CN116958952B - License plate target detection method suitable for expressway monitoring video - Google Patents
License plate target detection method suitable for expressway monitoring video Download PDFInfo
- Publication number
- CN116958952B CN116958952B CN202310845871.2A CN202310845871A CN116958952B CN 116958952 B CN116958952 B CN 116958952B CN 202310845871 A CN202310845871 A CN 202310845871A CN 116958952 B CN116958952 B CN 116958952B
- Authority
- CN
- China
- Prior art keywords
- license plate
- feature
- plate target
- target detection
- expressway
- 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
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 65
- 238000012544 monitoring process Methods 0.000 title claims abstract description 17
- 238000012549 training Methods 0.000 claims abstract description 14
- 230000007246 mechanism Effects 0.000 claims abstract description 7
- 238000013461 design Methods 0.000 claims abstract description 4
- 239000013598 vector Substances 0.000 claims description 5
- 230000006870 function Effects 0.000 claims description 4
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 3
- 238000010801 machine learning Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 238000000034 method Methods 0.000 abstract description 4
- 230000000694 effects Effects 0.000 description 7
- 230000004927 fusion Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000001502 supplementing effect Effects 0.000 description 1
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
- G06V20/625—License plates
-
- 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
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- General Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a license plate target detection method suitable for expressway monitoring video, which comprises the following steps: improving the detection branch of the SSD by adopting an attention mechanism, and performing element-by-element point multiplication on a feature mask obtained by an attention module of a high-level feature map and a shallow-level feature map; fusing the feature graphs of each layer in a superposition mode, and then combining license plate characteristics to design a attention feature pyramid applicable to expressway scenes; training a license plate target detection model on the self-built data set; and detecting the license plate target in the expressway scene by using the trained model, and acquiring the position and the size of the license plate target. The method is suitable for detecting the license plate target under the expressway monitoring video, has good multi-scale target detection performance and high positioning accuracy, and can be directly applied to detecting the license plate target under the expressway monitoring video.
Description
Technical Field
The invention belongs to the technical field of image processing, and particularly relates to a license plate target detection method suitable for expressway monitoring video.
Background
License plate target detection aims at finding license plate targets in images from monitoring video images and determining their positions and sizes. The automatic license plate recognition can be realized only by combining the license plate target detection task and the license plate recognition task, and the license plate target detection effect directly influences the license plate recognition effect. The existing algorithm is mainly suitable for scenes with good image acquisition conditions, such as parking lots, and has poor adaptability to complex scenes, such as highways.
The license plate targets in the expressway scene have the characteristics of small scale and multiple scale, and the characteristic extraction and fusion strategy of SSD aims at targets with smaller scale and larger span such as license plates, so that the detection effect is poor, a large number of license plate targets exist in practical application, the recall rate is reduced, and the problem is more serious particularly in the expressway scene. Meanwhile, the real-time requirement on the detection of the license plate target of the expressway is high, the existing license plate target detection algorithm mostly adopts a one-stage framework, but the multi-scale target detection performance and the positioning accuracy of the one-stage detection algorithm are low.
Therefore, a license plate target detection method suitable for expressway monitoring videos is needed, the detection speed of the license plate target is ensured, and meanwhile, the multi-scale target detection performance and the positioning accuracy of the license plate target detection are improved, so that the license plate target detection method has important practical significance for expressway vehicle management and control.
Disclosure of Invention
Therefore, the present invention is directed to a license plate target detection method suitable for use in expressway surveillance video. The invention aims to solve the problems of poor applicability of the existing license plate target detection algorithm and low multi-scale target detection performance and positioning accuracy for complex scenes such as highways. The invention improves on the basis of the target detection network SSD, aims at the problem of poor effect of SSD in detecting the multi-scale license plate of the expressway, and combines the attention mechanism and the feature fusion to construct an attention feature pyramid, so that the detection precision of the multi-scale license plate target is improved.
In order to achieve the above purpose, the invention provides a license plate target detection method suitable for expressway monitoring video, comprising the following steps:
S1, improving a detection branch of an SSD by adopting an attention mechanism, and performing element-by-element point multiplication on a feature mask obtained by an attention module of a high-level feature map and a shallow feature map;
S2, fusing the feature graphs of all layers in a superposition mode, and then combining license plate characteristics to design a attention feature pyramid applicable to expressway scenes;
S3, training a license plate target detection model on a self-built expressway license plate target detection data set;
s4, detecting the license plate target in the expressway scene by using the trained model, and acquiring the position and the size of the license plate target.
