CN111414911A - Card number identification method and system based on deep learning - Google Patents
Card number identification method and system based on deep learning Download PDFInfo
- Publication number
- CN111414911A CN111414911A CN202010207960.0A CN202010207960A CN111414911A CN 111414911 A CN111414911 A CN 111414911A CN 202010207960 A CN202010207960 A CN 202010207960A CN 111414911 A CN111414911 A CN 111414911A
- Authority
- CN
- China
- Prior art keywords
- card number
- picture
- deep learning
- data set
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- 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
Abstract
The invention discloses a card number identification method and a system based on deep learning, and the method comprises the following steps: step S1, performing data enhancement on the original data set to obtain an expanded data set; step S2, fitting the data in the extended data set in a synthetic mode to form a training data set; step S3, performing labeling processing on the training data set obtained in the step S2 to obtain a training sample set; step S4, training the deep learning model by adopting the training sample set obtained in the step S3 to obtain a card number identification model; in step S5, the card number recognition model is used to perform card number recognition on the picture to be recognized. The method of the invention adopts the deep learning technology, not only omits the complex preprocessing and post-processing work in the traditional method, but also greatly reduces the influence degree of the external environment, not only can identify the clear and upright picture, but also can identify the inclined and relatively fuzzy picture, and greatly improves the card number identification effect.
Description
Technical Field
The invention relates to the technical fields of artificial intelligence, computer image processing and the like, in particular to a card number identification method and system based on deep learning.
Background
In recent years, internet finance is rapidly developed, and online fund transaction gradually becomes a main consumption mode in daily life. With the development of mobile interconnection, more and more commercial mobile applications are provided, mobile payment becomes one of the most mainstream payment modes, the commercial mobile applications relate to the binding of personal bank card accounts, for example, bank cards are required to be provided, the bank cards are required to be scanned, card number authentication operation is required before bank payment binding and bank business, and 16-19 bank card numbers are manually input in actual tests, so that the speed is low, errors are easy to occur, and the user experience is very poor.
At present, the technology mainly adopted by bank card identification is the traditional OCR identification technology. OCR, Optical Character Recognition, is a technique for recognizing print characters as electronic text. In the current mainstream OCR recognition technology, preprocessing such as definition judgment, layout analysis, histogram equalization, graying, binaryzation, inclination correction, character cutting and the like is firstly carried out on an image to obtain an upright and clear single character image; then using a character template; and finally, outputting a text result in a mode of template matching and the like. Because the method excessively depends on an image processing algorithm so as to perform adaptive adjustment and processing on the image in different scenes, higher requirements are placed on the external environments such as the placement position of paper, the light environment of photographing, the accuracy of a scanner and the like, and the improvement of the character recognition accuracy is limited to a great extent.
Disclosure of Invention
In order to solve the technical problem that the prior card number identification technology is influenced by external environment, so that the identification effect and the identification rate are limited, the invention provides a card number identification method based on deep learning.
The invention is realized by the following technical scheme:
the card number identification method based on deep learning comprises the following steps:
step S1, performing data enhancement on the original data set to obtain an expanded data set;
step S2, fitting the data in the extended data set in a synthetic mode to form a training data set;
step S3, performing labeling processing on the training data set obtained in the step S2 to obtain a training sample set;
step S4, training the deep learning model by adopting the training sample set obtained in the step S3 to obtain a card number identification model;
in step S5, the card number recognition model is used to perform card number recognition on the picture to be recognized.
Preferably, step S1 of the present invention specifically includes the following steps:
step S11, random rotation, translation, filling and color change processing are carried out on the original data set by adopting a data enhancement tool so as to expand the original data set;
and step S12, adding salt and pepper noise and Gaussian blur processing to the image data in the expanded original data set to obtain an expanded data set.
Preferably, step S2 of the present invention specifically includes the following steps:
step S21, sequentially taking out 1 picture from the extended data set and transversely splicing the 1 picture with 4 randomly selected pictures together, and performing enlargement or reduction processing on the spliced pictures, and then synthesizing the pictures into an 800 × 100 background picture;
and S22, repeating the step S21 until all the pictures are taken out in sequence for splicing and fitting, and generating a training data set.
