CN113487570A - High-temperature continuous casting billet surface defect detection method based on improved yolov5x network model - Google Patents
High-temperature continuous casting billet surface defect detection method based on improved yolov5x network model Download PDFInfo
- Publication number
- CN113487570A CN113487570A CN202110763410.1A CN202110763410A CN113487570A CN 113487570 A CN113487570 A CN 113487570A CN 202110763410 A CN202110763410 A CN 202110763410A CN 113487570 A CN113487570 A CN 113487570A
- Authority
- CN
- China
- Prior art keywords
- continuous casting
- casting billet
- network model
- yolov5x
- temperature continuous
- 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.)
- Granted
Links
- 238000009749 continuous casting Methods 0.000 title claims abstract description 83
- 230000007547 defect Effects 0.000 title claims abstract description 62
- 238000001514 detection method Methods 0.000 title abstract description 24
- 238000012549 training Methods 0.000 claims abstract description 21
- 238000004519 manufacturing process Methods 0.000 claims abstract description 19
- 238000007781 pre-processing Methods 0.000 claims abstract description 5
- 238000002372 labelling Methods 0.000 claims abstract description 3
- 238000000034 method Methods 0.000 claims description 17
- 238000012545 processing Methods 0.000 claims description 5
- 230000004927 fusion Effects 0.000 claims description 4
- 238000003708 edge detection Methods 0.000 claims description 3
- 238000007689 inspection Methods 0.000 abstract description 3
- 238000012360 testing method Methods 0.000 description 8
- 238000005516 engineering process Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000012795 verification Methods 0.000 description 4
- 229910000831 Steel Inorganic materials 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 239000010959 steel Substances 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000005266 casting Methods 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000001816 cooling Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000003631 expected effect Effects 0.000 description 1
- 238000010191 image analysis Methods 0.000 description 1
- 230000005764 inhibitory process Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
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
- G06T7/0006—Industrial image inspection using a design-rule based approach
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- 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
- G06T2207/10016—Video; Image sequence
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30136—Metal
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Quality & Reliability (AREA)
- Radiation Pyrometers (AREA)
- Continuous Casting (AREA)
Abstract
The invention discloses a high-temperature continuous casting billet surface defect detection method based on an improved yolov5x network model, and belongs to the technical field of high-temperature continuous casting billet surface defect identification. Collecting the required number of high-temperature continuous casting billet surface images on a continuous casting billet production line; preprocessing the acquired surface images of the high-temperature continuous casting billet, and labeling the defects in each preprocessed image to obtain a training set of a yolov5x network model; the yolov5x network model is improved, and comprises the following steps: adding a GhostBottleneck module in the yolov5x network model to replace the Bottleneck module; training the improved yolov5x network model by using the training set to obtain a yolov5-Ghost network model; and detecting the surface defects of the high-temperature continuous casting billet on the continuous casting billet production line in real time by using the yolov5-Ghost network model. The model volume is reduced, a lighter yolov5-Ghost network model is established, the detection efficiency is improved, and the operation cost of quality inspection work is effectively reduced.
Description
Technical Field
The invention belongs to the technical field of high-temperature continuous casting billet surface defect identification, and particularly relates to a high-temperature continuous casting billet surface defect detection method based on an improved yolov5x network model.
Background
The production of continuous casting billets is a key link of the whole steel production line, and at present, in the continuous casting production process, along with the development of the process, the casting billet drawing speed is increased, and the high drawing speed can cause various defects to be generated on the surface of the high-temperature continuous casting billets, wherein the defects such as surface cracks can seriously affect the quality of the continuous casting billets and can also greatly affect the downstream steel rolling process. And present main detection mode is for carrying out artifical the detection after breaking away from the production line cooling to the room temperature with high temperature continuous casting billet, can consume a lot of manpower, material resources like this and cause unnecessary energy resource consumption. The traditional image analysis technology cannot accurately distinguish the surface defects of the high-temperature continuous casting billet, so that the related information of the surface defects of the high-temperature continuous casting billet cannot be effectively acquired, and the surface defects can be timely found. With the gradual maturity of artificial intelligence technology in recent years, the combination of artificial intelligence technology and defect detection technology has become the trend of high temperature continuous casting billet surface defect detection in the future.