Further, the step S1 includes the following substeps:
S1.1, calculating an attention mask corresponding to a deeper layer of features by using a Softmax function, wherein the attention mask comprises the Softmax probability of a channel and the Softmax probability of a feature position;
s1.2, the features of each feature layer are respectively fused with the attention mask corresponding to the deep layer features in an element-by-element dot multiplication mode.
Further, in the step S1.1, the Softmax probability of the channel is obtained by using the pixel value corresponding to any channel position in the feature map, and the Softmax probability C-Softmax of the channel is represented by the following formula:
in the formula, the size of the feature map is K multiplied by N, namely the width and the height of the feature map are K, and N channels are all arranged; representing the pixel value at c-channel position (i, j) in the feature map; lambda ij represents the pixel a priori eigenvalues, set to 1 by default.
Further, in the step S1.1, the Softmax probability of the feature position is obtained by using the pixel value of the corresponding point in the feature map, and the Softmax probability F-Softmax of the feature position is represented by the following formula:
Wherein, the size of the feature map is K multiplied by N; Representing the pixel value at c-channel position (i, j) in the feature map; lambda ij represents the a priori eigenvalues of the pixel, default to 1.
Further, the step S2 includes the following sub-steps:
s2.1, summing the features to be fused according to a superposition mode to obtain a new feature vector;
S2.2, forming an attention feature pyramid by a plurality of feature vectors, designing the layer number and the resolution of each layer of the feature pyramid by combining the characteristics of the license plate scale and the span, and then training and predicting the machine learning model of the attention feature pyramid.
Further, the specific steps of the step S3 are as follows:
and training on the built license plate target detection model on a training set by using an ADAM optimization algorithm to obtain and store the weight of the license plate target detection model.
Further, the step S4 includes the following substeps:
s4.1, loading the model weight obtained in the step S3 by using a designed license plate target detection model;
and S4.2, detecting the input monitoring video image in the expressway scene by using a license plate target detection model loaded with weight to obtain the position and the size of a license plate target.
The invention has the beneficial effects that:
Unlike the prior art, the present invention only considers supplementing semantic context information to shallow features, and ignores the task of fusing the low-level feature map and the high-level feature map. The invention uses the attention feature map generated by the high-level features as the attention area mask of the low-level features, so that the representation capability of the low-level features can be effectively enhanced; then, the attention feature pyramid is constructed by combining the characteristics of license plate targets in the expressway scene, so that the feature map of each level simultaneously contains rich position information and high-level semantic context information; finally, a complete license plate target detection method suitable for expressway monitoring videos is formed, the effect of detecting the multi-scale license plate targets in expressway scenes by using a practical algorithm is achieved, the multi-scale target detection performance is good, and the positioning accuracy is high.
Additional advantages, objects, and features of the invention will be set forth in part in the description which follows and in part will become apparent to those having ordinary skill in the art upon examination of the following or may be learned from practice of the invention. The objects and other advantages of the invention may be realized and obtained by means of the instrumentalities and combinations particularly pointed out in the specification.
Drawings
FIG. 1 is a flow chart of a license plate target detection method of the present invention;
FIG. 2 is a comparison of the improved detection branch and the original detection branch of SSD;
FIG. 3 is a schematic diagram of feature fusion in a superimposed manner;
fig. 4 is a diagram of a network structure after SSD modification.
Detailed Description
In order to make the technical scheme, advantages and objects of the present application more clear, the technical scheme of the embodiment of the present application will be clearly and completely described below with reference to the accompanying drawings of the embodiment of the present application. It will be apparent that the described embodiments are some, but not all, embodiments of the application. All other embodiments, which can be obtained by a person skilled in the art without creative efforts, based on the described embodiments of the present application belong to the protection scope of the present application.
As shown in fig. 1, the invention provides a license plate target detection method suitable for a highway monitoring video, which comprises the following steps:
s1, improving a detection branch of SSD by adopting an attention mechanism, and performing element-by-element dot multiplication on a high-level feature map through a feature mask obtained by an attention module and a shallow-level feature map, so that a feature selection effect can be achieved;
S2, fusing the feature graphs of all layers in a superposition mode to keep more detail information, avoiding overfitting, emphasizing the shallow feature layers, and designing a attention feature pyramid, namely a license plate target detection model, suitable for expressway scenes by combining the characteristics of license plates;
S3, training a license plate target detection model on a self-built expressway license plate target detection data set;
S4, detecting the license plate target in the expressway scene by using the trained license plate target detection model, and acquiring the position and the size of the license plate target.
Step S1: the attention mechanism is adopted to improve the detection branch of the SSD, as shown in fig. 2, after the further features pass through the upsampling and attention module, the corresponding attention mask is obtained, and the attention mask comprises the Softmax probability of a channel and the Softmax probability of the feature position, and specifically comprises the following substeps:
S1.1, calculating an attention mask corresponding to a deeper layer of features by using a Softmax function, wherein the attention mask comprises the Softmax probability of a channel and the Softmax probability of a feature position;
The Softmax probability of the channel is obtained by adopting a pixel value corresponding to any channel position in the feature map, and the Softmax probability C-Softmax of the channel is shown as follows:
in the formula, the size of the feature map is K multiplied by N, namely the width and the height of the feature map are K, and N channels are all arranged; representing the pixel value at c-channel position (i, j) in the feature map; lambda ij represents the pixel a priori eigenvalues, set to 1 by default.
The Softmax probability of the feature position is obtained by adopting the pixel value of the corresponding point in the feature map, and the Softmax probability F-Softmax of the feature position is shown in the following formula:
Wherein, the size of the feature map is K multiplied by N; Representing the pixel value at c-channel position (i, j) in the feature map; lambda ij represents the prior eigenvalue of the pixel, default to 1;
S1.2, fusing the features of each feature layer with the attention mask corresponding to the corresponding deeper features in a way of element-by-element dot multiplication;
Assuming that F i represents features of different levels in the SSD and F i′+1 represents an attention mask corresponding to a feature of a deeper level, the multi-scale attention feature can be calculated by:
in the method, in the process of the invention, The two representations are fused in a way of element-by-element dot multiplication.
Step S2: the feature graphs of all layers are fused in a superposition mode so as to keep more detail information, avoid overfitting, emphasize the shallow feature layers, and finally design an attention feature pyramid by combining the characteristics of license plates as shown in fig. 3, and specifically comprises the following substeps:
S2.1 supposing n feature maps F 1,F2,...,Fn, each feature map has a size of h×w×c, where H represents a height, W represents a width, and C represents a channel number, the formula of feature map superposition can be expressed as:
Wherein F out denotes an output feature map, (i, j) denotes a position in the feature map, k denotes the number of channels, and w l denotes a weight of the first feature map;
S2.2, forming an attention feature pyramid by a plurality of feature vectors, designing the layer number and the resolution of each layer of the feature pyramid by combining the characteristics of license plate dimensions and spans, and then training and predicting a machine learning model for the attention feature pyramid;
the invention improves the network structure of SSD based on license plate characteristics, takes Conv4_3 and Conv7 characteristics based on attention improvement as references, and additionally adds Conv8 and Conv9 formed by 1X 1 convolution with a step length of 2 and 3X 3 convolution; the added detection branches are improved by using the attention module, 4 feature graphs with different scales are obtained, so that an attention feature pyramid is constructed, and the result is shown in fig. 4; because license plate sizes in expressway scenes are generally smaller, input resolution is modified to 512×512, and the resolutions of the layers of the attention feature pyramid are shown in table 1.
TABLE 1
Step S3: training a license plate target detection model on a self-built expressway license plate target detection data set until a loss function converges, wherein the specific steps are as follows:
and training the model by using an ADAM optimization algorithm on the built network model to obtain and store the weight of the license plate target detection model.
Step S4: detecting a license plate target in a highway scene by using the trained model, and acquiring the position and the size of the license plate target, wherein the method specifically comprises the following substeps:
s4.1, using a designed license plate target detection model, and loading the weight of the model obtained in the step S3;
and S4.2, detecting the input monitoring video image in the expressway scene by using a license plate target detection model loaded with weight to obtain the position and the size of a license plate target.
Aiming at the problems that the SSD type 'bottom-up' feature pyramid structure is not perfect enough and the multi-scale feature extraction is insufficient, the attention feature pyramid is constructed by combining the comprehensive attention mechanism and the feature fusion and combining license plate characteristics, so that each layer of features can be ensured to pay attention to important information in an image better and all the features contain rich detail information and contextual semantic information, and the detection effect of the multi-scale license plate target is improved. Then training a designed model on a self-built expressway license plate target detection data set until the model converges, and storing the trained model weight; finally, license plate target detection in the expressway scene is realized according to the improved license plate target detection model suitable for the expressway scene and the trained weight.
Finally, it is noted that the above embodiments are only for illustrating the technical solution of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention may be modified or substituted without departing from the spirit and scope of the technical solution, and the present invention is intended to be covered in the scope of the present invention.
Claims (3)
1. The license plate target detection method suitable for the expressway monitoring video is characterized by comprising the following steps of:
S1, improving a detection branch of an SSD by adopting an attention mechanism, and performing element-by-element point multiplication on a feature mask obtained by an attention module of a high-level feature map and a shallow feature map;
S1.1, calculating an attention mask corresponding to deep features by using a Softmax function, wherein the attention mask comprises Softmax probability of a channel and Softmax probability of feature positions;
The Softmax probability of the channel is obtained by adopting a pixel value corresponding to any channel position in the feature map, and the Softmax probability C-Softmax of the channel is shown as follows:
in the formula, the size of the feature map is K multiplied by N, namely the width and the height of the feature map are K, and N channels are all arranged; representing the pixel value at c-channel position (i, j) in the feature map; lambda ij represents the pixel prior eigenvalue, default to 1;
s1.2, fusing the features of each feature layer with the attention mask corresponding to the deep layer features in an element-by-element dot multiplication mode;
The Softmax probability of the feature position is obtained by adopting the pixel value of the corresponding point in the feature map, and the Softmax probability F-Softmax of the feature position is shown in the following formula:
in the formula, the size of the feature map is K multiplied by N, namely the width and the height of the feature map are K, and N channels are all arranged; Representing the pixel value at c-channel position (i, j) in the feature map; lambda ij represents the prior eigenvalue of the pixel, default to 1;
S2, fusing the feature graphs of all layers in a superposition mode, and then combining license plate characteristics to design a attention feature pyramid applicable to expressway scenes;
s2.1, summing the features to be fused according to a superposition mode to obtain a new feature vector;
S2.2, forming an attention feature pyramid by a plurality of feature vectors, designing the layer number and the resolution of each layer of the feature pyramid by combining the characteristics of license plate dimensions and spans, and then training and predicting a machine learning model for the attention feature pyramid;
S3, training a license plate target detection model on a self-built expressway license plate target detection data set;
s4, detecting the license plate target in the expressway scene by using the trained model, and acquiring the position and the size of the license plate target.
2. The license plate target detection method applicable to the expressway monitoring video according to claim 1, wherein the specific steps of the step S3 are as follows:
and training on the built license plate target detection model on a training set by using an ADAM optimization algorithm to obtain and store the weight of the license plate target detection model.
3. The license plate target detection method applicable to expressway monitoring video according to claim 2, wherein said step S4 comprises the following sub-steps:
s4.1, loading the model weight obtained in the step S3 by using a designed license plate target detection model;
and S4.2, detecting the input monitoring video image in the expressway scene by using a license plate target detection model loaded with weight to obtain the position and the size of a license plate target.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310845871.2A CN116958952B (en) | 2023-07-11 | 2023-07-11 | License plate target detection method suitable for expressway monitoring video |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310845871.2A CN116958952B (en) | 2023-07-11 | 2023-07-11 | License plate target detection method suitable for expressway monitoring video |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116958952A CN116958952A (en) | 2023-10-27 |
| CN116958952B true CN116958952B (en) | 2024-04-30 |
Family
ID=88447091
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310845871.2A Active CN116958952B (en) | 2023-07-11 | 2023-07-11 | License plate target detection method suitable for expressway monitoring video |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116958952B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110533084A (en) * | 2019-08-12 | 2019-12-03 | 长安大学 | A kind of multiscale target detection method based on from attention mechanism |
| CN112308092A (en) * | 2020-11-20 | 2021-02-02 | 福州大学 | Light-weight license plate detection and identification method based on multi-scale attention mechanism |
| CN112580536A (en) * | 2020-12-23 | 2021-03-30 | 深圳市捷顺科技实业股份有限公司 | High-order video vehicle and license plate detection method and device |
| CN113989794A (en) * | 2021-11-12 | 2022-01-28 | 珠海安联锐视科技股份有限公司 | License plate detection and recognition method |
| CN115063785A (en) * | 2022-08-17 | 2022-09-16 | 深圳联友科技有限公司 | Method and device for positioning license plate in expressway scene by using target recognition model |
-
2023
- 2023-07-11 CN CN202310845871.2A patent/CN116958952B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110533084A (en) * | 2019-08-12 | 2019-12-03 | 长安大学 | A kind of multiscale target detection method based on from attention mechanism |
| CN112308092A (en) * | 2020-11-20 | 2021-02-02 | 福州大学 | Light-weight license plate detection and identification method based on multi-scale attention mechanism |
| CN112580536A (en) * | 2020-12-23 | 2021-03-30 | 深圳市捷顺科技实业股份有限公司 | High-order video vehicle and license plate detection method and device |
| CN113989794A (en) * | 2021-11-12 | 2022-01-28 | 珠海安联锐视科技股份有限公司 | License plate detection and recognition method |
| CN115063785A (en) * | 2022-08-17 | 2022-09-16 | 深圳联友科技有限公司 | Method and device for positioning license plate in expressway scene by using target recognition model |
Non-Patent Citations (3)
| Title |
|---|
| Jingwei Ren ; Hong Li.Implementation of Vehicle and License Plate Detection on Embedded Platform.2020 12th International Conference on Measuring Technology and Mechatronics Automation (ICMTMA).2020,全文. * |
| 一种基于SSD改进的目标检测算法;苏蒙;李为;;计算机与现代化;20200215(第02期);全文 * |
| 基于特征金字塔的多尺度特征融合网络;郭启帆，刘磊，张珹，徐文娟，靖稳峰;工程数学学报;20201031;第37卷(第5期);摘要、正文第1-3节 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116958952A (en) | 2023-10-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2021135254A1 (en) | License plate number recognition method and apparatus, electronic device, and storage medium | |
| CN111695448B (en) | Roadside vehicle identification method based on visual sensor | |
| CN111008633B (en) | License plate character segmentation method based on attention mechanism | |
| CN110766723B (en) | Unmanned aerial vehicle target tracking method and system based on color histogram similarity | |
| CN113159466B (en) | Short-time photovoltaic power generation prediction system and method | |
| Xiang et al. | Lightweight fully convolutional network for license plate detection | |
| Xie et al. | A binocular vision application in IoT: Realtime trustworthy road condition detection system in passable area | |
| CN112016512A (en) | Remote sensing image small target detection method based on feedback type multi-scale training | |
| CN114913498A (en) | Parallel multi-scale feature aggregation lane line detection method based on key point estimation | |
| CN114330529A (en) | Real-time pedestrian shielding detection method based on improved YOLOv4 | |
| CN113850136A (en) | Yolov5 and BCNN-based vehicle orientation identification method and system | |
| CN111553242B (en) | Training method for generating countermeasure network for predicting driving behavior and electronic device | |
| CN116994236A (en) | Low-quality image license plate detection method based on deep neural network | |
| CN114943888B (en) | Sea surface small target detection method based on multi-scale information fusion | |
| CN111126303B (en) | Multi-parking-place detection method for intelligent parking | |
| CN116958952B (en) | License plate target detection method suitable for expressway monitoring video | |
| CN116129386A (en) | Method, system and computer readable medium for detecting a travelable region | |
| CN114639084A (en) | Road side end vehicle sensing method based on SSD (solid State disk) improved algorithm | |
| CN112199539A (en) | Automatic labeling method, system and equipment for contents of unmanned aerial vehicle three-dimensional map photographic image | |
| CN116486203B (en) | Single-target tracking method based on twin network and online template updating | |
| CN113609904B (en) | Single-target tracking algorithm based on dynamic global information modeling and twin network | |
| CN117315614B (en) | Traffic target detection method based on improved YOLOv7 | |
| WO2023029704A1 (en) | Data processing method, apparatus and system | |
| Cong et al. | Object Detection and Image Segmentation for Autonomous Vehicles | |
| CN114495160A (en) | Pedestrian detection method and system based on improved RFBNet algorithm |
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 |