Preferably, step S5 of the present invention specifically includes the following steps:
step S51, the card number in the input original picture is positioned and extracted to obtain the picture to be identified:
and step S52, recognizing the picture to be recognized by using the card number recognition model.
Preferably, step S51 of the present invention specifically includes the following steps:
step S511, inputting an original picture to be identified, extracting an interest region in the picture, and carrying out graying processing on the extracted interest region;
step S512, performing morphological top hat operation on the grayed picture to highlight the card number row;
step S513, calculating a Sobel value in the X direction and protruding the card number edge;
step S514, performing morphology closing operation after the card number edge is protruded, so that the card number area is adhered;
step S515, performing cyclic binarization processing after morphological closing operation until the white pixel point proportion is between 0.2 and 0.6;
step S516, performing mean filtering on the image after the binarization processing to remove isolated pixel blocks, and then performing a second morphological closing operation;
step S517, after the second morphological closing operation, calculating a transverse projection, and denoising the rows according to 1.5 times of a transverse average pixel value; then searching coordinates y1 and y2 at the upper end and the lower end of the card number row area;
step S518, after the line denoising treatment, removing useless pixels and interference pixels on the upper side and the lower side of the card number line; calculating longitudinal projection, and removing independent small pixel blocks; then the coordinates x1 and x2 at the left and right ends of the card number array area are searched
Step S519, an initial card number area is formed by x1, x2, y1 and y2, and the initial card number area is finely adjusted to obtain a card number area; and cutting the card number area and synthesizing the picture to be identified.
Preferably, the deep learning model used in step S4 of the present invention is the dark net-yolov3 model.
Preferably, step S3 of the present invention specifically includes the following steps:
step S31, generating one-to-one corresponding label data from the picture data in the training data set;
a step S32 of converting the format of the tag data generated at the step S31 into a yolo tag data format;
step S33, the picture data in the training data set and the corresponding label data in the yolo label data format form a training sample set.
Preferably, the format of the yolo tag data in step S32 of the present invention is: object-class > < x _ center > < y _ center > < width > < height >;
wherein object-class represents the object type of the annotation, x _ center and < y _ center > represent the normalized center coordinates of the annotation frame, and < width > and < height > represent the normalized width and height of the annotation frame.
On the other hand, the invention also provides a card number identification system based on deep learning, which comprises a server, wherein the server is used for receiving the collected picture with the card number, and identifying and outputting the card number on the picture by adopting the card number identification method based on deep learning.
Preferably, the card number identification system of the present invention further includes at least one client, where the at least one client is configured to collect the picture data and send the picture data to the server, and the at least one client is further configured to receive an identification result fed back by the server.
Compared with the prior art, the invention has the following advantages and beneficial effects:
compared with the traditional OCR recognition technology, the method provided by the invention adopts the deep learning technology, so that the complex preprocessing and post-processing work in the traditional method is saved, and the model training time is reduced from several days to several hours. And the influence degree of the external environment is greatly reduced, not only can clear and correct pictures be identified, but also inclined and relatively fuzzy pictures can be identified, and the card number identification effect is greatly improved.
The method can be applied to the identification of the bank card number, and can also be applied to the application fields of license plate identification, certificate number identification, bill number identification and the like. In addition, the invention adopts the deep learning technology, the response time of a single identification request is about 100ms on average, the identification precision on the existing verification set reaches 99 percent on average, and meanwhile, the invention can be used for any program access of a mobile terminal and a PC terminal by carrying background structure API (Http mode) service, thereby providing service for the society with higher, faster and more complete service efficiency.
Drawings
The accompanying drawings, which are included to provide a further understanding of the embodiments of the invention and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the invention and together with the description serve to explain the principles of the invention. In the drawings:
FIG. 1 is a schematic flow chart of the method of the present invention.
Fig. 2 is a gray scale image of the interest area of the bank card picture to be identified according to the invention.
Fig. 3 is an image in which the card number row is highlighted for the grayscale image of fig. 2.
Fig. 4 is an image of the image of fig. 3 with the card number edge protruding.
Fig. 5 shows the result of the first morphological closing operation performed on the image of fig. 4.
