CN110414633B - System and method for recognizing handwritten fonts - Google Patents

System and method for recognizing handwritten fonts Download PDF

Info

Publication number
CN110414633B
CN110414633B CN201910598723.9A CN201910598723A CN110414633B CN 110414633 B CN110414633 B CN 110414633B CN 201910598723 A CN201910598723 A CN 201910598723A CN 110414633 B CN110414633 B CN 110414633B
Authority
CN
China
Prior art keywords
neural network
convolutional neural
training
layer
flexible substrate
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.)
Active
Application number
CN201910598723.9A
Other languages
Chinese (zh)
Other versions
CN110414633A (en
Inventor
段升顺
张志恒
吴俊�
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Southeast University
Original Assignee
Southeast University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Southeast University filed Critical Southeast University
Priority to CN201910598723.9A priority Critical patent/CN110414633B/en
Publication of CN110414633A publication Critical patent/CN110414633A/en
Application granted granted Critical
Publication of CN110414633B publication Critical patent/CN110414633B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/04Architecture, e.g. interconnection topology
    • G06N3/045Combinations of networks
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/08Learning methods
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V30/00Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
    • G06V30/10Character recognition
    • G06V30/24Character recognition characterised by the processing or recognition method
    • G06V30/242Division of the character sequences into groups prior to recognition; Selection of dictionaries
    • G06V30/244Division of the character sequences into groups prior to recognition; Selection of dictionaries using graphical properties, e.g. alphabet type or font
    • G06V30/2455Discrimination between machine-print, hand-print and cursive writing
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/20Movements or behaviour, e.g. gesture recognition
    • G06V40/28Recognition of hand or arm movements, e.g. recognition of deaf sign language

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Biophysics (AREA)
  • Molecular Biology (AREA)
  • Biomedical Technology (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Computational Linguistics (AREA)
  • Software Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Multimedia (AREA)
  • Psychiatry (AREA)
  • Social Psychology (AREA)
  • Human Computer Interaction (AREA)
  • Character Discrimination (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses a system and a method for recognizing handwritten fonts. The method is based on the handwriting font recognition of the flexible wearable bracelet, the problems of environment light and resolution are solved, and the method is more stable and lower in cost; the invention uses the trained model to identify the handwritten font by means of the convolutional neural network, and has the advantages of short time consumption, high identification rate and stronger robustness; the invention can be suitable for various specific occasions by flexibly modifying the structure of the output layer and by transfer learning.

