CN113191220A - Deep learning-based double-layer license plate recognition method - Google Patents
Deep learning-based double-layer license plate recognition method Download PDFInfo
- Publication number
- CN113191220A CN113191220A CN202110406424.8A CN202110406424A CN113191220A CN 113191220 A CN113191220 A CN 113191220A CN 202110406424 A CN202110406424 A CN 202110406424A CN 113191220 A CN113191220 A CN 113191220A
- Authority
- CN
- China
- Prior art keywords
- layer
- license plate
- double
- feature map
- deep learning
- 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
- 238000000034 method Methods 0.000 title claims abstract description 30
- 238000013135 deep learning Methods 0.000 title claims abstract description 19
- 238000011176 pooling Methods 0.000 claims abstract description 29
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 13
- 238000000605 extraction Methods 0.000 claims abstract description 10
- 238000004364 calculation method Methods 0.000 claims abstract description 7
- 238000010606 normalization Methods 0.000 claims abstract description 3
- 239000010410 layer Substances 0.000 claims description 114
- 230000006870 function Effects 0.000 claims description 15
- 230000008569 process Effects 0.000 claims description 9
- 239000002356 single layer Substances 0.000 claims description 8
- 230000004913 activation Effects 0.000 claims description 7
- 210000002569 neuron Anatomy 0.000 claims description 6
- 238000004422 calculation algorithm Methods 0.000 description 4
- 238000004590 computer program Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 230000011218 segmentation Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 239000002355 dual-layer Substances 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
- G06V20/54—Surveillance or monitoring of activities, e.g. for recognising suspicious objects of traffic, e.g. cars on the road, trains or boats
-
- 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
- G06V20/625—License plates
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
- Character Discrimination (AREA)
Abstract
The invention discloses a deep learning-based double-layer license plate recognition method, which is characterized in that the double-layer license plate is recognized, and the recognized double-layer license plate is input into a constructed convolutional neural network; performing feature extraction on the double-layer license plate through a backbone network of the convolutional neural network and performing normalization processing on input data of a partial network, wherein the feature extraction backbone network comprises eight convolutional layers and four pooling layers; dividing the feature map obtained by the network calculation into an upper feature map and a lower feature map according to the height of the feature map, obtaining the feature map with the width of a first preset width by the upper feature map through mean pooling, obtaining the feature map with the width of a second preset width by the lower feature map, combining the two processed feature maps into the feature map with the width of a third preset width, obtaining an output result through a layer of full connection layer, and finally combining n output results to obtain a final double-layer license plate recognition result.
Description
Technical Field
The invention relates to the field of image recognition and machine learning, in particular to a deep learning-based double-layer license plate recognition method.
Background
License plate identification is one of important components in modern intelligent traffic systems, and is very widely applied. The method is based on technologies such as digital image processing, mode recognition and computer vision, and analyzes vehicle images or video sequences shot by a camera to obtain a unique license plate number of each vehicle, so that the recognition process is completed.
In China, various license plate formats are specified according to different vehicle types and purposes, wherein the license plates include single-layer license plates of common cars and double-layer license plates of large trucks and motorcycles.
The common existing license plate algorithm only aims at the recognition of a single-layer blue license plate of a car and cannot be directly used for recognizing a double-layer yellow license plate of a large truck or a motorcycle; and the traditional license plate recognition algorithm comprises: character segmentation and character recognition, wherein the detected license plate region needs to be subjected to character segmentation, then each character is recognized, and finally the characters are combined together. The data marking used by the method is complex to obtain, and the reasoning speed has a space for improving.
Based on the defects, the invention mainly aims at the recognition of the double-layer yellow license plate, adopts the deep learning neural network algorithm and realizes the purpose of simple, convenient and efficient double-layer license plate recognition.
Disclosure of Invention
Aiming at the technical problem, the invention constructs a lightweight convolutional neural network for extracting features, uses the global features of a double-layer license plate to be processed into one-layer feature distribution, and uses two full-connection layers to carry out global classification on license plate information. That is, the present invention is directed to solving at least one of the problems occurring in the prior art. Therefore, the invention discloses a deep learning-based double-layer license plate recognition method, which comprises the following steps of:
step 1: identifying the double-layer license plate, and inputting the identified double-layer license plate into the constructed convolutional neural network;
step 2: performing feature extraction on the double-layer license plate through a backbone network of the convolutional neural network and performing normalization processing on input data of a partial network, wherein the feature extraction backbone network comprises eight convolutional layers and four pooling layers;
and step 3: dividing the feature map obtained by the network calculation into an upper layer feature map and a lower layer feature map according to the height of the feature map, obtaining the feature map with the width of a first preset width by the upper layer feature map through mean pooling, obtaining the feature map with the width of a second preset width by the lower layer, merging the two processed feature maps into the feature map with the width of a third preset width, and processing the features of the upper layer and the lower layer into single-layer features;
and 4, step 4: the processed single-layer features with the third preset width are equally divided into n parts, each part of the evenly divided features is reconstructed to obtain a two-dimensional tensor, the output dimension is reduced to 128 through a full connection layer, then an output result is obtained through the full connection layer, and finally the n parts of the output results are combined to obtain a final double-layer license plate recognition result.
Further, convolutional neural network implementations are done using the deep learning framework PyTorch.
Further, the target loss function of the convolutional neural network adopts AM-softmax, and the formula is as follows:
and cos (theta) is the area of the calculation sample on the category, cos (theta) -m is the interval of at least a super parameter m in the area between the categories, and s (cos (theta) -m) introduces the super parameter s and expands the interval of the cos value by s times.
Still further, the step 2 further comprises: the first layer convolution to the sixth layer convolution of the feature extraction backbone network are respectively composed of 64, 128, 256, 512 and 512 convolution kernels of 3x3, the step length stride is 1, the padding is 1, the seventh layer convolution is composed of 512 convolution kernels of 3x3, the step length stride is 1, the eighth layer convolution is composed of 512 convolution kernels of 2x2, the step length stride is 1, each layer convolution adopts the following formula to normalize input data:
in the training process, a batch random gradient descending mode is adopted, wherein E [ x (k) ] refers to the average value of each batch of training data neurons x (k); the denominator is a standard deviation of the activation of each data neuron x (k).
Still further, the step 2 further comprises: the activation function used after each layer of convolution is the ReLU function:
a=g(x)=max(0,z)
the four pooling layers are maximum pooling layers Max padding, the size of sliding windows of the first pooling layer and the second pooling layer is 2x2, the step length stride is 2, the step length variance of an element in each window is 1, the size of sliding windows of the third pooling layer and the fourth pooling layer is 2x2, the step length stride of rightward sliding is 2, the step length stride of downward sliding is 1, and the step length variance of the element in each window is 1; the four largest pooling layers are behind the first, second, fourth, and sixth convolution layers, respectively.
Still further, the step 3 further comprises: the third preset width is the sum of the first preset width and the second preset width.
Still further, the step 3 further comprises: the first preset width takes the value of 2, and the second preset width is 5.
Furthermore, the double-layer license plate is of a double-layer structure, wherein the upper layer is a combination of one letter of a Chinese character and the lower layer is a combination of five letters and numbers.
The invention further discloses an electronic device comprising:
a processor; and the number of the first and second groups,
a memory for storing executable instructions of the processor;
wherein the processor is configured to execute the above-described deep learning-based dual-layer license plate recognition method via execution of the executable instructions.
The invention also discloses a computer readable storage medium, on which a computer program is stored, wherein the computer program is executed by a processor to realize the deep learning-based double-layer license plate recognition method.
Compared with the prior art, the invention has the beneficial effects that: firstly, constructing a lightweight convolutional neural network for extracting features, processing the global features of a double-layer license plate into a one-layer feature distribution, and performing global classification on license plate information by using two fully-connected layers; and secondly, the purpose of reducing the class inner distance and increasing the class interval is achieved by improving the Softmax function, and the function can obtain a better effect under the condition of regularization of the features and the weight. The network uses a weighted sum mode in the training process to particularly pay attention to the problem of sample long tail distribution of the first Chinese character of the license plate.
Drawings
The invention will be further understood from the following description in conjunction with the accompanying drawings. The components in the figures are not necessarily to scale, emphasis instead being placed upon illustrating the principles of the embodiments. In the drawings, like reference numerals designate corresponding parts throughout the different views.
FIG. 1 is a block diagram of the overall neural network architecture provided in one embodiment of the present invention;
FIG. 2 is a diagram of a feature extraction backbone network in one embodiment of the invention;
fig. 3 is a block diagram of an output network in one embodiment of the invention.
Detailed Description
Example one
In order to realize the purpose of simple and efficient double-layer license plate recognition, a lightweight convolutional neural network is constructed to be used for extracting features, the global features of the double-layer license plate are processed into one-layer feature distribution, and two full-connection layers are used for globally classifying license plate information. The network implementation is completed by using a deep learning framework PyTorch, the target loss function of the network adopts AM-softmax, and the formula is as follows:
cos (θ) is the area of the computed samples over the classes, cos (θ) -m is the interval between classes requiring the area to be at least over the parameter m, s*(cos (θ) -m) introduction of the hyper-parameter s, expanding the interval of cos values by a factor of s, since the interval of cos values is [0,1 ]]The value of this interval is too small to effectively distinguish the differences. After the expansion of s times, the difference of distribution can be improved, obvious Martian effect is generated, and the convergence speed is improved.
The function is based on the improvement of the Softmax function, the purpose of reducing the class inner distance and increasing the class interval is achieved, and the function can achieve a better effect under the condition of regularization of the features and the weight. The network uses a weighted sum mode in the training process to particularly pay attention to the problem of sample long tail distribution of the first Chinese character of the license plate.
The specific network structure and implementation are as follows:
the overall structure of the neural network structure is shown in fig. 1, wherein the feature extraction backbone network comprises eight convolutional layers and four pooling layers, as shown in fig. 2.
Wherein the first layer convolution to the sixth layer convolution are respectively composed of 64, 128, 256, 512 convolution kernels of 3x3 (and step size stride of 1, padding of 1), the seventh layer convolution is composed of 512 convolution kernels of 3x3 (and step size stride of 1), the eighth layer convolution is composed of 512 convolution kernels of 2x2 (and step size stride of 1), each layer convolution has a convolution using the BatchNormalization algorithm (as shown below),
the formula is used for normalizing the input data of the hierarchical network. A batch random gradient descending mode is adopted in the training process, and E [ x (k) ] refers to the average value of each batch of training data neurons x (k); the denominator is then a standard deviation of the activation of each batch of data neurons x (k).
So as to improve the robustness of the model during training, prevent overfitting and accelerate the training of the model. The activation function used after each layer of convolution is the ReLU function (shown below).
a=g(x)=max(0,z)
The four pooling layers are maximum pooling layers Max padding, the sliding windows of the first pooling layer and the second pooling layer are 2x2, the step length stride is 2, the step length variance of an element in each window is 1, the sliding windows of the third pooling layer and the fourth pooling layer are also 2x2, the step length stride of rightward sliding is 2, the step length stride of downward sliding is 1, and the step length variance of the element in each window is also 1; the four largest pooling layers are behind the first, second, fourth, and sixth convolution layers, respectively.
The above is a detailed composition of the backbone network, and a series of convolution characteristic graphs are output from the backbone network for the next calculation.
And then according to the characteristics of the double-layer license plate, dividing the feature graph obtained by the network calculation into an upper feature graph and a lower feature graph according to the height of the feature graph, obtaining the feature graph with the width of two by the upper feature graph through mean pooling operation, obtaining the feature graph with the width of five by the lower feature graph through the same way, combining the two feature graphs obtained by processing into the feature graph with the width of seven, and processing the features of the upper layer and the lower layer into single-layer features.
Dividing the processed single-layer features with the width of seven into seven parts, reconstructing each part of the features to obtain a two-dimensional tensor, reducing the output dimensionality to 128 through a layer of full-connection layer (adding a ReLU activation function), finally obtaining an output result through a layer of full-connection layer, and finally combining the seven output results to obtain a final seven-bit double-layer license plate recognition result. As shown in figure three.
It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
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.
Although the invention has been described above with reference to various embodiments, it should be understood that many changes and modifications may be made without departing from the scope of the invention. It is therefore intended that the foregoing detailed description be regarded as illustrative rather than limiting, and that it be understood that it is the following claims, including all equivalents, that are intended to define the spirit and scope of this invention. The above examples are to be construed as merely illustrative and not limitative of the remainder of the disclosure. After reading the description of the invention, the skilled person can make various changes or modifications to the invention, and these equivalent changes and modifications also fall into the scope of the invention defined by the claims.
Claims (8)
1. A double-layer license plate recognition method based on deep learning is characterized by comprising the following steps:
step 1: identifying the double-layer license plate, and inputting the identified double-layer license plate into the constructed convolutional neural network;
step 2: performing feature extraction on the double-layer license plate through a backbone network of the convolutional neural network and performing normalization processing on input data of a partial network, wherein the feature extraction backbone network comprises eight convolutional layers and four pooling layers;
and step 3: dividing the feature map obtained by the network calculation into an upper layer feature map and a lower layer feature map according to the height of the feature map, obtaining the feature map with the width of a first preset width by the upper layer feature map through mean pooling, obtaining the feature map with the width of a second preset width by the lower layer, merging the two processed feature maps into the feature map with the width of a third preset width, and processing the features of the upper layer and the lower layer into single-layer features;
and 4, step 4: the processed single-layer features with the third preset width are equally divided into n parts, each part of the evenly divided features is reconstructed to obtain a two-dimensional tensor, the output dimension is reduced to 128 through a full connection layer, then an output result is obtained through the full connection layer, and finally the n parts of the output results are combined to obtain a final double-layer license plate recognition result.
2. The deep learning-based double-layer license plate recognition method of claim 1, wherein the convolutional neural network implementation is accomplished using a deep learning framework PyTorch.
3. The deep learning-based double-layer license plate recognition method of claim 1,
the target loss function of the convolutional neural network adopts AM-softmax, and the formula is as follows:
and cos (theta) is the area of the calculation sample on the category, cos (theta) -m is the interval of at least a super parameter m in the area between the categories, and s (cos (theta) -m) introduces the super parameter s and expands the interval of the cos value by s times.
4. The deep learning-based double-layer license plate recognition method of claim 1, wherein the step 2 further comprises: the first layer convolution to the sixth layer convolution of the feature extraction backbone network are respectively composed of 64, 128, 256, 512 and 512 convolution kernels of 3x3, the step length stride is 1, the padding is 1, the seventh layer convolution is composed of 512 convolution kernels of 3x3, the step length stride is 1, the eighth layer convolution is composed of 512 convolution kernels of 2x2, the step length stride is 1, each layer convolution adopts the following formula to normalize input data:
in the training process, a batch random gradient descending mode is adopted, wherein E [ x (k) ] refers to the average value of each batch of training data neurons x (k); the denominator is a standard deviation of the activation of each data neuron x (k).
5. The deep learning-based double-layer license plate recognition method of claim 1, wherein the step 2 further comprises: the activation function used after each layer of convolution is the ReLU function:
a=g(x)=max(0,z)
the four pooling layers are maximum pooling layers Max padding, the size of sliding windows of the first pooling layer and the second pooling layer is 2x2, the step length stride is 2, the step length variance of an element in each window is 1, the size of sliding windows of the third pooling layer and the fourth pooling layer is 2x2, the step length stride of rightward sliding is 2, the step length stride of downward sliding is 1, and the step length variance of the element in each window is 1; the four largest pooling layers are behind the first, second, fourth, and sixth convolution layers, respectively.
6. The deep learning-based double-layer license plate recognition method of claim 1, wherein the step 3 further comprises: the third preset width is the sum of the first preset width and the second preset width.
7. The deep learning-based double-layer license plate recognition method of claim 1, wherein the step 3 further comprises: the first preset width takes the value of 2, and the second preset width is 5.
8. The deep learning-based double-layer license plate recognition method of claim 1, wherein the double-layer license plate is of a double-layer structure, wherein the upper layer is a combination of one letter of a Chinese character and the lower layer is a combination of five letters and numbers.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110406424.8A CN113191220A (en) | 2021-04-15 | 2021-04-15 | Deep learning-based double-layer license plate recognition method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110406424.8A CN113191220A (en) | 2021-04-15 | 2021-04-15 | Deep learning-based double-layer license plate recognition method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113191220A true CN113191220A (en) | 2021-07-30 |
Family
ID=76977357
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110406424.8A Pending CN113191220A (en) | 2021-04-15 | 2021-04-15 | Deep learning-based double-layer license plate recognition method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113191220A (en) |
Citations (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109063666A (en) * | 2018-08-14 | 2018-12-21 | 电子科技大学 | The lightweight face identification method and system of convolution are separated based on depth |
CN109508715A (en) * | 2018-10-30 | 2019-03-22 | 南昌大学 | A kind of License Plate and recognition methods based on deep learning |
CN109766805A (en) * | 2018-12-28 | 2019-05-17 | 安徽清新互联信息科技有限公司 | A kind of double-deck license plate character recognition method based on deep learning |
CN109948573A (en) * | 2019-03-27 | 2019-06-28 | 厦门大学 | A kind of noise robustness face identification method based on cascade deep convolutional neural networks |
CN110163199A (en) * | 2018-09-30 | 2019-08-23 | 腾讯科技(深圳)有限公司 | Licence plate recognition method, license plate recognition device, car license recognition equipment and medium |
CN110889323A (en) * | 2019-10-10 | 2020-03-17 | 平安科技(深圳)有限公司 | Universal license plate recognition method and device, computer equipment and storage medium |
CN110956957A (en) * | 2019-12-23 | 2020-04-03 | 苏州思必驰信息科技有限公司 | Training method and system of speech enhancement model |
CN111652230A (en) * | 2020-05-25 | 2020-09-11 | 浙江大华技术股份有限公司 | License plate recognition method, electronic device and storage medium |
CN111724794A (en) * | 2020-06-17 | 2020-09-29 | 哈尔滨理工大学 | Speaker recognition method |
CN111832568A (en) * | 2020-06-12 | 2020-10-27 | 北京百度网讯科技有限公司 | License plate recognition method, and training method and device of license plate recognition model |
CN111914838A (en) * | 2020-07-28 | 2020-11-10 | 同济大学 | License plate recognition method based on text line recognition |
CN112101076A (en) * | 2019-06-18 | 2020-12-18 | 虹软科技股份有限公司 | Method and device for identifying pigs |
CN112232184A (en) * | 2020-10-14 | 2021-01-15 | 南京邮电大学 | Multi-angle face recognition method based on deep learning and space conversion network |
-
2021
- 2021-04-15 CN CN202110406424.8A patent/CN113191220A/en active Pending
Patent Citations (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109063666A (en) * | 2018-08-14 | 2018-12-21 | 电子科技大学 | The lightweight face identification method and system of convolution are separated based on depth |
CN110163199A (en) * | 2018-09-30 | 2019-08-23 | 腾讯科技(深圳)有限公司 | Licence plate recognition method, license plate recognition device, car license recognition equipment and medium |
CN109508715A (en) * | 2018-10-30 | 2019-03-22 | 南昌大学 | A kind of License Plate and recognition methods based on deep learning |
CN109766805A (en) * | 2018-12-28 | 2019-05-17 | 安徽清新互联信息科技有限公司 | A kind of double-deck license plate character recognition method based on deep learning |
CN109948573A (en) * | 2019-03-27 | 2019-06-28 | 厦门大学 | A kind of noise robustness face identification method based on cascade deep convolutional neural networks |
CN112101076A (en) * | 2019-06-18 | 2020-12-18 | 虹软科技股份有限公司 | Method and device for identifying pigs |
CN110889323A (en) * | 2019-10-10 | 2020-03-17 | 平安科技(深圳)有限公司 | Universal license plate recognition method and device, computer equipment and storage medium |
CN110956957A (en) * | 2019-12-23 | 2020-04-03 | 苏州思必驰信息科技有限公司 | Training method and system of speech enhancement model |
CN111652230A (en) * | 2020-05-25 | 2020-09-11 | 浙江大华技术股份有限公司 | License plate recognition method, electronic device and storage medium |
CN111832568A (en) * | 2020-06-12 | 2020-10-27 | 北京百度网讯科技有限公司 | License plate recognition method, and training method and device of license plate recognition model |
CN111724794A (en) * | 2020-06-17 | 2020-09-29 | 哈尔滨理工大学 | Speaker recognition method |
CN111914838A (en) * | 2020-07-28 | 2020-11-10 | 同济大学 | License plate recognition method based on text line recognition |
CN112232184A (en) * | 2020-10-14 | 2021-01-15 | 南京邮电大学 | Multi-angle face recognition method based on deep learning and space conversion network |
Non-Patent Citations (2)
Title |
---|
QIUYU ZHU ET AL.: "Improving Classification Performance of Softmax", 《APPLIED SCIENCES》 * |
朱文和: "基于卷积神经网络的双行车牌识别系统", 《前沿探讨》 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP6557783B2 (en) | Cascade neural network with scale-dependent pooling for object detection | |
Jiang et al. | A pedestrian detection method based on genetic algorithm for optimize XGBoost training parameters | |
Selmi et al. | Deep learning system for automatic license plate detection and recognition | |
CN111401148A (en) | Road multi-target detection method based on improved multilevel YO L Ov3 | |
Prakash et al. | Face recognition with convolutional neural network and transfer learning | |
CN110321967B (en) | Image classification improvement method based on convolutional neural network | |
Du et al. | Fused deep neural networks for efficient pedestrian detection | |
CN101859382A (en) | License plate detection and identification method based on maximum stable extremal region | |
Sheng et al. | Learning-based road crack detection using gradient boost decision tree | |
Shen et al. | In teacher we trust: Learning compressed models for pedestrian detection | |
Kumar et al. | Face spoofing detection using improved SegNet architecture with a blur estimation technique | |
CN112580480B (en) | Hyperspectral remote sensing image classification method and device | |
CN106845458B (en) | Rapid traffic sign detection method based on nuclear overrun learning machine | |
Ayachi et al. | To perform road signs recognition for autonomous vehicles using cascaded deep learning pipeline | |
CN112084890A (en) | Multi-scale traffic signal sign identification method based on GMM and CQFL | |
Ding et al. | Noise-resistant network: a deep-learning method for face recognition under noise | |
Dorbe et al. | FCN and LSTM based computer vision system for recognition of vehicle type, license plate number, and registration country | |
Thubsaeng et al. | Vehicle logo detection using convolutional neural network and pyramid of histogram of oriented gradients | |
Uthaib et al. | Multiclassification of license plate based on deep convolution neural networks | |
CN111950583A (en) | Multi-scale traffic signal sign identification method based on GMM clustering | |
Pham et al. | CNN-based character recognition for license plate recognition system | |
JP2000259766A (en) | Pattern recognizing method | |
CN112070116B (en) | Automatic artistic drawing classification system and method based on support vector machine | |
CN111461002B (en) | Sample processing method for thermal imaging pedestrian detection | |
Sharma et al. | Deep convolutional neural network with ResNet-50 learning algorithm for copy-move forgery detection |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210730 |