Fig. 6 shows the result of the first binarization processing performed on fig. 5.
Fig. 7 shows the result of the cyclic binarization process of fig. 6.
Fig. 8 is an image after mean filtering of fig. 7.
Fig. 9 shows the result of the second morphological closing operation performed on fig. 8.
Fig. 10 shows the line denoising processing result performed on fig. 9.
Fig. 11 shows the result of the operation of removing the unnecessary pixels and the interference pixels from fig. 10.
Fig. 12 shows the result of the small pixel block clearing operation performed on fig. 11.
FIG. 13 is a diagram illustrating cropping and synthesizing a card number region into a picture to be recognized.
FIG. 14 is a schematic diagram of the system of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail below with reference to examples and accompanying drawings, and the exemplary embodiments and descriptions thereof are only used for explaining the present invention and are not meant to limit the present invention.
Example 1
The embodiment provides a card number identification method based on deep learning.
Specifically, as shown in fig. 1, the method of this embodiment specifically includes the following steps:
step S1, performing data enhancement on the original data set to obtain an expanded data set.
The embodiment adopts a keras self-contained data enhancement tool to carry out the following enhancement operation on a given data set containing a thousand-card-number picture: (1) rotating (randomly rotating the image within the range of 0-15 degrees); (2) translating; (3) filling; (4) color change (changing the overall color of the picture by cheaply setting the value of the picture color channel) processing to expand the original data set; then, salt and pepper noise and Gaussian blur processing are added to the image data in the expanded original data set, and an expanded data set of ten thousand pictures is generated.
And step S2, fitting the data in the extended data set in a synthetic mode to form a training data set.
The synthesis idea of this embodiment is: sequentially taking out 1 picture from the extended data set and transversely splicing the 1 picture with 4 randomly selected pictures together, carrying out amplification or reduction processing on the spliced pictures, and then synthesizing the pictures into an 800 × 100 background picture; and repeating the synthesis operation until all the pictures are taken out in sequence for splicing and fitting to generate a training data set.
Step S3, performing labeling processing on the training data set obtained in step S2 to obtain a training sample set.
The deep learning model adopted in this embodiment is a dark net-yolov3 model. yolov3 has borrowed residual error network structure, forms deeper network level to and the multiscale detects, has promoted mAP and small object detection effect.
In this embodiment, since the numbers in the pictures in the training data set are the same in size and position in the pictures, the tag data can be generated according to this feature. The specific process is as follows:
(1) generating one-to-one corresponding label data from the picture data in the training data set; (2) the generated tag data is an xml file in the form of Pascal-voc, and therefore, the format of the generated tag data needs to be converted into the yolo tag data format. Namely, the picture data in the training data set and the corresponding label data in the yolo label data format form a training sample set.
Specifically, the data format of the yolo tag in this embodiment is: object-class > < x _ center > < y _ center > < width > < height >;
wherein object-class represents the object type of the annotation, < x _ center > and < y _ center > represent the normalized center coordinates of the annotation frame, and < width > and < height > represent the normalized width and height of the annotation frame.
The calculation method comprises the following steps:
x-center＝(x1+x2)/(2*img-w)
y_center＝(y1+y2)/(2*img_h)
width＝(x2-x1)/img-w
Height＝(y2-y1)/img_h
where img _ w, img _ h are the width and height of the picture.
And S4, training the deep learning model by adopting the training sample set obtained in the step S3 to obtain a card number identification model.
In this embodiment, before model training, training configuration needs to be performed. After the file configuration is completed, model training begins. And finally, testing on a given test set of 500 real bank card samples to achieve 99% of identification accuracy.
That is, the present embodiment implements training and testing of the card number identification model through the above steps S1-S4, and then identifies the actual bank card to be tested by using the card number identification model obtained through training. In another preferred embodiment, the card number recognition model of the present embodiment may also be applied to the technical fields of license plate recognition, certificate number recognition, bill number, and the like.
In step S5, the card number recognition model is used to perform card number recognition on the actual picture to be recognized.
In this embodiment, the card identifying the picture to be identified by using the card number identification model further includes:
(1) positioning and extracting the card number in the input original picture to obtain a picture to be identified:
the specific process of locating and extracting the card number in the embodiment is as follows:
step one, inputting a bank card picture to be identified, then extracting an interest area (an image area containing a bank card number) in the bank card picture, and finally performing gray processing on the extracted interest area to obtain a gray image shown in fig. 2.
And step two, performing morphological top hat operation on the grayed picture to highlight the card number row, as shown in fig. 3.
Step three, calculating the Sobel value in the X direction, and protruding the card number edge, as shown in FIG. 4.
And step four, performing a first morphological closing operation after the card number edge is protruded to enable the card number area to be adhered, and preparing for the following pixel statistics as shown in fig. 5.
And fifthly, circularly binarizing the picture after the first morphological closing operation until the white pixel point proportion is between 0.2 and 0.6. In this process, after the first binarization processing, as shown in fig. 6, at this time, because the threshold value of binarization is too high, the information loss is serious, and the white pixel occupation ratio at this time is 0.0650595, the threshold value is subjected to a reduction cyclic binarization processing until a condition is met, so that the final result of the cyclic binarization processing shown in fig. 7 is obtained. The card number rows can be positioned according to the characteristic that the white pixels of the card number rows are the most after the cyclic binarization processing, and the card number rows can be positioned according to the adhesion of the card number pixels.
And step six, performing mean filtering on the image after the binarization processing to remove some isolated pixel blocks, and performing a second morphological closing operation after the operation is performed as shown in fig. 8, as shown in fig. 9.
Step seven, after the second morphological closing operation, calculating a transverse projection, and performing denoising processing on the rows according to 1.5 times of the transverse average pixel value, as shown in fig. 10; then searching coordinates y1 and y2 at the upper end and the lower end of the card number row area;
step eight, removing useless pixels and interference pixels on the upper side and the lower side of the card number row after the row denoising treatment, as shown in fig. 11; computing the longitudinal projection, eliminating the independent small pixel blocks, as shown in fig. 12; the coordinates x1 and x2 at the left and right ends of the card number array area are then found.
Step nine, an initial card number region is formed by x1, x2, y1 and y2, and the card number region is obtained by fine adjustment of the initial card number region (the x1, x2, y1 and y2 are slightly increased); the card number region is cut and synthesized into a picture to be recognized, as shown in fig. 13.
(2) And identifying the picture to be identified by using the card number identification model.
Example 2
The embodiment provides a card number identification system based on deep learning. The system includes a server for executing the deep learning-based card number identification method set forth in embodiment 1 above.
As shown in fig. 14, the server of this embodiment is configured to receive the collected picture with the card number, and identify and output the card number on the picture by using the card number identification method based on deep learning.
The system of the embodiment further comprises at least one client, wherein the at least one client is used for collecting the picture data and sending the picture data to the server, and the at least one client is also used for receiving the identification result fed back by the server.
The client in this embodiment may be a mobile device terminal such as a mobile phone and a PAD, or an equipment terminal such as a computer PC.
The system of the embodiment can simultaneously identify the card number of the bank card images uploaded by a plurality of clients, thereby greatly improving the identification efficiency and the identification precision of the card number of the bank card.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The above-mentioned embodiments are intended to illustrate the objects, technical solutions and advantages of the present invention in further detail, and it should be understood that the above-mentioned embodiments are merely exemplary embodiments of the present invention, and are not intended to limit the scope of the present invention, and any modifications, equivalent substitutions, improvements and the like made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (10)
1. The card number identification method based on deep learning is characterized by comprising the following steps:
step S1, performing data enhancement on the original data set to obtain an expanded data set;
step S2, fitting the data in the extended data set in a synthetic mode to form a training data set;
step S3, performing labeling processing on the training data set obtained in the step S2 to obtain a training sample set;
step S4, training the deep learning model by adopting the training sample set obtained in the step S3 to obtain a card number identification model;
in step S5, the card number recognition model is used to perform card number recognition on the picture to be recognized.
2. The deep learning-based card number identification method according to claim 1, wherein the step S1 specifically includes the following steps:
step S11, random rotation, translation, filling and color change processing are carried out on the original data set by adopting a data enhancement tool so as to expand the original data set;
and step S12, adding salt and pepper noise and Gaussian blur processing to the image data in the expanded original data set to obtain an expanded data set.
3. The deep learning-based card number identification method according to claim 1, wherein the step S2 specifically includes the following steps:
step S21, sequentially taking out 1 picture from the extended data set and transversely splicing the 1 picture with 4 randomly selected pictures together, and performing enlargement or reduction processing on the spliced pictures, and then synthesizing the pictures into an 800 × 100 background picture;
and S22, repeating the step S21 until all the pictures are taken out in sequence for splicing and fitting, and generating a training data set.
4. The deep learning-based card number identification method according to claim 1, wherein the step S5 specifically includes the following steps:
step S51, the card number in the input original picture is positioned and extracted to obtain the picture to be identified:
and step S52, recognizing the picture to be recognized by using the card number recognition model.
5. The deep learning-based card number identification method according to claim 4, wherein the step S51 specifically includes the following steps:
step S511, inputting an original picture to be identified, extracting an interest region in the picture, and carrying out graying processing on the extracted interest region;
step S512, performing morphological top hat operation on the grayed picture to highlight the card number row;
step S513, calculating a Sobel value in the X direction and protruding the card number edge;
step S514, performing morphology closing operation after the card number edge is protruded, so that the card number area is adhered;
step S515, performing cyclic binarization processing after morphological closing operation until the white pixel point proportion is between 0.2 and 0.6;
step S516, performing mean filtering on the image after the binarization processing to remove isolated pixel blocks, and then performing a second morphological closing operation;
step S517, after the second morphological closing operation, calculating a transverse projection, and denoising the rows according to 1.5 times of a transverse average pixel value; then searching coordinates y1 and y2 at the upper end and the lower end of the card number row area;
step S518, after the line denoising treatment, removing useless pixels and interference pixels on the upper side and the lower side of the card number line; calculating longitudinal projection, and removing independent small pixel blocks; then the coordinates x1 and x2 at the left and right ends of the card number array area are searched
Step S519, an initial card number area is formed by x1, x2, y1 and y2, and the initial card number area is finely adjusted to obtain a card number area; and cutting the card number area and synthesizing the picture to be identified.
6. The deep learning-based card number identification method according to any one of claims 1-5, characterized in that the deep learning model adopted in the step S4 is the dark net-yolov3 model.
7. The deep learning-based card number identification method according to claim 6, wherein the step S3 specifically includes the following steps:
step S31, generating one-to-one corresponding label data from the picture data in the training data set;
a step S32 of converting the format of the tag data generated at the step S31 into a yolo tag data format;
step S33, the picture data in the training data set and the corresponding label data in the yolo label data format form a training sample set.
8. The deep learning-based card number identification method according to claim 7, characterized in that the yolo tag data format in step S32 is: object-class > < x _ center > < y _ center > < width > < height >;
wherein object-class represents the object type of the annotation, < x _ center > and < y _ center > represent the normalized center coordinates of the annotation frame, and < width > and < height > represent the normalized width and height of the annotation frame.
9. The deep learning-based card number identification system is characterized by comprising a server, wherein the server is used for receiving a collected picture with a card number, and identifying and outputting the card number on the picture by adopting the deep learning-based card number identification method of any one of claims 1 to 6.
10. The deep learning based card number identification system of claim 5 further comprising at least one client, wherein the at least one client is used for collecting picture data and sending the picture data to the server, and the at least one client is also used for receiving the identification result fed back by the server.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010207960.0A CN111414911A (en) | 2020-03-23 | 2020-03-23 | Card number identification method and system based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010207960.0A CN111414911A (en) | 2020-03-23 | 2020-03-23 | Card number identification method and system based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111414911A true CN111414911A (en) | 2020-07-14 |
Family
ID=71491429
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010207960.0A Pending CN111414911A (en) | 2020-03-23 | 2020-03-23 | Card number identification method and system based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111414911A (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101246551A (en) * | 2008-03-07 | 2008-08-20 | 北京航空航天大学 | Fast license plate locating method |
| US20190180467A1 (en) * | 2017-12-11 | 2019-06-13 | Beijing Didi Infinity Technology And Development Co., Ltd. | Systems and methods for identifying and positioning objects around a vehicle |
| CN110070552A (en) * | 2019-05-07 | 2019-07-30 | 西南石油大学 | A kind of rock image porosity type recognition methods based on semantic segmentation |
| WO2019174130A1 (en) * | 2018-03-14 | 2019-09-19 | 平安科技（深圳）有限公司 | Bill recognition method, server, and computer readable storage medium |
| CN110569839A (en) * | 2019-08-09 | 2019-12-13 | 河海大学常州校区 | Bank card number identification method based on CTPN and CRNN |
| CN110717493A (en) * | 2019-09-16 | 2020-01-21 | 浙江大学 | License plate recognition method containing stacked characters based on deep learning |
-
2020
- 2020-03-23 CN CN202010207960.0A patent/CN111414911A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101246551A (en) * | 2008-03-07 | 2008-08-20 | 北京航空航天大学 | Fast license plate locating method |
| US20190180467A1 (en) * | 2017-12-11 | 2019-06-13 | Beijing Didi Infinity Technology And Development Co., Ltd. | Systems and methods for identifying and positioning objects around a vehicle |
| WO2019174130A1 (en) * | 2018-03-14 | 2019-09-19 | 平安科技（深圳）有限公司 | Bill recognition method, server, and computer readable storage medium |
| CN110070552A (en) * | 2019-05-07 | 2019-07-30 | 西南石油大学 | A kind of rock image porosity type recognition methods based on semantic segmentation |
| CN110569839A (en) * | 2019-08-09 | 2019-12-13 | 河海大学常州校区 | Bank card number identification method based on CTPN and CRNN |
| CN110717493A (en) * | 2019-09-16 | 2020-01-21 | 浙江大学 | License plate recognition method containing stacked characters based on deep learning |
Non-Patent Citations (2)
| Title |
|---|
| SANGDOO YUN等: "\"CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Features\"", pages 2 * |
| 刘开心_8A6C: ""卷积神经网络实现多个数字识别"", 《HTTPS://WWW.JIANSHU.COM/P/79265078C95B》, pages 1 - 2 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Gatos et al. | Automatic table detection in document images | |
| CN109766890B (en) | Information identification method, equipment and system | |
| CN104008401A (en) | Method and device for image character recognition | |
| CN110598566A (en) | Image processing method, device, terminal and computer readable storage medium | |
| CN112070649A (en) | Method and system for removing specific character string watermark | |
| CN112307919A (en) | Improved YOLOv 3-based digital information area identification method in document image | |
| CN109741273A (en) | A kind of mobile phone photograph low-quality images automatically process and methods of marking | |
| CN114842524B (en) | Face false distinguishing method based on irregular significant pixel cluster | |
| CN112001200A (en) | Identification code identification method, device, equipment, storage medium and system | |
| CN113792659B (en) | Document identification method and device and electronic equipment | |
| CN113239807B (en) | Method and device for training bill identification model and bill identification | |
| CN112508000B (en) | Method and equipment for generating OCR image recognition model training data | |
| US20140086473A1 (en) | Image processing device, an image processing method and a program to be used to implement the image processing | |
| CN111199240A (en) | Training method of bank card identification model, and bank card identification method and device | |
| CN111507119A (en) | Identification code identification method and device, electronic equipment and computer readable storage medium | |
| CN111507181B (en) | Correction method and device for bill image and computer equipment | |
| CN112749696A (en) | Text detection method and device | |
| CN115035533B (en) | Data authentication processing method and device, computer equipment and storage medium | |
| CN111008635A (en) | OCR-based multi-bill automatic identification method and system | |
| CN111462388A (en) | Bill inspection method and device, terminal equipment and storage medium | |
| CN111414911A (en) | Card number identification method and system based on deep learning | |
| CN110428264A (en) | Fake method, device, equipment and medium are tested in identification based on dot matrix screen antifalsification label | |
| CN115565178A (en) | Font identification method and apparatus | |
| CN112308141B (en) | Scanning bill classification method, system and readable storage medium | |
| CN112101058B (en) | Automatic identification method and device for test paper bar code |
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 |