Description

System and method for recognizing handwritten fonts
Technical Field
The invention relates to a character recognition device, in particular to a system and a recognition method for recognizing handwritten characters.
Background
The earliest character recognition originated in the united states early in the 50 s as an extremely important part of industrial application, and the development of character recognition has been receiving attention.
At present, optical Character Recognition (OCR) is mainly applied to character recognition, and the general flow is as follows: the method comprises the steps of checking characters on paper through electronic equipment (a scanner, a camera and the like), processing the characters into a black-and-white image by utilizing specific binarization, then denoising, carrying out tilt correction on a document, carrying out layout analysis such as segmentation and line division on a navigation picture, then carrying out character segmentation, inputting the segmented single characters into a specific character recognition program such as a convolutional neural network, an SVM (support vector machine) and the like, and finally outputting the recognized characters.
At present, the problem of pixel distortion naturally exists in the field because the image obtained by shooting is utilized.
Disclosure of Invention
The purpose of the invention is as follows: the invention aims to provide a system and a method for recognizing a handwritten font, which solve the problems of easy environmental influence and unstable recognition rate.
The technical scheme is as follows: the system for recognizing the handwritten fonts comprises a flexible wearable bracelet, a sensor signal acquisition and processing unit and a convolutional neural network unit, wherein the flexible wearable bracelet is electrically connected with the sensor signal acquisition unit and the convolutional neural network unit.
The flexible wearable bracelet comprises a flexible substrate, wherein a pressure deformation sensor is arranged on the flexible substrate, and an insulating packaging layer covers the flexible substrate.
The invention relates to a recognition method of a system for recognizing handwritten fonts, which comprises the following steps:
(1) The flexible wearable bracelet acquires wrist dynamics parameter information at characteristic points during wrist bending movement through the change of the resistance of the sensor;
(2) The sensor signal acquisition processing unit converts the acquired wrist dynamics parameter information into an electric response signal and outputs the electric response signal;
(3) And the convolutional neural network unit trains the neural network model to obtain an optimal handwriting convolutional neural network discrimination model, then an electric response signal is input into the optimal handwriting convolutional neural network discrimination model, and recognition of the handwriting font is realized after prediction of a network middle layer.
Wherein the step (3) specifically comprises:
(a) Writing unified characters by different people in different pen holding postures, different strength and different fonts to construct a data set for training a neural network;
(b) Constructing a structure of a convolutional neural network model, setting the number of output units of a neural network to be T, and adopting a network data set to perform pre-training to initialize weight parameters;
(c) Inputting the constructed data set into the constructed convolutional neural network model, training to optimize weight parameters, and finally obtaining an optimal handwritten convolutional neural network discrimination model;
(d) And inputting the electric response signal obtained by the sensor signal acquisition and processing unit into the optimal handwriting convolution neural network discrimination model, and outputting a discrimination result.
The data set format in the step (a) is (source, target), the source is information of wrist dynamics parameters at the collected and processed characteristic points, and the target is a handwritten font coded by adopting a one-hot coding format.
And (c) in the training process, cross validation is adopted to check whether the model is over-fitted, and a cross entropy loss function softmax is adopted as an error loss function of the network to perform back propagation training until the error continuous N training period does not decrease any more, so that the optimal handwriting discrimination model is obtained.
Has the advantages that: the method is based on the handwriting font recognition of the flexible wearable bracelet, the problems of environment light watching and resolution ratio are eliminated, and the method is more stable and lower in cost; the invention uses the trained model to identify the handwritten font by means of the convolutional neural network, and has the advantages of short time consumption, high identification rate and stronger robustness; the invention can be suitable for various specific occasions, such as specific handwritten font recognition, by flexibly modifying the structure of the output layer and by transfer learning.
Drawings
FIG. 1 is a schematic block diagram of the present invention;
fig. 2 is a schematic structural diagram of a flexible wearable bracelet;
FIG. 3 is a flow chart of an electrical response acquisition module;
FIG. 4 is a flow chart of a neural network element;
fig. 5 is a flow chart of the operation of the present invention.
Detailed Description
The invention is further described below with reference to the accompanying drawings.
As shown in fig. 1-2, for example, a system for recognizing the characters "east", "south", "big", and "learning" can be used for recognizing the handwritten character, and includes:
(1) Flexible wearable bracelet
The flexible wearable bracelet can realize the acquisition of wrist dynamics parameter information of three characteristic points when the wrist bends to move through the change of sensor resistance, and for realizing above function, through placing flexible pressure/deformation sensor in three characteristic points A, B and C department at the bracelet and realize the acquisition to characteristic point information to the realization is to the acquisition of dynamic parameter when the representation wrist moves.
The flexible wearable bracelet is composed of a pressure/deformation sensor 2, a VHB double-sided hydrogel flexible substrate layer 1 and an insulating packaging layer 3, wherein the pressure/deformation sensor 2 is an MXene and multi-walled carbon nanotube (MWCNTs) three-dimensional cross-linked network.
When the pressure/deformation sensor is deformed by pressure and the like, a three-dimensional cross-linked network formed by MXene and multi-walled carbon nanotubes (MWCNTs) generates transverse cracks, so that a conductive path is reduced, the conductive resistance is increased, the resistance is increased, and the dynamic state of the wrist is represented by the degree of resistance increase.
The preparation method of the flexible wearable bracelet comprises the following steps: cutting the VHB double-sided hydrogel into a fixed cuboid, wherein the length is 18cm, the width is 1cm, and the thickness is 5mm; utilizing a laser engraving machine to engrave three characteristic points at fixed positions of a mask plate, wherein the length of the mask plate is 18.2cm, the width of the mask plate is 1.2cm, the thickness of the mask plate is 6mm, the positions of the three characteristic points are the two sides and the middle of the lower side of a wrist, and then fixing the mask plate on a VHB double-sided hydrogel flexible substrate layer; spraying the prepared MXene and multi-walled carbon nanotube dispersion liquid on a mask plate by using a spraying process, repeating for 5-6 times, and finishing spraying; spraying PVP (polyvinyl pyrrolidone) aqueous solution on the other side of the VHB double-sided hydrogel flexible substrate layer to serve as a packaging layer and an electric insulation layer; stripping the mask plate from the VHB double-sided hydrogel flexible substrate layer, and leading out lead wires at three characteristic points to serve as a conductive layer;
(2) Sensor signal acquisition processing unit
The sensor signal acquisition processing unit comprises an electric response acquisition module for acquiring electric signals sig _1, sig _2and sig _3and an electric response processing module for processing the electric response signals sig _1, sig _2and sig _3. As shown in fig. 3, when the AI chip detects that one of the electrical signals sig _1, sig _2, and sig _3is changed through the analog-to-digital converter, the analog-to-digital converter is controlled to collect the electrical signals sig _1, sig _2, and sig _3, and store the electrical signals sig _1, sig _2, and sig _3in the buffer memory, and then the collection is ended until all the electrical signals sig _1, sig _2, and sig _3are not changed, where the number of statistical collection points is W, and then the electrical signals are processed by the electrical response processing module, matrix spliced, and output the processed electrical signals.
The electric signal processing module utilizes the recorded initial voltage signal U 0 Processing the voltage signal acquired by the electric response acquisition module into (U) i -U 0 )/U 0 Wherein, U i I =1,2, …, W for the voltage signals acquired at the three characteristic points. Splicing the processed sig _1, sig _2and sig _3 into a two-dimensional matrix with the dimension of [ W, 3]]And the second-dimension first channel is a processed sig _1 electric signal, the second-dimension second channel is a processed sig _2 electric signal, and the second-dimension third channel is a processed sig _3 electric signal. Finally, 0 will be added at the end of each row until the length of each row of the matrix is filled to 2000.
(3) Convolutional neural network unit
In order to realize higher recognition rate, the invention utilizes a convolution neural network with a special structure. According to this example, the network input is [2000,3], and the softmax classification level is a 4-dimensional vector, i.e., 4 classifiers.
The structure of the convolutional neural network is as follows:
[2000,3] two-dimensional tensor input- > convolution layer- > Relu- > pooling layer- > convolution layer- > ReLu- > pooling layer- > convolution layer- > ReLu- > pooling layer- > full connecting layer 1- > full connecting layer 2- > softmax layer [4] dimension tensor output.
In order to further optimize the network structure, avoid the overfitting phenomenon and accelerate the training, the convolution layer adopts a one-dimensional convolution kernel with the length of 3; the pooling layer adopts one-dimensional average pooling with the nuclear length of 2; and constraining the model by adopting Dropout layers, wherein the Dropout value of the first convolution layer is 0.9, the Dropout value of the second convolution layer is 0.8, the Dropout value of the third convolution layer is 0.7, the Dropout value of the fourth convolution layer is 0.6, and the Dropout values of the fully-connected layer 1 and the fully-connected layer 2 are 0.5.
In the network training process, cross validation is adopted to check whether the model is over-fitted, and the specific flow is as follows: the training data is divided into one or more data sets, one part of the data set is used for training the model, the other part is used for verifying the accuracy of the model, and if the classification results are different from each other on the training set and the testing set, overfitting is generated. In this case, the samples in the database are divided into 5 sub-data sets, each word data set is used as a test sample, and the other 4 word data sets are used as training samples. Thus, 5 classifiers are obtained, 5 test results are obtained, and the average value of the 5 results is used for measuring the piezoresistive matching performance of the deep decomposable convolutional neural classification network model provided by the user.
As shown in fig. 4, the work flow of the convolutional neural network unit mainly includes the following steps: the data set used for training the neural network is constructed by writing by different people in different pen holding postures, different strength ways and different fonts for writing the four characters of east, south, big and learning. Wherein, the data format is (source, target). The dimension of the source is [2000,3], and the targets are y values in the supervised learning process, wherein the categories of all the targets are encoded in a one-hot encoding format, and the format of the targets after the one-hot encoding is correspondingly shown in the following table 1:
TABLE 1 format of target after one-hot coding
Writing character "Dong" 'Nannan' Big " "studying"
one-hot coding 1000 0100 0010 0001
And (3) constructing a structure of a roll-in neural network model, setting the number of output units of the neural network to be 4, and adopting a network data set to carry out pre-training to initialize weight parameters. And inputting the constructed data set into a constructed convolutional neural network for training to optimize the weight parameters. And (3) verifying whether the model is over-fitted by adopting cross validation, performing back propagation training as the error of the network until the error is not reduced for 5 continuous training periods, so as to obtain an optimal handwriting discrimination model, inputting the signals acquired and processed by the AI chip control analog-to-digital conversion module into the optimal handwriting convolution neural network discrimination model obtained by training in the step (3), and outputting the discrimination result.
(4) Integrated circuit design module
The integrated circuit design module is mainly used for supplying power to the flexible wearable bracelet, controlling the analog-to-digital conversion module to acquire three characteristic point signals, and realizing the processing of characteristic electric signals and the normal work of a neural network by the AI chip.
As shown in fig. 5, the working flow of the present invention is:
1) And (3) constructing a database, namely acquiring output signals sig _1, sig \u2 and sig \u3 of the flexible wearable bracelet under a specific gesture by using a sensor signal acquisition processing unit, processing the output signals into a matrix of [2000,3], using the matrix as a source of the database unit, and using a corresponding handwritten font as the database as a target. A certain number of (source, target) constructed databases are extracted, wherein each type of target is collected 50 times.
2) And training the constructed pre-trained convolutional neural network by using data in the database unit through transfer learning, and verifying whether the model is over-fitted or not through cross validation.
3) The AI chip acquires and processes electrical output signals sig _1, sig \u2 and sig \u3 from the flexible pressure sensor array through the dynamic signal acquisition and processing unit via the analog-to-digital conversion module, and then transmits data to the data training neural network discrimination unit for pressure matching identification.
4) And the value output by the neural network is the recognized handwriting font through one-hot inverse coding.