Disclosure of Invention
In order to solve the problems, the invention provides a high-temperature continuous casting billet surface defect detection method based on an improved yolov5x network model, and aims to improve the high-temperature continuous casting billet surface defect detection efficiency and effectively reduce the operation cost of the high-temperature continuous casting billet surface quality inspection work.
The technical scheme of the invention is as follows:
the method for detecting the surface defects of the high-temperature continuous casting billet based on the improved yolov5x network model comprises the following steps:
collecting the surface images of the high-temperature continuous casting billets in required quantity;
preprocessing the acquired surface images of the high-temperature continuous casting billet, and labeling the defects in each preprocessed image to obtain a training set of a yolov5x network model;
the yolov5x network model is improved, and comprises the following steps: adding a GhostBottleneck module in the yolov5x network model to replace the Bottleneck module;
training the improved yolov5x network model by using the training set to obtain a yolov5-Ghost network model;
and detecting the surface defects of the high-temperature continuous casting billet on the continuous casting billet production line in real time by using the yolov5-Ghost network model.
Further, according to the high-temperature continuous casting billet surface defect detection method based on the improved yolov5x network model, the required number of high-temperature continuous casting billet surface images are collected on a continuous casting billet production line.
Further, according to the method for detecting the surface defects of the high-temperature continuous casting billet based on the improved yolov5x network model, the pretreatment of the acquired surface image of the high-temperature continuous casting billet comprises the following steps: image details are improved through a multi-scale fusion algorithm, and surface defect edge detection of the high-temperature continuous casting billet is realized based on an OpenCV image processing algorithm.
Further, according to the method for detecting the surface defects of the high-temperature continuous casting billet based on the improved yolov5x network model, a graphic image annotation tool LabelImg is used for annotating the defects in each preprocessed image.
Further, according to the method for detecting the surface defects of the high-temperature continuous casting billet based on the improved yolov5x network model, when the yolo5x-Ghost network model detects the defects, the defects are marked by a rectangular frame, and the rectangular frame is two-point coordinates.
Further, according to the high-temperature continuous casting billet surface defect detection method based on the improved yolov5x network model, a camera is used for collecting the required number of high-temperature continuous casting billet surface images on a continuous casting billet production line.
Further, according to the method for detecting the surface defects of the high-temperature continuous casting billet based on the improved yolov5x network model, when a camera is used for acquiring the required number of high-temperature continuous casting billet surface images on a continuous casting billet production line, the height of the camera is set to be within the range of 2-3 meters, and the incident angle is set to be within the range of 15-25 degrees.
Compared with the prior art, the invention has the following beneficial effects: the method combines an artificial intelligence technology with a defect detection technology, shooting and sampling are carried out through detection equipment on a continuous casting billet production line site, collected data are preprocessed through an OpenCV image processing algorithm to serve as a data set for model training to train a yolov5x neural network model, defect information is highlighted, the adaptability of the algorithm in the actual environment is enhanced, a Ghostbottleneck module is added into a yolov5x network model to replace a Bottleneck module, and the model volume is reduced under the conditions that the model accuracy is not reduced and the accuracy is reduced to a minimum degree, so that a lighter model is established, the detection efficiency is greatly improved, and the operation cost of the surface quality inspection work of the high-temperature continuous casting billet is effectively reduced.
Drawings
FIG. 1 is a schematic flow chart of a high-temperature continuous casting billet surface defect detection method based on an improved yolov5x network model in the embodiment;
fig. 2(a) is a schematic diagram of the ghost bottleeck module when the convolution step stride is 1; (b) a schematic diagram of a GhostBottleneck module when the convolution step length equals 2;
FIG. 3 is a diagram showing the training results of yolov5-Ghost network model in this embodiment;
fig. 4 is a diagram showing the training result of the yolov5x network model in this embodiment.
Detailed Description
To facilitate an understanding of the present application, the present application will now be described more fully with reference to the accompanying drawings. Preferred embodiments of the present application are given in the accompanying drawings. This application may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.
Fig. 1 is a schematic flow chart of a high-temperature continuous casting billet surface defect detection method based on an improved yolov5x network model, which comprises the following steps:
step 1: collecting a surface image of a high-temperature continuous casting billet on a continuous casting billet production line;
as the current continuous casting process is mature, the surface defects of the high-temperature continuous casting billet are relatively few, and no acknowledged usable data set of the surface defects of the high-temperature continuous casting billet exists. Therefore, the embodiment adopts a field data acquisition mode to obtain a high-temperature continuous casting billet surface defect data set with sufficient data volume.
In the preferred embodiment, aiming at the surface defect condition of a high-temperature continuous casting billet in a certain steel mill, a high-definition camera with strong light inhibition and high temperature resistance is used for carrying out data acquisition on the surface of the high-temperature continuous casting billet, before shooting, the height of the camera and the incident angle of the camera are calculated through the actual shooting range and the camera view angle, an optimal imaging system is established, and the surface image of the high-temperature continuous casting billet with the defect is shot. In the embodiment, the height of the camera is determined to be in the range of 2-3 meters, and the imaging effect is best when the incident angle of the camera is in the range of 15-25 degrees.
Step 2: preprocessing the acquired high-temperature continuous casting billet surface image, marking the defects in each preprocessed image to obtain a high-temperature continuous casting billet surface defect data set, and dividing the data set into a training set for pre-training, a verification set for verification and a test set for testing.
Considering the influence of the possibly unclear defect details in the high-temperature continuous casting billet image and strong red light interference on the image, the shot image data is preprocessed, the image details are improved through a multi-scale fusion algorithm, and the high-temperature continuous casting billet surface defect edge detection is realized based on an OpenCV image processing algorithm. In the multi-scale fusion algorithm, the original image is filtered by a multi-scale Gaussian function, then the original image is subtracted from the original image to obtain detail information of different degrees, the detail information of different degrees is weighted, and then the weighted information is fused into the original image to reinforce the information of the original image.
And then finely marking the preprocessed high-temperature continuous casting billet surface image, marking the defects in each image by using a graphic image annotation tool LabelImg to manufacture a high-temperature continuous casting billet surface defect data set, selecting a part of data as a training set, a part of data as a verification set, and the rest of data as a test set, wherein the number ratio of the pictures in the training set, the verification set and the test set is 8:1: 1.
And step 3: the yolov5x network model was improved.
According to the preferred embodiment, a GhostBottleneck module is added to the yolov5x network model to replace a Bottleneck module, the image high-dimensional feature extraction function of the GhostBottleneck module is fully utilized, and the improvement can be understood through the structure of the GhostBottleneck module, so that the whole yolov5x network model is lighter and lighter, the calculated amount is less, the same effect as the original model can be achieved, and the improvement is more applicable.
As shown in fig. 2(a), the Ghost bottleeck module is composed of two stacked Ghost modules. The first Ghost module is used as an extension layer, and the number of channels is increased. In the present embodiment, the ratio of the number of output channels to the number of input channels of the Ghost module is referred to as an expansion ratio (expansion ratio). The second Ghost module reduces the number of channels to match the path shortcut. Then, shortcut is used to connect the inputs and outputs of the two Ghost modules. The second Ghost module in the Ghost Bottleneck does not use the ReLU activation function, and the other layers apply Batch Normalization (BN) and ReLu activation functions after each layer; in Ghost bottleeck, for the convolution step size stride of 2, as shown in fig. 2(b), two Ghost modules are connected by deep convolution with stride of 2.
And 4, step 4: training the improved yolov5x network model by using the training set to obtain a yolov5-Ghost network model;
in the preferred embodiment, epoch is set within 200-300 when the improved yolov5x network model is trained using the training set. And then carrying out parameter fine adjustment on the yolov5-Ghost network model, replacing a Bottleneck module in the yolov5x network model with a GhostBottleneck module to possibly reduce the detection accuracy of the yolov5-Ghost network model to different degrees, and carrying out parameter fine adjustment to possibly restore the model accuracy. Some models are high in accuracy after being lightened, and fine adjustment can be omitted, but the models are suddenly changed after the weight is lightened, calculation can generate deviation, unexpected detection results are better than the previous detection results, and fine adjustment is still necessary.
And 5: and detecting the surface defects of the high-temperature continuous casting billet on the continuous casting billet production line in real time by using the yolov5-Ghost network model.
After a relevant environment is configured on a computer on a production field, the yolov5-Ghost network model is adapted and debugged on the computer, meanwhile, the computer is adapted with a camera for detecting the surface defects of the high-temperature continuous casting billet, and image flow taking is well performed. And (3) sending the key frame extracted by the acquired video bare stream according to a certain frame extraction cleaning strategy to a yolov5-Ghost network model for processing, wherein the key frame is an image containing surface defects of the high-temperature continuous casting billet.
When the yolov5-Ghost network model detects a defect, it is marked with a brightly colored rectangular box, which is a two-point coordinate. Obj [ ] ═ Yolo _ Result [ Xn1, Yn1, Xn2, Yn2 ]. Xn1, Yn1 are the coordinates of the upper left corner of the nth rectangular frame; xn2, Yn2 are the coordinates of the bottom right corner of the nth rectangle.
In the preferred embodiment, the calculation times (GFLOPs) of yolov5-Ghost network model in the actual operation process in the addition, subtraction, multiplication and division calculation process are shown in Table 1. The training results of yolov5x model are shown in FIG. 3, and the training results of yolo5x-Ghost model are shown in FIG. 4.
TABLE 1
As can be seen from table 1 and fig. 3 and 4, yolov5-Ghost network model has fewer operations and is lighter. In fig. 3 and 4, Box is the average value of the loss function, and the smaller the value, the more accurate the rectangular Box marked by yolov5-Ghost network model is; objectness is the mean value of the target detection loss, and the smaller the value is, the more accurate the target detection is; the Classication is the mean value of the class, and the smaller the value is, the more accurate the class is; precision is the accuracy (correctly detected target in test set/actually found target in test set); recalling as Recall (correctly detected target in test set/marked target in test set); mAP @0.5 and mAP @0.5: 0.95: AP is the area enclosed after drawing by using Precision and Recall as two axes, m represents the average value, and the number behind @ represents the threshold value for judging IoU as a positive sample and a negative sample; mAP @0.5:0.95 represents the average mAP over different IoU thresholds (from 0.5 to 0.95, step size 0.05) (0.5, 0.55, 0.6, 0.65, 0.7, 0.75, 0.8, 0.85, 0.9, 0.95). Comparing fig. 3 and fig. 4, it can be seen that after the yolov5x network model is lightened by adding the ghost bottleeck module instead of the bottleeck module, the detection precision is hardly reduced, and the expected effect of the invention is achieved.
It should be understood that various changes and modifications can be made by those skilled in the art without departing from the spirit and scope of the invention.
Claims (7)
1. The method for detecting the surface defects of the high-temperature continuous casting billet based on the improved yolov5x network model is characterized by comprising the following steps of:
collecting the surface images of the high-temperature continuous casting billets in required quantity;
preprocessing the acquired surface images of the high-temperature continuous casting billet, and labeling the defects in each preprocessed image to obtain a training set of a yolov5x network model;
the yolov5x network model is improved, and comprises the following steps: adding a GhostBottleneck module in the yolov5x network model to replace the Bottleneck module;
training the improved yolov5x network model by using the training set to obtain a yolov5-Ghost network model;
and detecting the surface defects of the high-temperature continuous casting billet on the continuous casting billet production line in real time by using the yolov5-Ghost network model.
2. The method for detecting the surface defects of the high-temperature continuous casting billet based on the improved yolov5x network model is characterized in that the required number of high-temperature continuous casting billet surface images are collected on a continuous casting billet production line.
3. The method for detecting the surface defects of the high-temperature continuous casting billet based on the improved yolov5x network model according to claim 1, wherein the preprocessing the acquired surface images of the high-temperature continuous casting billet comprises the following steps: image details are improved through a multi-scale fusion algorithm, and surface defect edge detection of the high-temperature continuous casting billet is realized based on an OpenCV image processing algorithm.
4. The method for detecting the surface defects of the high-temperature continuous casting billet based on the improved yolov5x network model, according to claim 1, wherein a graphic image annotation tool LabelImg is used for marking the defects in each image after pretreatment.
5. The method for detecting the surface defects of the high-temperature continuous casting billet based on the improved yolov5x network model is characterized in that when the yolov5-Ghost network model detects the defects, the defects are marked by a rectangular frame, and the rectangular frame is a two-point coordinate.
6. The method for detecting the surface defects of the high-temperature continuous casting billet based on the improved yolov5x network model is characterized in that a camera is used for acquiring the required number of images of the surface of the high-temperature continuous casting billet on a continuous casting billet production line.
7. The method for detecting the surface defects of the high-temperature continuous casting billet based on the improved yolov5x network model is characterized in that when a camera is used for acquiring the required number of images of the surface of the high-temperature continuous casting billet on a continuous casting billet production line, the height of the camera is set to be within the range of 2-3 meters, and the incident angle is set to be within the range of 15-25 degrees.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110763410.1A CN113487570B (en) | 2021-07-06 | 2021-07-06 | High-temperature continuous casting billet surface defect detection method based on improved yolov5x network model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110763410.1A CN113487570B (en) | 2021-07-06 | 2021-07-06 | High-temperature continuous casting billet surface defect detection method based on improved yolov5x network model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113487570A true CN113487570A (en) | 2021-10-08 |
| CN113487570B CN113487570B (en) | 2024-01-30 |
Family
ID=77940514
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110763410.1A Active CN113487570B (en) | 2021-07-06 | 2021-07-06 | High-temperature continuous casting billet surface defect detection method based on improved yolov5x network model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113487570B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113935416A (en) * | 2021-10-15 | 2022-01-14 | 东北大学 | Continuous casting breakout prediction method for multi-state crystallizer |
| CN114418064A (en) * | 2021-12-27 | 2022-04-29 | 西安天和防务技术股份有限公司 | Target detection method, terminal equipment and storage medium |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015055060A1 (en) * | 2013-10-16 | 2015-04-23 | 湖南镭目科技有限公司 | Online detecting method for continuous casting slab surface quality |
| CN109215009A (en) * | 2017-06-29 | 2019-01-15 | 上海金艺检测技术有限公司 | Continuous casting billet surface image defect inspection method based on depth convolutional neural networks |
| CN110400296A (en) * | 2019-07-19 | 2019-11-01 | 重庆邮电大学 | The scanning of continuous casting blank surface defects binocular and deep learning fusion identification method and system |
| CN110728657A (en) * | 2019-09-10 | 2020-01-24 | 江苏理工学院 | Annular bearing outer surface defect detection method based on deep learning |
| CN111476756A (en) * | 2020-03-09 | 2020-07-31 | 重庆大学 | Method for identifying casting DR image loose defects based on improved YO L Ov3 network model |
| US20200292462A1 (en) * | 2019-03-15 | 2020-09-17 | Inventec (Pudong) Technology Corporation | Surface defect detection system and method thereof |
| CN112132828A (en) * | 2020-10-15 | 2020-12-25 | 浙江工商大学 | Film flaw detection method based on deep learning |
| CN112634237A (en) * | 2020-12-25 | 2021-04-09 | 福州大学 | Long bamboo strip surface defect detection method and system based on YOLOv3 improved network |
| CN112730425A (en) * | 2020-12-19 | 2021-04-30 | 滁州蜂云视觉科技有限公司 | Industrial scene oriented product surface defect automatic identification method |
| CN112926584A (en) * | 2021-05-11 | 2021-06-08 | 武汉珈鹰智能科技有限公司 | Crack detection method and device, computer equipment and storage medium |
| CN112926501A (en) * | 2021-03-23 | 2021-06-08 | 哈尔滨理工大学 | Traffic sign detection algorithm based on YOLOv5 network structure |
| CN112990392A (en) * | 2021-05-20 | 2021-06-18 | 四川大学 | New material floor defect target detection system based on improved YOLOv5 algorithm |
| CN113011333A (en) * | 2021-03-19 | 2021-06-22 | 哈尔滨工业大学 | System and method for obtaining optimal venipuncture point and direction based on near-infrared image |
-
2021
- 2021-07-06 CN CN202110763410.1A patent/CN113487570B/en active Active
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015055060A1 (en) * | 2013-10-16 | 2015-04-23 | 湖南镭目科技有限公司 | Online detecting method for continuous casting slab surface quality |
| CN109215009A (en) * | 2017-06-29 | 2019-01-15 | 上海金艺检测技术有限公司 | Continuous casting billet surface image defect inspection method based on depth convolutional neural networks |
| US20200292462A1 (en) * | 2019-03-15 | 2020-09-17 | Inventec (Pudong) Technology Corporation | Surface defect detection system and method thereof |
| CN110400296A (en) * | 2019-07-19 | 2019-11-01 | 重庆邮电大学 | The scanning of continuous casting blank surface defects binocular and deep learning fusion identification method and system |
| CN110728657A (en) * | 2019-09-10 | 2020-01-24 | 江苏理工学院 | Annular bearing outer surface defect detection method based on deep learning |
| CN111476756A (en) * | 2020-03-09 | 2020-07-31 | 重庆大学 | Method for identifying casting DR image loose defects based on improved YO L Ov3 network model |
| CN112132828A (en) * | 2020-10-15 | 2020-12-25 | 浙江工商大学 | Film flaw detection method based on deep learning |
| CN112730425A (en) * | 2020-12-19 | 2021-04-30 | 滁州蜂云视觉科技有限公司 | Industrial scene oriented product surface defect automatic identification method |
| CN112634237A (en) * | 2020-12-25 | 2021-04-09 | 福州大学 | Long bamboo strip surface defect detection method and system based on YOLOv3 improved network |
| CN113011333A (en) * | 2021-03-19 | 2021-06-22 | 哈尔滨工业大学 | System and method for obtaining optimal venipuncture point and direction based on near-infrared image |
| CN112926501A (en) * | 2021-03-23 | 2021-06-08 | 哈尔滨理工大学 | Traffic sign detection algorithm based on YOLOv5 network structure |
| CN112926584A (en) * | 2021-05-11 | 2021-06-08 | 武汉珈鹰智能科技有限公司 | Crack detection method and device, computer equipment and storage medium |
| CN112990392A (en) * | 2021-05-20 | 2021-06-18 | 四川大学 | New material floor defect target detection system based on improved YOLOv5 algorithm |
Non-Patent Citations (3)
| Title |
|---|
| HONG LI等: ""A UAV Detection Algorithm Combined With Lightweight Network"", 《IEEE》 * |
| 于科为;: "基于卷积神经网络的工件缺陷检测研究", 信息与电脑(理论版), no. 21 * |
| 肖瑞雪等: ""结合高效特征融合的可变尺寸图像隐写分析"", 《计算机工程与应用》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113935416A (en) * | 2021-10-15 | 2022-01-14 | 东北大学 | Continuous casting breakout prediction method for multi-state crystallizer |
| CN114418064A (en) * | 2021-12-27 | 2022-04-29 | 西安天和防务技术股份有限公司 | Target detection method, terminal equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113487570B (en) | 2024-01-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110390691A (en) | A kind of ore scale measurement method and application system based on deep learning | |
| CN110992349A (en) | Underground pipeline abnormity automatic positioning and identification method based on deep learning | |
| CN107966454A (en) | A kind of end plug defect detecting device and detection method based on FPGA | |
| CN113947731B (en) | Foreign matter identification method and system based on contact net safety inspection | |
| CN112967255A (en) | Shield segment defect type identification and positioning system and method based on deep learning | |
| CN113487570B (en) | High-temperature continuous casting billet surface defect detection method based on improved yolov5x network model | |
| CN112802016A (en) | Real-time cloth defect detection method and system based on deep learning | |
| CN112183313A (en) | SlowFast-based power operation field action identification method | |
| CN107024478A (en) | A kind of part processes automatic defect identifying system | |
| CN110490842A (en) | A kind of steel strip surface defect detection method based on deep learning | |
| CN105184229A (en) | Online learning based real-time pedestrian detection method in dynamic scene | |
| CN112419261B (en) | Visual acquisition method and device with abnormal point removing function | |
| CN116612106A (en) | Method for detecting surface defects of optical element based on YOLOX algorithm | |
| CN110728269B (en) | High-speed rail contact net support pole number plate identification method based on C2 detection data | |
| CN116385758A (en) | Detection method for damage to surface of conveyor belt based on YOLOv5 network | |
| CN117252815A (en) | Industrial part defect detection method, system, equipment and storage medium based on 2D-3D multi-mode image | |
| CN111507249A (en) | Transformer substation nest identification method based on target detection | |
| CN109815957A (en) | A kind of character recognition method based on color image under complex background | |
| CN109975307A (en) | Bearing surface defect detection system and detection method based on statistics projection training | |
| CN114078106A (en) | Defect detection method based on improved Faster R-CNN | |
| Wang et al. | Visual defect detection for substation equipment based on joint inspection data of camera and robot | |
| Hu et al. | Printed Circuit Board (PCB) Surface Micro Defect Detection Model Based on Residual Network with Novel Attention Mechanism. | |
| CN113420839A (en) | Semi-automatic labeling method and segmentation positioning system for stacking planar target objects | |
| Wang et al. | Research on appearance defect detection of power equipment based on improved faster-rcnn | |
| Li et al. | Weight Estimation Method of Coal and Gangue on Conveyor Belt Based on Instance Segmentation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |