better robustness.
Drawings
FIG. 1 is a flow chart of the casting defect X-ray diagram automatic identification method based on the improved neural network.
FIG. 2 is a network structure diagram of the casting defect X-ray diagram automatic identification method based on the improved neural network.
FIG. 3 is a histogram of the type and quantity of target images of an automatic casting defect X-ray image identification method based on an improved neural network.
FIG. 4 is a diagram of improved model parameter convergence of the casting defect X-ray diagram automatic identification method based on the improved neural network.
FIG. 5 is a comparison diagram of key indexes of three models of the casting defect X-ray diagram automatic identification method based on the improved neural network.
FIG. 6 is a diagram of the identification result of the casting defect X-ray diagram automatic identification method based on the improved neural network on the target defect.
Detailed Description
The present invention will be described in detail with reference to specific examples. The following examples will assist those skilled in the art in further understanding the invention, but are not intended to limit the invention in any way. It should be noted that variations and modifications can be made by persons skilled in the art without departing from the spirit of the invention. All falling within the scope of the present invention.
Referring to fig. 1 to 4, an automatic identification method of casting defect X-ray images based on an improved neural network, which is based on a convolution countermeasure network and an improved YOLOv5 network model, mainly comprises the following steps:
(1) inputting a plurality of acquired X-ray images of the castings;
(2) performing image preprocessing on the image;
(3) marking the defects in each preprocessed image, and randomly selecting and forming a training set and a verification set according to a ratio of 9: 1;
(4) constructing an improved convolutional neural network model based on YOLOv5, wherein the improved convolutional neural network model comprises adding a channel attention mechanism module SEnet and optimizing a loss function; training the improved convolutional neural network model based on Yolov5 using the training set;
(5) after iteration is carried out for N times, the trained model is used for completing defect identification of the X-ray images in the verification set, identification results of various comparison models are compared, if all indexes of the evaluation model are more comprehensive, the step (7) is carried out, and if not, the step (6) is carried out;
(6) judging the convergence condition of model parameters by using the training set, continuously improving the parameters of the network model, and returning to the step (4);
(7) obtaining a weight file of an improved YOLOv5 network model based on a training process;
(8) and inputting the X-ray image of the casting into a weight file of the improved YOLOv5 network model, and detecting and identifying the defect type and the position coordinates of the defect type of the casting.
In the step 1, the casting is an aluminum alloy casting commonly used in the automobile industry, and the defect types in the image are six types.
In step 2, the image preprocessing is to utilize convolution to generate a countermeasure network, learn the distribution of real samples through a group of random noises, enhance the characteristics of the defective images and generate new sample data. The convolution generation countermeasure network is a DCGAN network structure and mainly comprises a generator Generator (G) and a discriminator (D), the core idea is that Nash equilibrium is finally achieved through mutual competition of G and D, and a data set with obvious enhanced data characteristics is generated through continuously learning the mathematical distribution of real sample data.
And 3, marking the defects in each preprocessed image by using a graphical image annotation tool label.
In step 4, the improved convolutional neural network model based on YOLOv5 adopts YOLOv5s with smaller volume and weight as a model for detecting and identifying the defects of the casting, and is improved on the basis of the model, and the training method mainly comprises the following steps:
1) the input layer self-adapts to zoom pictures, adopts a Mosaic data enhancement mode, randomly calls the pictures, and automatically calculates the optimal anchor frame value of the data set;
2) adding a Focus structure at a Backbone stage, constructing a CSP structure, and obtaining a downsampling characteristic diagram with complete information through 4 slicing operations and 1 convolution operation of 32 convolution kernels;
3) the output layer adopts CIoU _ Loss, namely:
in the formula, ρ2(b，bgt) The Euclidean distance of the central points of the prediction frame and the real frame; c is the diagonal distance of the minimum closure area which simultaneously contains the prediction frame and the real frame; alpha is a balance parameter; β measures the uniformity of aspect ratio, where α, β can be expressed as follows:
the corresponding loss function is:
replacing the GIoU as a loss function, increasing the measure of the intersection scale;
4) a ShuffleNetv2 structure or a MobileNetv2 structure is adopted in a convolution layer to replace a convolution nerve network layer structure except a Focus structure in an original model backbone, and meanwhile, a channel attention mechanism module SEnet (Squeeze-and-Excitation Networks) is introduced into the tail end part of the backbone, so that improved YOLOv5-SNv2-SE and YOLOv5-MNv2-SE network models are constructed.
The indexes of the evaluation model in the steps 5-7 include GIoU Loss, Objectness Loss, Classification Loss, Precision, recalling rate, mAP @0.5 (average Precision value of multiple classes under 0.5 generalized cross-over threshold), mAP @0.5:0.95 (average Precision value of multiple classes under change cross-over threshold). The various contrast models include a Yolov5 prototype model, a modified Yolov5-SNv2-SE model, and a Yolov5-MNv2-SE network model.
In step 8, the detection and identification of the casting defects are carried out by utilizing the improved optimal YOLOv5-MNv2-SE model to identify the defects and coordinates.
Referring to fig. 5, after the original YOLOv5 model, the improved YOLOv5-SNv2-SE and the YOLOv5-MNv2-SE are iterated for thousands of times, the key indexes of the models are compared, and as the iteration times are increased, the target identification accuracy (Precision) and the average accuracy value (mAP @0.5) of the three improved models are continuously improved compared with the original YOLOv5, and in combination, the effect of X-ray casting defect image identification by using the YOLOv5-MNv2-SE is the best.
Referring to fig. 6, the target defect recognition result graph is a target defect recognition result obtained by loading a weight file obtained in a training process and inputting an image data set by using an improved YOLOv5-MNv2-SE network model.
Specific embodiments of the present invention have been described above with reference to the accompanying drawings. It is to be understood that the present invention is not limited to the specific embodiments described above, and that various changes or modifications may be made by one skilled in the art within the scope of the appended claims without departing from the spirit of the invention.
Claims (10)
1. An automatic identification method of casting defect X-ray diagrams based on an improved neural network is characterized in that a convolution countermeasure network and an improved YOLOv5 network model are based on, and the method mainly comprises the following steps:
(1) inputting a plurality of acquired X-ray images of the castings;
(2) performing image preprocessing on the image;
(3) marking the defects in each preprocessed image, and randomly selecting and forming a training set and a verification set according to a ratio of 9: 1;
(4) constructing an improved convolutional neural network model based on YOLOv5, wherein the improved convolutional neural network model comprises adding a channel attention mechanism module SEnet and optimizing a loss function; training the improved convolutional neural network model based on Yolov5 using the training set;
(5) after iteration is carried out for N times, the trained model is used for completing defect identification of the X-ray images in the verification set, identification results of various comparison models are compared, if all indexes of the evaluation model are more comprehensive, the step (7) is carried out, and if not, the step (6) is carried out;
(6) judging the convergence condition of model parameters by using the training set, continuously improving the parameters of the network model, and returning to the step (4);
(7) obtaining a weight file of an improved YOLOv5 network model based on a training process;
(8) and inputting the X-ray image of the casting into a weight file of the improved YOLOv5 network model, and detecting and identifying the defect type and the position coordinates of the defect type of the casting.
2. The method for automatically identifying the casting defect X-ray diagram based on the improved neural network as claimed in claim 1, wherein the image preprocessing is to use convolution to generate a countermeasure network, learn the distribution of real samples through a set of random noises, perform feature enhancement on the defect image and generate new sample data.
3. The method for automatically identifying the casting defect X-ray diagram based on the improved neural network as claimed in claim 1, wherein the defect marking is to mark the defect in each preprocessed image by using a graphical image annotation tool, label.
4. The method for automatically identifying the X-ray diagram of the defects of the casting based on the improved neural network as claimed in claim 1, wherein the casting is an aluminum alloy casting commonly used in the automobile industry.
5. The method for automatically identifying the casting defect X-ray diagram based on the improved neural network as claimed in claim 1, wherein the defect types are six types.
6. The method for automatically identifying the casting defect X-ray diagram based on the improved neural network as claimed in claim 1, wherein the improved convolutional neural network model based on the YOLOv5 adopts YOLOv5s with smaller volume and weight as a model for detecting and identifying the casting defect, and is improved on the basis of the model.
7. The method for automatically identifying the casting defect X-ray diagram based on the improved neural network as claimed in claim 1, wherein the training of the improved convolutional neural network model based on the YOLOv5 is mainly carried out by the following steps:
(4.1) the input layer adaptively zooms the pictures, randomly calls the pictures by adopting a Mosaic data enhancement mode, and automatically calculates the optimal anchor frame value of the data set;
(4.2) adding a Focus structure at the Backbone stage, constructing a CSP structure, and obtaining a downsampling characteristic diagram with complete information through 4 slicing operations and 1 convolution operation of 32 convolution kernels;
(4.3) the output layer adopts CIoU _ Loss, namely:
in the formula, ρ2(b，bgt) The Euclidean distance of the central points of the prediction frame and the real frame; c is the diagonal distance of the minimum closure area which simultaneously contains the prediction frame and the real frame; alpha is a balance parameter; β measures the uniformity of aspect ratio, where α, β can be expressed as follows:
the corresponding loss function is:
replacing the GIoU as a loss function, increasing the measure of the intersection scale;
(4.4) adopting a ShuffleNetv2 structure or a MobileNetv2 structure to replace a convolutional neural network layer structure except for a Focus structure in the original model backbone at the convolutional layer, and simultaneously selecting to introduce the channel attention mechanism module SENet (Squeeze-and-Excitation Networks) at the tail end part of the backbone to construct improved YOLOv5-SNv2-SE and YOLOv5-MNv2-SE network models.
8. The method as claimed in claim 7, wherein the various contrast models include YOLOv5 original model, the improved YOLOv5-SNv2-SE and YOLOv5-MNv2-SE network models.
9. The method for automatically identifying casting defects based on the improved neural network X-ray diagram of claim 7, wherein the improved YOLOv5 network model is the YOLOv5-MNv2-SE model.
10. The method for automatically identifying the casting defect X-ray diagram based on the improved neural network as claimed in claim 1, wherein the indexes of the evaluation model comprise GIoU Loss (generalized cross-over ratio Loss), Objectness Loss (target detection Loss), Classification Loss (quasi-target Classification Loss), Precision, Recall, mAP @0.5(0.5 generalized cross-over ratio threshold lower multi-class average Precision value), mAP @0.5:0.95 (variation cross-over ratio threshold lower multi-class average Precision value).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111474014.3A CN114372955A (en) | 2021-12-02 | 2021-12-02 | Casting defect X-ray diagram automatic identification method based on improved neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111474014.3A CN114372955A (en) | 2021-12-02 | 2021-12-02 | Casting defect X-ray diagram automatic identification method based on improved neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114372955A true CN114372955A (en) | 2022-04-19 |
Family
ID=81139295
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111474014.3A Pending CN114372955A (en) | 2021-12-02 | 2021-12-02 | Casting defect X-ray diagram automatic identification method based on improved neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114372955A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114549997A (en) * | 2022-04-27 | 2022-05-27 | 清华大学 | X-ray image defect detection method and device based on regional feature extraction |
| CN114882410A (en) * | 2022-05-11 | 2022-08-09 | 华东交通大学 | Tunnel ceiling lamp fault detection method and system based on improved positioning loss function |
| CN114972261A (en) * | 2022-05-27 | 2022-08-30 | 东北大学 | Method for identifying surface quality defects of plate strip steel |
| CN115266774A (en) * | 2022-07-29 | 2022-11-01 | 中国特种设备检测研究院 | Weld ray detection and evaluation method based on artificial intelligence |
| CN115273072A (en) * | 2022-06-13 | 2022-11-01 | 南京林业大学 | Apple leaf disease detection method based on improved Yolov5s model |
| CN117197146A (en) * | 2023-11-08 | 2023-12-08 | 北京航空航天大学江西研究院 | Automatic identification method for internal defects of castings |
| CN117437190A (en) * | 2023-10-18 | 2024-01-23 | 湖北汽车工业学院 | YOLOv5 s-based aluminum surface defect detection method |
| CN118032820A (en) * | 2024-04-02 | 2024-05-14 | 深圳市艾兰特科技有限公司 | SMT full-automatic online X-RAY nondestructive testing system |
-
2021
- 2021-12-02 CN CN202111474014.3A patent/CN114372955A/en active Pending
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114549997A (en) * | 2022-04-27 | 2022-05-27 | 清华大学 | X-ray image defect detection method and device based on regional feature extraction |
| CN114882410A (en) * | 2022-05-11 | 2022-08-09 | 华东交通大学 | Tunnel ceiling lamp fault detection method and system based on improved positioning loss function |
| CN114882410B (en) * | 2022-05-11 | 2023-09-12 | 华东交通大学 | Tunnel dome lamp fault detection method and system based on improved positioning loss function |
| CN114972261A (en) * | 2022-05-27 | 2022-08-30 | 东北大学 | Method for identifying surface quality defects of plate strip steel |
| CN115273072A (en) * | 2022-06-13 | 2022-11-01 | 南京林业大学 | Apple leaf disease detection method based on improved Yolov5s model |
| CN115273072B (en) * | 2022-06-13 | 2023-05-19 | 南京林业大学 | Apple leaf disease detection method based on improved Yolov5s model |
| CN115266774A (en) * | 2022-07-29 | 2022-11-01 | 中国特种设备检测研究院 | Weld ray detection and evaluation method based on artificial intelligence |
| CN115266774B (en) * | 2022-07-29 | 2024-02-13 | 中国特种设备检测研究院 | Artificial intelligence-based weld joint ray detection and evaluation method |
| CN117437190A (en) * | 2023-10-18 | 2024-01-23 | 湖北汽车工业学院 | YOLOv5 s-based aluminum surface defect detection method |
| CN117197146A (en) * | 2023-11-08 | 2023-12-08 | 北京航空航天大学江西研究院 | Automatic identification method for internal defects of castings |
| CN118032820A (en) * | 2024-04-02 | 2024-05-14 | 深圳市艾兰特科技有限公司 | SMT full-automatic online X-RAY nondestructive testing system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114372955A (en) | Casting defect X-ray diagram automatic identification method based on improved neural network | |
| CN111179251B (en) | Defect detection system and method based on twin neural network and by utilizing template comparison | |
| CN107437245B (en) | High-speed railway contact net fault diagnosis method based on deep convolutional neural network | |
| CN108918536B (en) | Tire mold surface character defect detection method, device, equipment and storage medium | |
| CN109509187B (en) | Efficient inspection algorithm for small defects in large-resolution cloth images | |
| CN113239930B (en) | Glass paper defect identification method, system, device and storage medium | |
| CN111462076B (en) | Full-slice digital pathological image fuzzy region detection method and system | |
| CN110543878A (en) | pointer instrument reading identification method based on neural network | |
| CN107194418B (en) | Rice aphid detection method based on antagonistic characteristic learning | |
| CN111667455A (en) | AI detection method for various defects of brush | |
| CN111027443B (en) | Bill text detection method based on multitask deep learning | |
| CN116012291A (en) | Industrial part image defect detection method and system, electronic equipment and storage medium | |
| CN115082922A (en) | Water meter digital picture processing method and system based on deep learning | |
| CN109815957A (en) | A kind of character recognition method based on color image under complex background | |
| CN112396580A (en) | Circular part defect detection method | |
| CN117036314A (en) | Method for detecting oxidation area of high-density flexible IC substrate | |
| CN113128556B (en) | Deep learning test case sequencing method based on mutation analysis | |
| CN114782322A (en) | YOLOv5 model arc additive manufacturing molten pool defect detection method | |
| Lin et al. | Image denoising of printed circuit boards using conditional generative adversarial network | |
| CN111369508A (en) | Defect detection method and system for metal three-dimensional lattice structure | |
| CN116993721B (en) | Steel plate surface defect detection method based on weak supervision neural network | |
| CN114897901B (en) | Battery quality detection method and device based on sample expansion and electronic equipment | |
| CN111724352B (en) | Patch LED flaw labeling method based on kernel density estimation | |
| CN117671339A (en) | Casting defect identification method based on regional convolution network | |
| CN117853417A (en) | Weld defect identification method and device, electronic equipment and storage medium |
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 |