CN115494092A - Carbon fiber gas cylinder nondestructive testing method and system based on fast convolutional neural network - Google Patents
Carbon fiber gas cylinder nondestructive testing method and system based on fast convolutional neural network Download PDFInfo
- Publication number
- CN115494092A CN115494092A CN202211221392.5A CN202211221392A CN115494092A CN 115494092 A CN115494092 A CN 115494092A CN 202211221392 A CN202211221392 A CN 202211221392A CN 115494092 A CN115494092 A CN 115494092A
- Authority
- CN
- China
- Prior art keywords
- image
- gas cylinder
- carbon fiber
- original
- neural network
- 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
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N23/00—Investigating or analysing materials by the use of wave or particle radiation, e.g. X-rays or neutrons, not covered by groups G01N3/00 – G01N17/00, G01N21/00 or G01N22/00
- G01N23/02—Investigating or analysing materials by the use of wave or particle radiation, e.g. X-rays or neutrons, not covered by groups G01N3/00 – G01N17/00, G01N21/00 or G01N22/00 by transmitting the radiation through the material
- G01N23/04—Investigating or analysing materials by the use of wave or particle radiation, e.g. X-rays or neutrons, not covered by groups G01N3/00 – G01N17/00, G01N21/00 or G01N22/00 by transmitting the radiation through the material and forming images of the material
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N23/00—Investigating or analysing materials by the use of wave or particle radiation, e.g. X-rays or neutrons, not covered by groups G01N3/00 – G01N17/00, G01N21/00 or G01N22/00
- G01N23/02—Investigating or analysing materials by the use of wave or particle radiation, e.g. X-rays or neutrons, not covered by groups G01N3/00 – G01N17/00, G01N21/00 or G01N22/00 by transmitting the radiation through the material
- G01N23/06—Investigating or analysing materials by the use of wave or particle radiation, e.g. X-rays or neutrons, not covered by groups G01N3/00 – G01N17/00, G01N21/00 or G01N22/00 by transmitting the radiation through the material and measuring the absorption
- G01N23/18—Investigating the presence of flaws defects or foreign matter
-
- 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
-
- 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
Abstract
The invention discloses a carbon fiber gas cylinder nondestructive testing method and system based on a fast convolution neural network, which comprises the following steps: acquiring an original X-ray image; processing the data of the original X-ray image to obtain a processed image; extracting image features based on the processed image; obtaining a candidate region based on the image characteristics; classifying and regressing the candidate region to obtain a defect region image; and (5) extracting the outline of the image of the defect area to obtain a determined defect area, and marking the determined defect area on the original X-ray image to finish detection. The identification accuracy is high, the performance is good, after the system is deployed to a production environment and is actually applied, the system can stably run for a long time, the requirement of the industrial production environment can be met, and the system can help the production flow of industrial enterprises to advance towards intellectualization and digitization.
Description
Technical Field
The invention relates to the technical field of nondestructive testing, in particular to a carbon fiber gas cylinder nondestructive testing method and system based on a fast convolutional neural network.
Background
The gas cylinder is a movable pressure container with a main body structure in a bottle shape and generally filled with gas (such as compressed gas, liquefied gas, dissolved and adsorbed gas and the like). The carbon fiber gas cylinder is formed by processing a high-quality and high-efficiency carbon fiber wire and a high-strength aluminum alloy inner container by using a full winding technology. It has the excellent performances of corrosion resistance, long service life, light weight, high strength and no breakage, and can be widely applied to the fields of fire fighting, chemical engineering, mining, automobile industry and other high and new fields. However, due to technical barriers, limitations of the technological level and the influence of the use time, various defect problems occur to the carbon fiber composite gas cylinder, and the common defect problems include debonding, cracking, inclusion and the like.
In view of these problems, in recent years, with the wide application of the nondestructive testing technology in the field of composite materials, the advantages of the nondestructive testing technology in the aspect of defect detection are gradually remarkable, the core of the nondestructive testing technology lies in that the nondestructive testing technology does not damage or change the physical state and the chemical state of the object to be tested, and meanwhile, by applying the technology, the various contents such as the property, the state, the structure and the like of the object to be tested can be effectively detected, so that necessary data support is provided for the work of various industries, and the work efficiency and the work quality of the various industries are effectively improved.
The X-ray nondestructive testing technology is one of key components in the nondestructive testing technology, the defects on the carbon fiber gas cylinder can be detected by using the X-ray nondestructive testing technology, the detection method used in the current industry is mainly a manual detection method, namely, workers classify and manually position the defects by naked eyes, and the defects have the following defects:
1. the gas cylinder defect detection process comprises a plurality of steps, wherein most of the steps depend on manual operation, for example, workers need to observe an X-ray image of the gas cylinder in front of a screen to judge defect points existing on the gas cylinder, however, the mode is time-consuming and labor-consuming, the detection work is complicated, the operators are difficult to concentrate on for a long time, the detection efficiency is low, and false detection and missed detection are easily caused.
2. The existing defect monitoring is already used for detecting a defect image, and actual inspection shows that the defect detection rate of the existing method is still acceptable, but the overall effect is not good, and the following reasons exist: firstly, in the using process, system equipment must be deployed beside a production line to ensure the image transmission efficiency, and workers need to operate computer equipment nearby the production line, so that the use is inconvenient; secondly, the stability of the method for processing abnormal conditions is poor, and the operation efficiency of the system cannot be ensured after long-time use.
Disclosure of Invention
In order to solve the problems in the prior art, the invention provides a carbon fiber gas cylinder nondestructive testing method and system based on a fast convolutional neural network, the defects on the carbon fiber composite gas cylinder are calibrated by utilizing an X-ray imaging and new method, the carbon fiber composite gas cylinder can stably run in an industrial production environment in real time, an alarm is given in time when the defects are detected, the detection work is effectively completed by workers, the workload is reduced, the problems of manual detection false detection and high omission factor are solved, the work efficiency of enterprises is improved, and the intelligent and digital manufacturing transformation is carried out.
On one hand, in order to achieve the technical purpose, the invention provides a carbon fiber gas cylinder nondestructive testing method based on a fast convolutional neural network, which comprises the following steps:
s1, acquiring an original X-ray image;
s2, performing data processing on the original X-ray image to obtain a processed image;
s3, extracting image features based on the processed image;
s4, obtaining a candidate region based on the image characteristics;
s5, classifying and regressing the candidate region to obtain a defect region image;
s6, extracting the outline of the image of the defect area to obtain a determined defect area, and marking the determined defect area on the original X-ray image to finish detection.
Optionally, the data processing includes:
s21, performing data amplification on the original X-ray image by adopting a data enhancement method;
and S22, carrying out data format processing on the original X-ray image after data amplification to obtain a processed image.
Optionally, the data amplification comprises flipping, rotating, and shearing.
Optionally, the S3 includes:
and pre-training the processed image by adopting a VGG16 pre-training model of a Caffe framework to obtain the image characteristics.
Optionally, the obtaining process of the candidate region is:
and performing region recommendation on the image features by adopting a region generation network to obtain the candidate regions.
Optionally, the S5 includes:
s51, selecting 128 recommended area samples from the candidate areas by adopting a suggestion box generator;
s52, pooling the recommended region samples to the same scale by using a pooling layer, and resetting the samples to be a one-dimensional vector;
and S53, initializing and connecting the front two full-connection layers by adopting the weight of the VGG16 pre-training model to obtain a defect area image.
Optionally, the S6 includes:
s61, extracting the outline of the image of the defect area, and storing the position of the outline in an array form;
s62, setting the minimum side length, the minimum perimeter and the minimum area of the defect region to obtain a set threshold;
s63, traversing the image of the defect area according to the set threshold value, and screening the defect area which does not meet the condition to obtain a determined defect area;
s64, marking the determined defect area on the original X-ray image to finish detection
On the other hand, in order to achieve the technical purpose, the invention provides a carbon fiber gas cylinder nondestructive testing system based on a fast convolution neural network, which comprises: the system comprises an image transmission module, a client core module, a server core module, a parameter configuration module and a data management module;
the image transmission module is used for acquiring an original X-ray image;
the client core module is used for creating an X-ray machine monitoring thread and an alarm device monitoring thread;
the server core module is used for receiving the image, executing an image recognition algorithm and triggering an action instruction.
The parameter configuration module is used for configuring parameters of a system client and a server;
the data management module is used for managing data in the detection process.
The invention has the following technical effects:
the defects on the carbon fiber composite gas cylinder are calibrated by utilizing X-ray imaging and a new method, the carbon fiber composite gas cylinder stably runs in an industrial production environment in real time, an alarm is given in time when the defects are detected, workers are effectively assisted to finish detection work, the workload is reduced, the problems of manual detection false detection and high omission factor are solved, the improvement of the working efficiency of enterprises is facilitated, and the transformation to intelligent and digital manufacturing is realized.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a flow chart of a nondestructive testing method for a carbon fiber gas cylinder based on a fast convolutional neural network according to an embodiment of the present invention;
FIG. 2 is a diagram illustrating the result of a carbon fiber gas cylinder nondestructive testing method based on a fast convolutional neural network according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a carbon fiber gas cylinder nondestructive testing system based on a fast convolutional neural network according to a second embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Example one
As shown in FIG. 1, the invention discloses a carbon fiber gas cylinder nondestructive testing method based on a fast convolutional neural network, which comprises the following steps:
s1, acquiring original X-ray images, wherein 51 carbon fiber gas cylinder X-ray images in VOOC2007 format manufactured by labelme are adopted as a data set.
S2, carrying out data processing on the original X-ray image to obtain a processed image;
s21, performing data amplification on the original X-ray image by adopting a data enhancement method, wherein the data amplification method comprises the following steps:
s211, turning, namely turning the original X-ray image up and down and turning the original X-ray image left and right respectively;
s212, rotating, namely sequentially rotating the original X-ray images by 15 degrees each time;
and S213, cutting, namely performing translational cutting from eight directions, namely upper, lower, left, right, upper left, lower left, upper right and lower right by taking 50 pixels as step length. To ensure that the original X-ray image does not change in size, the translated portion is replaced with a black area.
S22, carrying out data format processing on the original X-ray image after data amplification to enable the original X-ray image to be in a data format meeting requirements, and obtaining a processed image, wherein the data format processing comprises the following steps:
s221, zooming the original X-ray image after data amplification to ensure that the long side is less than or equal to 1000 and the short side is less than or equal to 600 (at least one side is equal to or less than 600);
and S222, scaling the corresponding target frame in the same scale.
For the VGG16 pre-training model using Caffe framework, the processed image is in the BGR format of 0-255, and a mean value is subtracted, so that the mean value of the pixels of the processed image is 0.
S3, extracting image features based on the processed image;
pre-training the processed image by adopting a VGG16 pre-training model of a Caffe framework, setting the learning rate of the first four layers of convolution layers as 0 to save video memory, and obtaining a C × (H/16) × (W/16) feature map: taking the output of the Conv5_3 as the picture feature, the Conv5_3 is downsampled by 16 times compared with the input, that is, the input picture size is 3 × H × W, and the feature map size is C × (H/16) × (W/16).
The VGG16 pre-trains the first two layers of the last three fully connected layers of the model to initialize partial parameters of the candidate frame classification module (ROIHead).
S4, based on the image features, adopting a region generation network to perform region recommendation on the image features to obtain candidate regions, wherein the candidate regions comprise:
s41, for each gas cylinder picture to be detected (namely an original X-ray image), calculating the probability that (H/16) X (W/16) X9 prior frames belong to the foreground and corresponding position parameters by using the characteristic diagram obtained in S3;
s42, selecting 12000 prior frames with higher probability;
s43, correcting the positions of the 12000 prior frames by using the regressed position parameters to obtain recommended regions ROIs;
s44, non-maximum suppression (NMS) is used to select 2000 candidate regions with the highest probability.
S5, classifying and regressing the candidate regions to obtain a defect region image;
s51, selecting 128 recommended region samples from the candidate regions by adopting a suggestion box generator (ProposaTargetCreator);
s52, pooling the recommended region samples to the same scale (7 x 7) by using a Pooling layer (ROI Pooling) of the VGG16 pre-training model, and resetting the samples to be a one-dimensional vector;
s53, initializing the previous two full connection layers by adopting the weight of the VGG16 pre-training model, and then connecting the two full connection layers, namely the full connection layer (FC 21) for predicting the category of the recommended region and the full connection layer (FC 84) for regressing the position to obtain a defect region image.
S6, extracting the outline of the image of the defect area to obtain a determined defect area, and marking the determined defect area on the original X-ray image to finish detection, wherein the method comprises the following steps:
s61, extracting the outline of the image of the defect area, and storing the outline position in an array form;
s62, setting the minimum side length, the minimum perimeter and the minimum area of the defect area to be determined to obtain a set threshold;
s63, traversing all defect outlines extracted from the defect area image according to a set threshold, and screening out the defect areas which do not meet the conditions (set threshold) to obtain a determined defect area;
and S64, marking the determined defect area on the original X-ray image and displaying to finish detection.
The data set adopted in the embodiment is 51 carbon fiber gas cylinder X-ray images in VOOC2007 format manufactured by labelme, 2040 processed images are obtained after data amplification, three defects of inclusion, fracture and debonding are shared, resNet50 is used as a feature extraction network, and 50 times of training are performed in total. As shown in fig. 2, (a) is an original image, and (b) is a model prediction result image, it can be seen that the Faster R-CNN method can well predict defects in a carbon fiber gas cylinder image, compared with a conventional algorithm, the Faster R-CNN method has higher confidence and detection efficiency, and the more image data, the more the method has the more remarkable advantages.
Example two
As shown in fig. 3, the invention discloses a carbon fiber gas cylinder nondestructive testing system based on a fast convolutional neural network, comprising: the system comprises an image transmission module, a client core module, a server core module, a parameter configuration module and a data management module;
the image transmission module is used for acquiring an original X-ray image and comprises buffer area management, a message frame packaging mechanism and WebSocket communication.
The client core module is used for creating an X-ray machine monitoring thread and an alarm device monitoring thread, and comprises GUI drawing, a multithreading mechanism and hardware state maintenance.
The server core module is used for receiving the image, executing an image recognition algorithm and triggering an action instruction.
The parameter configuration module is used for configuring parameters of the client and the server, including system state parameters, detection module parameters and the like.
The data management module is used for managing data generated in the operation process of the system, such as detection records, inquiry of detection conclusions, generation of detection reports and the like.
The detection system disclosed by the embodiment is used for carrying out communication between the client and the server based on Websocket and HTTP, constructing a client program by adopting C + + and Qt frames and constructing a server program by adopting node.
The detection system disclosed by the embodiment has a good test effect in the data set of the defect images of 1000 carbon fiber gas cylinders, the average detection time of each image is not more than 2s, and the occupation condition of system resources is kept stable. The test result shows that the system can stably run in various system operation environments and abnormal states, has good detection performance, and can meet the requirements of industrial production environments.
The foregoing illustrates and describes the principles, general features, and advantages of the present invention. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, which are given by way of illustration of the principles of the present invention, but that various changes and modifications may be made without departing from the spirit and scope of the invention, and such changes and modifications are within the scope of the invention as claimed. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (8)
1. The carbon fiber gas cylinder nondestructive testing method based on the fast convolutional neural network is characterized by comprising the following steps of:
s1, acquiring an original X-ray image;
s2, performing data processing on the original X-ray image to obtain a processed image;
s3, extracting image features based on the processed image;
s4, obtaining a candidate region based on the image characteristics;
s5, classifying and regressing the candidate region to obtain a defect region image;
s6, extracting the outline of the image of the defect area to obtain a determined defect area, and marking the determined defect area on the original X-ray image to finish detection.
2. The carbon fiber gas cylinder nondestructive testing method based on the fast convolutional neural network as claimed in claim 1, wherein the data processing comprises:
s21, performing data amplification on the original X-ray image by adopting a data enhancement method;
and S22, carrying out data format processing on the original X-ray image after data amplification to obtain a processed image.
3. The carbon fiber gas cylinder nondestructive testing method based on the fast convolutional neural network as claimed in claim 2, wherein the data amplification comprises flipping, rotating and shearing.
4. The carbon fiber gas cylinder nondestructive testing method based on the fast convolutional neural network of claim 1, wherein the S3 comprises:
and pre-training the processed image by adopting a VGG16 pre-training model of a Caffe framework to obtain the image characteristics.
5. The carbon fiber gas cylinder nondestructive testing method based on the fast convolutional neural network as claimed in claim 1, wherein the candidate region is obtained by the following steps:
and performing region recommendation on the image features by adopting a region generation network to obtain the candidate regions.
6. The carbon fiber gas cylinder nondestructive testing method based on the fast convolutional neural network as claimed in claim 1, wherein the S5 comprises:
s51, selecting 128 recommended area samples from the candidate areas by adopting a suggestion box generator;
s52, pooling the recommended region samples to the same scale by using a pooling layer, and resetting the samples to be a one-dimensional vector;
and S53, initializing and connecting the previous two full connection layers by adopting the weight of the VGG16 pre-training model to obtain a defect area image.
7. The carbon fiber gas cylinder nondestructive testing method based on the fast convolutional neural network as claimed in claim 1, wherein the S6 comprises:
s61, extracting the outline of the image of the defect area, and storing the position of the outline in an array form;
s62, setting the minimum side length, the minimum perimeter and the minimum area of the defect region to obtain a set threshold;
s63, traversing the image of the defect area according to the set threshold value, and screening the defect area which does not meet the condition to obtain a determined defect area;
and S64, marking the determined defect area on the original X-ray image to finish detection.
8. Carbon fiber gas cylinder nondestructive test system based on fast convolution neural network, its characterized in that includes: the system comprises an image transmission module, a client core module, a server core module, a parameter configuration module and a data management module;
the image transmission module is used for acquiring an original X-ray image;
the client core module is used for creating an X-ray machine monitoring thread and an alarm device monitoring thread;
the server core module is used for receiving the image, executing the detection task and triggering an action instruction;
the parameter configuration module is used for configuring parameters of a system client and a server;
the data management module is used for managing data in the detection process.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211221392.5A CN115494092A (en) | 2022-10-08 | 2022-10-08 | Carbon fiber gas cylinder nondestructive testing method and system based on fast convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211221392.5A CN115494092A (en) | 2022-10-08 | 2022-10-08 | Carbon fiber gas cylinder nondestructive testing method and system based on fast convolutional neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115494092A true CN115494092A (en) | 2022-12-20 |
Family
ID=84473389
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211221392.5A Pending CN115494092A (en) | 2022-10-08 | 2022-10-08 | Carbon fiber gas cylinder nondestructive testing method and system based on fast convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115494092A (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109285139A (en) * | 2018-07-23 | 2019-01-29 | 同济大学 | A kind of x-ray imaging weld inspection method based on deep learning |
| CN109767427A (en) * | 2018-12-25 | 2019-05-17 | 北京交通大学 | The detection method of train rail fastener defect |
| CN110222681A (en) * | 2019-05-31 | 2019-09-10 | 华中科技大学 | A kind of casting defect recognition methods based on convolutional neural networks |
| CN110570410A (en) * | 2019-09-05 | 2019-12-13 | 河北工业大学 | Detection method for automatically identifying and detecting weld defects |
| CN113012123A (en) * | 2021-03-13 | 2021-06-22 | 山东大学 | Classification recognition and quantitative analysis method and system for defect and damage of carbon fiber composite material |
| CN113724204A (en) * | 2021-08-03 | 2021-11-30 | 上海卫星装备研究所 | Method and system for positioning and identifying defects of aerospace composite material |
| CN114693657A (en) * | 2022-04-06 | 2022-07-01 | 重庆大学 | Intelligent detection method and system for multi-size and multi-category defects on surface of large complex structural member based on Faster R-CNN |
-
2022
- 2022-10-08 CN CN202211221392.5A patent/CN115494092A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109285139A (en) * | 2018-07-23 | 2019-01-29 | 同济大学 | A kind of x-ray imaging weld inspection method based on deep learning |
| CN109767427A (en) * | 2018-12-25 | 2019-05-17 | 北京交通大学 | The detection method of train rail fastener defect |
| CN110222681A (en) * | 2019-05-31 | 2019-09-10 | 华中科技大学 | A kind of casting defect recognition methods based on convolutional neural networks |
| CN110570410A (en) * | 2019-09-05 | 2019-12-13 | 河北工业大学 | Detection method for automatically identifying and detecting weld defects |
| CN113012123A (en) * | 2021-03-13 | 2021-06-22 | 山东大学 | Classification recognition and quantitative analysis method and system for defect and damage of carbon fiber composite material |
| CN113724204A (en) * | 2021-08-03 | 2021-11-30 | 上海卫星装备研究所 | Method and system for positioning and identifying defects of aerospace composite material |
| CN114693657A (en) * | 2022-04-06 | 2022-07-01 | 重庆大学 | Intelligent detection method and system for multi-size and multi-category defects on surface of large complex structural member based on Faster R-CNN |
Non-Patent Citations (1)
| Title |
|---|
| 陈云霁 等: "《智能计算系统》", 机械工业出版社, pages: 76 - 77 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110059694B (en) | Intelligent identification method for character data in complex scene of power industry | |
| KR20230124713A (en) | Fault detection methods, devices and systems | |
| EP3899508A1 (en) | Automated inspection system and associated method for assessing the condition of shipping containers | |
| CN113421263B (en) | Part defect detection method, device, medium and computer program product | |
| CN111597945B (en) | Target detection method, device, equipment and medium | |
| CN111222478A (en) | Construction site safety protection detection method and system | |
| CN111807183A (en) | Elevator door state intelligent detection method based on deep learning | |
| CN112560816A (en) | Equipment indicator lamp identification method and system based on YOLOv4 | |
| CN113888514A (en) | Method and device for detecting defects of ground wire, edge computing equipment and storage medium | |
| CN111368636A (en) | Object classification method and device, computer equipment and storage medium | |
| CN112700444B (en) | Bridge bolt detection method based on self-attention and central point regression model | |
| CN111667476A (en) | Cloth flaw detection method and device, electronic equipment and readable storage medium | |
| CN110910360A (en) | Power grid image positioning method and image positioning model training method | |
| CN115471487A (en) | Insulator defect detection model construction and insulator defect detection method and device | |
| CN112101138A (en) | Bridge inhaul cable surface defect real-time identification system and method based on deep learning | |
| CN115829995A (en) | Cloth flaw detection method and system based on pixel-level multi-scale feature fusion | |
| CN116883801A (en) | YOLOv8 target detection method based on attention mechanism and multi-scale feature fusion | |
| CN114373162A (en) | Dangerous area personnel intrusion detection method and system for transformer substation video monitoring | |
| CN113361473B (en) | Image processing method, model training method, image processing device, model training apparatus, storage medium, and program | |
| CN113507605A (en) | Dangerous area monitoring video compression storage method and system based on artificial intelligence | |
| CN115494092A (en) | Carbon fiber gas cylinder nondestructive testing method and system based on fast convolutional neural network | |
| CN113160209A (en) | Target marking method and target identification method for building facade damage detection | |
| CN111582012A (en) | Method and device for detecting small target ship | |
| CN110569717A (en) | partial discharge detection method, device, system, equipment and readable storage medium | |
| CN113177566B (en) | Feature extraction model training method and device and computer equipment |
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 |