Claims (4)

1. A recognition method for a system for handwritten font recognition, comprising the steps of:
(1) The identification system is constructed and comprises a flexible wearable bracelet, a sensor signal acquisition and processing unit and a convolutional neural network unit, wherein the flexible wearable bracelet is electrically connected with the sensor signal acquisition unit and the convolutional neural network unit;
the flexible wearable bracelet comprises a flexible substrate, wherein a pressure deformation sensor is arranged on the flexible substrate, and an insulating packaging layer covers the flexible substrate;
the flexible wearable bracelet consists of a pressure deformation sensor, a VHB double-sided hydrogel flexible substrate layer and an insulating packaging layer, wherein the pressure deformation sensor is an MXene and multi-walled carbon nanotube MWCNTs three-dimensional cross-linked network;
the preparation method of the flexible wearable bracelet comprises the following steps: cutting the VHB double-sided hydrogel into fixed cuboids; carving three characteristic points at the fixed position of the mask plate by using a laser carving machine, wherein the positions of the three characteristic points are the two sides and the middle of the lower side of the wrist; fixing a mask plate on the VHB double-sided hydrogel flexible substrate layer; spraying the prepared MXene and multi-walled carbon nanotube dispersion liquid on a mask plate by using a spraying process; spraying PVP aqueous solution on the other side of the VHB double-sided hydrogel flexible substrate layer to serve as an encapsulation layer and an electric insulation layer; stripping the mask plate from the VHB double-sided hydrogel flexible substrate layer, and leading out lead wires at three characteristic points to serve as a conductive layer;
(2) The flexible wearable bracelet acquires wrist dynamics parameter information at a characteristic point during wrist bending motion through the change of the resistance of the sensor;
(3) The sensor signal acquisition processing unit converts the acquired wrist dynamics parameter information into an electric response signal and outputs the electric response signal;
(4) And the convolutional neural network unit trains the convolutional neural network model to obtain an optimal handwriting convolutional neural network discrimination model, then an electric response signal is input into the optimal handwriting convolutional neural network discrimination model, and recognition of the handwriting font is realized after prediction of a network middle layer.
2. The recognition method of the system for handwritten font recognition according to claim 1, wherein said step (4) is specifically:
(a) Writing unified characters by different people in different pen holding postures, different strength and different fonts to construct a data set for training a neural network;
(b) Constructing a structure of a convolutional neural network model, setting the number of output units of a neural network as T, and adopting a network data set to perform pre-training to initialize weight parameters;
(c) Inputting the constructed data set into the constructed convolutional neural network model, training to optimize weight parameters, and finally obtaining an optimal handwritten convolutional neural network discrimination model;
(d) And inputting the electric response signal obtained by the sensor signal acquisition processing unit into the optimal handwritten convolutional neural network discrimination model, and outputting a discrimination result.
3. The recognition method of the system for handwritten font recognition according to claim 2, wherein the data set format in step (a) is (source, target), the source is used for collecting the wrist dynamics parameter information at the processed feature points, and the target is the handwritten font coded by a one-hot coding format.
4. The method of claim 2, wherein cross validation is used to verify whether the model is over-fitted during the training process in step (c), and cross entropy loss function is used as the error loss function of the network for back propagation training until the N training periods of error no longer decrease, thereby obtaining the optimal handwriting discrimination model.
CN201910598723.9A 2019-07-04 2019-07-04 System and method for recognizing handwritten fonts Active CN110414633B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910598723.9A CN110414633B (en) 2019-07-04 2019-07-04 System and method for recognizing handwritten fonts

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910598723.9A CN110414633B (en) 2019-07-04 2019-07-04 System and method for recognizing handwritten fonts

Publications (2)

Publication Number Publication Date
CN110414633A CN110414633A (en) 2019-11-05
CN110414633B true CN110414633B (en) 2022-10-14

Family

ID=68360284

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910598723.9A Active CN110414633B (en) 2019-07-04 2019-07-04 System and method for recognizing handwritten fonts

Country Status (1)

Country Link
CN (1) CN110414633B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110864828B (en) * 2019-11-08 2021-05-28 五邑大学 Preparation method of silver nanowire/MXene flexible stress sensor
CN113903043B (en) * 2021-12-11 2022-05-06 绵阳职业技术学院 Method for identifying printed Chinese character font based on twin metric model
CN116597458B (en) * 2023-07-14 2023-09-08 厦门达宸信教育科技有限公司 Handwritten letter recognition method, system and application

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP3142545A1 (en) * 2014-05-15 2017-03-22 Nokia Technologies OY An apparatus, method and computer program for a wearable device
CN108267078A (en) * 2018-03-18 2018-07-10 吉林大学 A kind of flexible wearable resistance strain and preparation method thereof
CN108428566A (en) * 2018-01-23 2018-08-21 浙江工业大学 A kind of high efficiency preparation method of the planar miniature electrode of super capacitor of interdigital structure
CN109238522A (en) * 2018-09-21 2019-01-18 南开大学 A kind of wearable flexibility stress sensor and its preparation method and application

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8847229B2 (en) * 2011-03-24 2014-09-30 Panasonic Corporation Flexible semiconductor device, method for manufacturing the same, image display device using the same and method for manufacturing the image display device
CN106648050A (en) * 2016-09-20 2017-05-10 浙江理工大学 Multimedia computer gesture control system and control method based on flexible electronic skin
CN107085730A (en) * 2017-03-24 2017-08-22 深圳爱拼信息科技有限公司 A kind of deep learning method and device of character identifying code identification
CN107644006B (en) * 2017-09-29 2020-04-03 北京大学 Automatic generation method of handwritten Chinese character library based on deep neural network
CN109549649A (en) * 2018-11-19 2019-04-02 东南大学 A kind of movable wearable device of detection neck
CN109753566B (en) * 2019-01-09 2020-11-24 大连民族大学 Model training method for cross-domain emotion analysis based on convolutional neural network

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP3142545A1 (en) * 2014-05-15 2017-03-22 Nokia Technologies OY An apparatus, method and computer program for a wearable device
CN108428566A (en) * 2018-01-23 2018-08-21 浙江工业大学 A kind of high efficiency preparation method of the planar miniature electrode of super capacitor of interdigital structure
CN108267078A (en) * 2018-03-18 2018-07-10 吉林大学 A kind of flexible wearable resistance strain and preparation method thereof
CN109238522A (en) * 2018-09-21 2019-01-18 南开大学 A kind of wearable flexibility stress sensor and its preparation method and application

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
"CapBand: Battery-free Successive Capacitance Sensing Wristband for Hand Gesture Recognition";Hoang Truong等;《SenSys "18: Proceedings of the 16th ACM Conference on Embedded Networked Sensor Systems》;20181104;全文 *
"应用于可穿戴电子设备的柔性拉伸电路的制备";汪明月等;《中国力学大会-2017暨庆祝中国力学学会成立60周年大会论文集(C)》;20170831;全文 *

Also Published As

Publication number Publication date
CN110414633A (en) 2019-11-05

Similar Documents

Publication Publication Date Title
CN110414633B (en) System and method for recognizing handwritten fonts
CN110232341B (en) Semi-supervised learning image identification method based on convolution-stacking noise reduction coding network
CN111209910A (en) Systems, methods, and non-transitory computer-readable media for semantic segmentation
CN113761976A (en) Scene semantic analysis method based on global guide selective context network
CN112686345B (en) Offline English handwriting recognition method based on attention mechanism
Asthana et al. Handwritten multiscript numeral recognition using artificial neural networks
CN113837151B (en) Table image processing method and device, computer equipment and readable storage medium
CN112926462B (en) Training method and device, action recognition method and device and electronic equipment
CN114092938B (en) Image recognition processing method and device, electronic equipment and storage medium
CN110866900A (en) Water body color identification method and device
CN111401322A (en) Station entering and exiting identification method and device, terminal and storage medium
CN116797533B (en) Appearance defect detection method and system for power adapter
CN111079665A (en) Morse code automatic identification method based on Bi-LSTM neural network
CN115620312A (en) Cross-modal character handwriting verification method, system, equipment and storage medium
CN114220178A (en) Signature identification system and method based on channel attention mechanism
CN106897770B (en) Method and device for establishing license plate recognition model
CN116310349A (en) Large-scale point cloud segmentation method, device, equipment and medium based on deep learning
CN115311664A (en) Method, device, medium and equipment for identifying text type in image
CN112183669B (en) Image classification method, device, equipment and storage medium
CN112801960A (en) Image processing method and device, storage medium and electronic equipment
CN112949672A (en) Commodity identification method, commodity identification device, commodity identification equipment and computer readable storage medium
CN110910356A (en) Method for generating image noise detection model, image noise detection method and device
CN111967287A (en) Pedestrian detection method based on deep learning
CN220491300U (en) Air handwriting letter recognition system based on hydrogel sensor
CN108875929B (en) Font code generation method and apparatus, font code, storage medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant