CN112818970A - General detection method for steel coil code spraying identification - Google Patents
General detection method for steel coil code spraying identification Download PDFInfo
- Publication number
- CN112818970A CN112818970A CN202110120186.4A CN202110120186A CN112818970A CN 112818970 A CN112818970 A CN 112818970A CN 202110120186 A CN202110120186 A CN 202110120186A CN 112818970 A CN112818970 A CN 112818970A
- Authority
- CN
- China
- Prior art keywords
- steel coil
- image
- code spraying
- character
- detection method
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/22—Character recognition characterised by the type of writing
- G06V30/224—Character recognition characterised by the type of writing of printed characters having additional code marks or containing code marks
-
- 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
-
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/22—Image preprocessing by selection of a specific region containing or referencing a pattern; Locating or processing of specific regions to guide the detection or recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/14—Image acquisition
- G06V30/148—Segmentation of character regions
- G06V30/153—Segmentation of character regions using recognition of characters or words
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Molecular Biology (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Character Discrimination (AREA)
Abstract
The invention provides a general detection method for steel coil code spraying recognition, and belongs to the technical field of machine vision OCR detection. According to the method, an infrared monitoring camera, an image processing workstation and a specific detection and recognition algorithm are adopted to build an intelligent coil number recognition system, the detection algorithm firstly obtains a coil area through a semantic segmentation model, straight lines are uniformly radiated to the outer ring of a coil on the basis of the center of the coil and the gray distribution of the corresponding straight lines is counted, the area with characters is taken out through threshold value truncation, then the deformed characters are converted into linear characters through conversion from a polar coordinate system to a rectangular coordinate system, the subsequent processing is facilitated, on the character recognition algorithm, a single anchor point detection mechanism is determined according to the consistency of the character size through an improved yolo-v3 algorithm, meanwhile, too large and too small target frames are removed, the detection automation speed can be accelerated, the recognition accuracy can be improved, the final recognition rate can reach more than 99%, and the requirement of line production is met.
Description
Technical Field
The invention relates to the technical field of machine vision OCR detection, in particular to a general detection method for steel coil code spraying recognition.
Background
At present, steel mills increase the progress of pushing an automatic production line, reduce repetitive labor and liberate productivity. The steel coil post-processing such as leveling, crosscut and the like needs to be handled from the steel coil of specified coil number of dispatching in the reservoir area, and in the past operation, the steel coil number is mostly checked manually, and the number is recorded into the system, and then specific steel coil information is obtained by the second grade. Therefore, the people need to check repeatedly, the labor cost is wasted, and potential safety hazards also exist. The design method adopts a mode of detecting the steel coil code spraying based on the image, and the steel coil number in the steel coil code spraying is identified and then automatically sent to a secondary system through a network, so that the steel coil information is obtained for subsequent process treatment. Through solving the problem of coil of strip number discernment with machine vision, the cost only needs several thousand yuan, has saved huge cost of labor, and the price/performance ratio is high. Meanwhile, the machine identification can guarantee accuracy while working uninterruptedly, and detection errors caused by manual work negligence can be avoided.
Disclosure of Invention
The invention aims to provide a general detection method for identifying a steel coil code spraying.
The method can realize code spraying identification and send the code spraying identification to an upper layer interface by adopting a video monitoring camera, a graphic workstation and a specific identification algorithm which are relatively low in price, so that the tracking of the steel coil is realized.
The method specifically comprises the following steps:
(1) placing an infrared monitoring camera beside a steel coil conveying roller way, wherein the infrared monitoring camera shoots a steel coil code spraying side to obtain an original collected image, and transmitting the original collected image to an image processing workstation through a network for storage and detection;
(2) firstly, extracting a steel coil ROI from an original acquired image through a self-defined semantic segmentation network to obtain a binary image of a steel coil region, and performing affine transformation processing on a deformed steel coil in the original acquired image through a deformation processing module to enable the steel coil region to present an original standard circle;
(3) aiming at a standard circular steel coil image, taking a pixel point in the center of a steel coil as a reference, uniformly drawing 180 rays to the outer ring of the steel coil at intervals of 2 degrees, counting the gray distribution of all pixel points on the rays, retaining the rays when the gray is greater than a specified threshold value, and traversing all the rays to obtain a region defined by the middles of two outermost rays, namely a character region;
(4) through the conversion from a polar coordinate system to a rectangular coordinate system, the character area is stretched from an arc state to a rectangular state, namely the character is converted from a bent character to a universal linear character;
(5) and finally, training an improved target detection model based on yolo-v3, designing and adopting a single anchor point frame aiming at the size invariance of the character area, and filtering out a part of larger or smaller useless target frames according to the size of the characters corresponding to the characteristic layer to finish the identification of the steel coil code spraying.
Wherein, the infrared monitoring camera is adopted to collect the image in the step (1), so that an external light source is not needed, the maintenance cost is reduced, and the image with uniform gray scale can be normally collected day and night.
In the step (2), a steel coil area is extracted from an original collected image by using a segmentation algorithm, and the circular form of the steel coil is restored by using a deformation correction algorithm, wherein the segmentation algorithm depends on a semantic segmentation model obtained by training, the input of the model is the original collected image, and the output of the model is a binary image extracted from the steel coil area; the deformation correction reduction algorithm is to respectively count an x-axis gray level mean value and a y-axis gray level mean value according to the extracted binary image by taking the center of the steel coil as an origin, the width direction as an x axis and the height direction as a y axis, and the gray level ratio is used as the length-width ratio of deformation correction to adjust the image size.
The self-defined semantic segmentation network in the step (2) is a network model which is built by utilizing a deep learning technology and is used for semantic segmentation, and comprises a coding layer and a decoding layer, wherein the coding layer mainly adopts operations such as convolution, expansion convolution and the like to fully extract image features, the decoding layer performs up-sampling on the image features, and finally a two-value segmentation graph with the same size as an input image is obtained.
And (4) setting a threshold value between the gray level of the character area and the gray level of the background in the step (3), and adjusting according to the actual condition.
The improved target detection model based on yolo-v3 in the step (5) is that a single anchor block is adopted in yolo-v3 to carry out selective filtering of a feature layer.
In the step (5), the anchor point frames are designed to be equal in size to the characters, and meanwhile, the larger and smaller feature levels in the network structure are filtered, so that the recognition speed and accuracy can be improved at the same time.
The larger useless target frame in the step (5) means a target frame larger than (m +0.5 x min (m, n)) × (n +0.5 x min (m, n)), and the smaller useless target frame means a target frame smaller than (m-0.5 x min (m, n)) × (n-0.5 x min (m, n)), where m is an average width of characters and n is an average height of characters.
The method has the identification accuracy rate of more than 99%.
The technical scheme of the invention has the following beneficial effects:
according to the technical scheme, the infrared monitoring camera and the graphic workstation are used as hardware bases, the character recognition algorithm is added to carry out steel coil code spraying recognition, recognized steel coil numbers can be directly sent to the secondary system to call data information, the field automation requirement is met, meanwhile, the labor cost is reduced, and the recognition accuracy rate of the recognition algorithm can meet the field requirement.
Drawings
FIG. 1 is an overall scheme diagram of a general detection method for coil inkjet code identification according to the present invention;
FIG. 2 is a system hardware layout of the method of the present invention;
FIG. 3 is a schematic diagram of character region extraction according to the present invention.
Detailed Description
In order to make the technical problems, technical solutions and advantages of the present invention more apparent, the following detailed description is given with reference to the accompanying drawings and specific embodiments.
The invention provides a general detection method for identifying a code sprayed on a steel coil.
As shown in fig. 1 and fig. 2, the method can realize code spraying identification and send the code spraying identification to an upper layer interface by adopting a relatively cheap video monitoring camera, a graphic workstation and a specific identification algorithm, so as to realize steel coil tracking.
The method specifically comprises the following steps:
(1) placing an infrared monitoring camera beside a steel coil conveying roller way, wherein the infrared monitoring camera shoots a steel coil code spraying side to obtain an original collected image, and transmitting the original collected image to an image processing workstation through a network for storage and detection;
(2) the identification algorithm firstly extracts a coil ROI from an original acquired image through a self-defined semantic segmentation network to obtain a binary image of a coil area, and performs affine transformation processing on a deformed coil in the original acquired image through a deformation processing module to enable the coil area to present an original standard circle;
(3) aiming at a standard circular steel coil image, taking a pixel point in the center of a steel coil as a reference, uniformly drawing 180 rays to the outer ring of the steel coil at intervals of 2 degrees, counting the gray distribution of all pixel points on the rays, retaining the rays when the gray is greater than a specified threshold value, and traversing all the rays to obtain a region defined by the middles of two outermost rays, namely a character region;
(4) through the conversion from a polar coordinate system to a rectangular coordinate system, the character area is stretched from an arc state to a rectangular state, namely the character is converted from a bent character to a universal linear character;
(5) and finally, training an improved target detection model based on yolo-v3, designing and adopting a single anchor point frame aiming at the size invariance of the character area, and filtering out a part of larger or smaller useless target frames according to the size of the characters corresponding to the characteristic layer to finish the identification of the steel coil code spraying.
In the step (2), a steel coil area is extracted from the original collected image by using a segmentation algorithm, and the circular form of the steel coil is restored by using a deformation correction algorithm.
And (5) designing the anchor point frame to be equal in size to the character.
The method has the identification accuracy rate of more than 99%.
The following description is given with reference to specific examples.
In the specific implementation process, after the steel coil passes through the identification camera in the advancing process, the camera acquires the complete image of the steel coil, the camera adopts a 400 ten thousand pixel infrared monitoring camera, the lens adopts 8mm, the field angle is 60 degrees, and the camera is arranged at the position 1m away from the code spraying end face of the steel coil.
The method comprises the steps of extracting a steel coil region by adopting a semantic segmentation-based detection mode, realizing accurate positioning of the steel coil region by a built semantic segmentation model based on deep learning, and being suitable for extraction of field conditions such as gray level change, complex background and the like. The problem of poor effect of simple segmentation through the gray threshold is avoided.
The segmentation model extracted from the steel coil area adopts a coding-decoding structure (see table 1), firstly, the characteristics of the image are extracted through a coding layer, then, the characteristics are up-sampled through a decoding layer, and a two-value segmentation graph consistent with the input size of the model is obtained. Wherein conv2d _3x3 represents a convolutional layer of 3x3, down _ sampling represents a maximum pooling layer, conv2d _1x1 represents a convolutional layer of 1x1, and conv2d _ scaled represents a dilated convolution; up _ sampling represents upsampling, and the values in the output size represent height, width, and channel number, respectively.
The segmentation model needs to be trained through about 100 groups of marked samples to obtain optimized model parameters, images collected by a camera are normalized to 512x512 in the prediction process, then the images are input into the semantic segmentation model, the model outputs a two-value segmentation map with 512x512 size (wherein a white area is a steel coil area, and a black part is a background), the two-value segmentation map is mapped to the same size of an original image, and product operation is carried out on the two-value segmentation map and the original image to obtain a steel coil area map.
TABLE 1 semantic segmentation model structure composition
And combining the two-value segmentation graph with the original image, performing deformation calculation by using the two-value segmentation graph, and transforming the steel coil which is not completely a standard circle originally in an affine change manner to obtain a standard circle graph for subsequent processing.
Based on the processed standard circular steel coil, 180 straight lines are emitted to all directions of the circle by taking the circle center of the standard circular steel coil as a base point, a gray level distribution curve from the radius to the radius on each straight line is respectively counted, all rays with gray levels meeting a threshold value are obtained by limiting the gray level range of an area, the starting angle and the ending angle of the two outermost rays are calculated, and the area between the two angles is a character area on the steel coil, as shown in fig. 3.
The method comprises the following steps of converting a character area in an arc shape into a linear character area in a polar coordinate conversion mode, wherein the conversion coordinate is as follows:
wherein x is the abscissa under rectangular coordinate, y is the ordinate under rectangular coordinate, θ is the angle under polar coordinate, r is the radius under polar coordinate, α is the starting angle of the character region, β is the ending angle of the character region, r is the ending angle of the character regioninIs the inner diameter dimension r of the coil imageoutThe outer diameter of the steel coil image.
And then assigning the gray value of (theta, r) under the polar coordinate to the gray value of (x, y) under the corresponding rectangular coordinate system to realize the conversion of the character form, and the character is easier to detect after being converted into a linear character.
The model recognition character adopts the improved structure based on yolo-v3, and the size of the character is consistent, so that only a single anchor point frame can be set when the model is constructed, and meanwhile, the characteristic layers with the part of the distance between the character and the character are oversize and undersize are shielded, the detection speed can be effectively accelerated, and the detection accuracy is improved.
While the foregoing is directed to the preferred embodiment of the present invention, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the invention as defined in the appended claims.
Claims (8)
1. A general detection method for steel coil code spraying identification is characterized in that: the method comprises the following steps:
(1) placing an infrared monitoring camera beside a steel coil conveying roller way, wherein the infrared monitoring camera shoots a steel coil code spraying side to obtain an original collected image, and transmitting the original collected image to an image processing workstation through a network for storage and detection;
(2) firstly, extracting a steel coil ROI from an original acquired image through a self-defined semantic segmentation network to obtain a binary image of a steel coil region, and performing affine transformation processing on a deformed steel coil in the original acquired image through a deformation processing module to enable the steel coil region to present an original standard circle;
(3) aiming at a standard circular steel coil image, taking a pixel point in the center of a steel coil as a reference, uniformly drawing 180 rays to the outer ring of the steel coil at intervals of 2 degrees, counting the gray distribution of all pixel points on the rays, retaining the rays when the gray is greater than a specified threshold value, and traversing all the rays to obtain a region defined by the middles of two outermost rays, namely a character region;
(4) through the conversion from a polar coordinate system to a rectangular coordinate system, the character area is stretched from an arc state to a rectangular state, namely the character is converted from a bent character to a universal linear character;
(5) and finally, training an improved target detection model based on yolo-v3, designing and adopting a single anchor point frame aiming at the size invariance of the character area, and filtering out a part of larger or smaller useless target frames according to the size of the characters corresponding to the characteristic layer to finish the identification of the steel coil code spraying.
2. The steel coil code spraying identification universal detection method according to claim 1, characterized in that: the self-defined semantic segmentation network in the step (2) is a network model which is built by utilizing a deep learning technology and is used for semantic segmentation, and comprises a coding layer and a decoding layer, wherein the coding layer fully extracts image features by adopting convolution and expansion convolution operation, the decoding layer performs up-sampling on the image features, and finally a two-value segmentation graph with the same size as an input image is obtained.
3. The steel coil code spraying identification universal detection method according to claim 1, characterized in that: in the step (2), a steel coil area is extracted from the original collected image by using a segmentation algorithm, and the circular form of the steel coil is restored by using a deformation correction algorithm, wherein the segmentation algorithm depends on a semantic segmentation model obtained by training, the input of the model is the original collected image, and the output is a binary image extracted from the steel coil area; the deformation correction reduction algorithm is that according to the extracted binary image, the center of a steel coil is taken as an original point, the width direction is an x axis, the height direction is a y axis, the x axis gray level mean value and the y axis gray level mean value are respectively counted, the gray level ratio is taken as the length-width ratio of deformation correction, and image size adjustment is carried out.
4. The steel coil code spraying identification universal detection method according to claim 1, characterized in that: and (4) setting a threshold value between the gray level of the character area and the gray level of the background in the step (3), and adjusting according to the actual condition.
5. The steel coil code spraying identification universal detection method according to claim 1, characterized in that: the improved target detection model based on yolo-v3 in the step (5) is to adopt a single anchor block in yolo-v3 to perform selective filtering of a feature layer.
6. The steel coil code spraying identification universal detection method according to claim 1, characterized in that: and (5) designing the anchor frame to be equal in size to the character.
7. The steel coil code spraying identification universal detection method according to claim 1, characterized in that: the larger useless target frame in the step (5) refers to a target frame larger than (m +0.5 x min (m, n)) × (n +0.5 x min (m, n)), and the smaller useless target frame refers to a target frame smaller than (m-0.5 x min (m, n)) × (n-0.5 x min (m, n)), wherein m is the average width of the characters and n is the average height of the characters.
8. The steel coil code spraying identification universal detection method according to claim 1, characterized in that: the method has the identification accuracy rate of more than 99%.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110120186.4A CN112818970B (en) | 2021-01-28 | 2021-01-28 | Universal detection method for identifying steel coil spray code |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110120186.4A CN112818970B (en) | 2021-01-28 | 2021-01-28 | Universal detection method for identifying steel coil spray code |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112818970A true CN112818970A (en) | 2021-05-18 |
| CN112818970B CN112818970B (en) | 2023-07-21 |
Family
ID=75860225
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110120186.4A Active CN112818970B (en) | 2021-01-28 | 2021-01-28 | Universal detection method for identifying steel coil spray code |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112818970B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113392833A (en) * | 2021-06-10 | 2021-09-14 | 沈阳派得林科技有限责任公司 | Method for identifying type number of industrial radiographic negative image |
| CN114581911A (en) * | 2022-03-07 | 2022-06-03 | 柳州钢铁股份有限公司 | Steel coil label identification method and system |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109389121A (en) * | 2018-10-30 | 2019-02-26 | 金现代信息产业股份有限公司 | A kind of nameplate recognition methods and system based on deep learning |
| CN109636815A (en) * | 2018-12-19 | 2019-04-16 | 东北大学 | A kind of metal plate and belt Product labelling information identifying method based on computer vision |
| CN109871938A (en) * | 2019-01-21 | 2019-06-11 | 重庆大学 | A kind of components coding detection method based on convolutional neural networks |
| WO2020103676A1 (en) * | 2018-11-23 | 2020-05-28 | 腾讯科技（深圳）有限公司 | Image identification method and apparatus, system, and storage medium |
-
2021
- 2021-01-28 CN CN202110120186.4A patent/CN112818970B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109389121A (en) * | 2018-10-30 | 2019-02-26 | 金现代信息产业股份有限公司 | A kind of nameplate recognition methods and system based on deep learning |
| WO2020103676A1 (en) * | 2018-11-23 | 2020-05-28 | 腾讯科技（深圳）有限公司 | Image identification method and apparatus, system, and storage medium |
| CN109636815A (en) * | 2018-12-19 | 2019-04-16 | 东北大学 | A kind of metal plate and belt Product labelling information identifying method based on computer vision |
| CN109871938A (en) * | 2019-01-21 | 2019-06-11 | 重庆大学 | A kind of components coding detection method based on convolutional neural networks |
Non-Patent Citations (1)
| Title |
|---|
| 付城祥;仇润鹤;: "一种基于YOLOv3算法的车牌识别系统", 科技与创新, no. 03 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113392833A (en) * | 2021-06-10 | 2021-09-14 | 沈阳派得林科技有限责任公司 | Method for identifying type number of industrial radiographic negative image |
| CN114581911A (en) * | 2022-03-07 | 2022-06-03 | 柳州钢铁股份有限公司 | Steel coil label identification method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112818970B (en) | 2023-07-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110728200B (en) | Real-time pedestrian detection method and system based on deep learning | |
| CN111860348A (en) | Deep learning-based weak supervision power drawing OCR recognition method | |
| CN112818970B (en) | Universal detection method for identifying steel coil spray code | |
| CN103295009B (en) | Based on the license plate character recognition method of Stroke decomposition | |
| CN112966631A (en) | License plate detection and identification system and method under unlimited security scene | |
| CN109993154B (en) | Intelligent identification method for single-pointer sulfur hexafluoride instrument of transformer substation | |
| CN102646193A (en) | Segmentation method of character images distributed in ring shape | |
| CN110598698B (en) | Natural scene text detection method and system based on adaptive regional suggestion network | |
| CN103413271A (en) | Document image rectifying method based on local information | |
| CN113392669A (en) | Image information detection method, detection device and storage medium | |
| CN113780276A (en) | Text detection and identification method and system combined with text classification | |
| CN115810197A (en) | Multi-mode electric power form recognition method and device | |
| CN112364862A (en) | Method for matching disturbance deformation Chinese character picture based on histogram similarity | |
| CN112529001A (en) | License plate recognition method based on neural network | |
| CN111783608A (en) | Face changing video detection method | |
| CN108288037B (en) | Tire code identification system | |
| CN104346596A (en) | Identification method and identification device for QR (Quick Response) code | |
| CN111079749B (en) | End-to-end commodity price tag character recognition method and system with gesture correction | |
| CN117058182A (en) | Target detection frame optimization method for visual ranging | |
| CN112163502B (en) | Visual positioning method under indoor dynamic scene | |
| CN104268536A (en) | Face detection method through images | |
| CN111985482B (en) | Multi-mode multi-scene automatic license plate recognition method based on deep learning | |
| CN112381088A (en) | License plate recognition method and system for oil tank truck | |
| CN113888590A (en) | Video target tracking method based on data enhancement and twin network | |
| Yaohui et al. | Visual alignment system for PCB production based on yolov5 |
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 |