A method and device for traffic sign recognition that improves YOLOv3

Abstract

The present invention provides a method and device for improving traffic sign recognition of YOLOv3. The method includes collecting a number of traffic sign pictures to obtain a data set; preprocessing the data set based on a preset data augmentation method; dividing the preprocessed data set into a training set and a test set; introducing channel attention to enhance the depth network in the YOLOv3 network model to obtain an improved YOLOv3 network model; inputting the training set into the improved YOLOv3 network model for training and using the test set for verification to obtain a trained YOLOv3 network model; and using the trained YOLOv3 network model to recognize traffic signs. By performing data augmentation processing on the pictures, overfitting can be effectively prevented while the generalization ability of the model is increased. By introducing channel attention to enhance the depth network in the network model, the accuracy of traffic sign recognition can be improved and the efficiency of traffic sign recognition can be enhanced.

Description

Technical Field

[0001] The present invention belongs to the technical field of traffic sign recognition, and in particular relates to a method and device for improving YOLOv3 traffic sign recognition. Background Art

[0002] The increasing popularity of self-driving cars necessitates more accurate traffic sign recognition technology. Traditional traffic sign recognition methods mostly employ convolutional neural network models. These models are renowned for their exceptionally high accuracy, but also for their high computational cost and memory requirements. While they can address performance issues in areas such as image classification and object detection, they are still subject to significant challenges in terms of hardware requirements. Summary of the Invention

[0003] The present invention aims to solve at least one of the technical problems existing in the prior art and provide a method and device for improving traffic sign recognition of YOLOv3.

[0004] In one aspect of the present invention, a method for improving traffic sign recognition using YOLOv3 is provided, the method comprising:

[0005] Collect several traffic sign images to obtain a data set;

[0006] Preprocessing the data set based on a preset data enhancement method;

[0007] Divide the preprocessed dataset into training set and test set;

[0008] The channel attention enhanced deep network is introduced into the YOLOv3 network model to obtain the improved YOLOv3 network model;

[0009] Inputting the training set into the improved YOLOv3 network model for training, and using the test set for verification to obtain a trained YOLOv3 network model;

[0010] The trained YOLOv3 network model is used to identify traffic signs.

[0011] In some embodiments, the preprocessing of the dataset based on a preset data enhancement method includes:

[0012] The dataset is preprocessed based on a mosaic data enhancement method.

[0013] In some embodiments, the mosaic-based data enhancement method preprocesses the dataset, including:

[0014] Randomly select multiple traffic sign images from the dataset;

[0015] Perform flipping and scaling operations on the selected multiple traffic sign images respectively;

[0016] Randomly select a center point on the new image, and perform size cropping on the selected multiple traffic sign images along the vertical and horizontal lines where the center point is located;

[0017] Arrange the cropped images in a preset order to obtain a preprocessed dataset.

[0018] In some embodiments, the arranging the cropped images in a preset order includes:

[0019] Arrange the cropped images in the order of upper left, lower left, upper right, and lower right.

[0020] On the other hand, the present invention provides a device for improving traffic sign recognition of YOLOv3, and the device includes:

[0021] An acquisition module, configured to acquire a plurality of traffic sign images to obtain a dataset;

[0022] A preprocessing module, configured to preprocess the dataset based on a preset data augmentation method;

[0023] A partitioning module, configured to partition the preprocessed dataset into a training set and a test set;

[0024] A design module, configured to introduce channel attention to enhance the depth network in the YOLOv3 network model to obtain an improved YOLOv3 network model;

[0025] A training module, configured to input the training set into the improved YOLOv3 network model for training, and use the test set for verification to obtain a trained YOLOv3 network model;

[0026] An identification module, configured to identify traffic signs by using the trained YOLOv3 network model.

[0027] In some embodiments, the preprocessing module is specifically further configured to:

[0028] Preprocess the dataset based on the mosaic data augmentation method.

[0029] In some embodiments, the preprocessing module is specifically further configured to:

[0030] Randomly select multiple traffic sign images from the dataset;

[0031] Perform flipping and scaling operations on the selected multiple traffic sign pictures respectively;

[0032] Randomly select a center point on the new picture, and perform size cropping on the selected multiple traffic sign pictures along the vertical and horizontal lines where the center point is located;

[0033] Arrange the cropped pictures in a preset order to obtain a preprocessed data set.

[0034] In some embodiments, the preprocessing module is further specifically configured to:

[0035] Arrange the cropped pictures in the order of upper left, lower left, upper right, and lower right.

[0036] On the other hand, the present invention provides an electronic device, including:

[0037] One or more processors;

[0038] A storage unit for storing one or more programs, which when executed by the one or more processors, can enable the one or more processors to implement the method described above.

[0039] On the other hand, the present invention provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it can implement the method described above.

[0040] The method and device for traffic sign recognition that improves YOLOv3 of the present invention, through data augmentation processing on the collected traffic sign pictures, can effectively prevent overfitting while increasing the generalization ability of the model. And by introducing channel attention in the YOLOv3 network model to enhance the depth network, it is possible to focus on what kind of features are meaningful, summarize spatial features, thereby improving the accuracy and efficiency of traffic sign recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a flowchart of the method for traffic sign recognition that improves YOLOv3 proposed by the present invention;

[0042] Figure 2 It is a schematic structural diagram of the device for traffic sign recognition that improves YOLOv3 proposed by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0043] To enable those skilled in the art to better understand the technical solutions of the present invention, the present invention will be further described in detail below with reference to the drawings and specific embodiments.

[0044] On the one hand, in this embodiment, asFigure 1 As shown, it relates to a method S100 for traffic sign recognition that improves YOLOv3. The method S100 includes:

[0045] S110. Collect a number of traffic sign pictures to obtain a dataset.

[0046] S120. Preprocess the dataset based on a preset data augmentation method.

[0047] Specifically, in this step, preprocess the dataset based on the mosaic data augmentation method. In order to effectively prevent overfitting while increasing the generalization ability of the model, the data augmentation method in this embodiment specifically includes the following steps:

[0048] Randomly select multiple traffic sign pictures from the dataset.

[0049] Perform flipping and scaling operations on the selected multiple traffic sign pictures respectively.

[0050] Randomly select a center point on the new picture, and perform size cropping on the selected multiple traffic sign pictures along the vertical and horizontal lines where the center point is located.

[0051] Arrange the cropped pictures in a preset order to obtain a preprocessed dataset.

[0052] Furthermore, the arranging the cropped pictures in a preset order includes: arranging the cropped pictures in the order of upper left, lower left, upper right, and lower right.

[0053] S130. Divide the preprocessed dataset into a training set and a test set.

[0054] S140. Introduce channel attention to enhance the depth network in the YOLOv3 network model to obtain an improved YOLOv3 network model.

[0055] S150. Input the training set into the improved YOLOv3 network model for training, and use the test set for verification to obtain a trained YOLOv3 network model.

[0056] S160. Use the trained YOLOv3 network model to recognize traffic signs.

[0057] The method for traffic sign recognition by improving YOLOv3 in this embodiment can effectively prevent overfitting while increasing the generalization ability of the model by performing data augmentation on the collected traffic sign pictures. Moreover, by introducing channel attention to enhance the deep network in the YOLOv3 network model, it is possible to focus on what features are meaningful and summarize spatial features, thereby improving the accuracy and efficiency of traffic sign recognition.

[0058] On the other hand, as Figure 2 shown, a device 100 for traffic sign recognition by improving YOLOv3 is provided. The device 100 can be applied to the method described above. The device 100 includes:

[0059] An acquisition module 110, configured to acquire a number of traffic sign pictures to obtain a data set;

[0060] A preprocessing module 120, configured to preprocess the data set based on a preset data augmentation method;

[0061] A partitioning module 130, configured to partition the preprocessed data set into a training set and a test set;

[0062] A design module 140, configured to introduce channel attention to enhance the deep network in the YOLOv3 network model to obtain an improved YOLOv3 network model;

[0063] A training module 150, configured to input the training set into the improved YOLOv3 network model for training and use the test set for verification to obtain a trained YOLOv3 network model;

[0064] An identification module 160, configured to identify traffic signs using the trained YOLOv3 network model.

[0065] The device for traffic sign recognition by improving YOLOv3 in this embodiment can effectively prevent overfitting while increasing the generalization ability of the model by performing data augmentation on the collected traffic sign pictures. Moreover, by introducing channel attention to enhance the deep network in the YOLOv3 network model, it is possible to focus on what features are meaningful and summarize spatial features, thereby improving the accuracy and efficiency of traffic sign recognition.

[0066] In some embodiments, the preprocessing module 120 is further specifically configured to:

[0067] Preprocess the data set based on the mosaic data augmentation method.

[0068] In some embodiments, the pre-processing module 120 is further configured to:

[0069] Randomly selecting multiple traffic sign images from the dataset;

[0070] Performing flipping and scaling operations on the selected multiple traffic sign images respectively;

[0071] A center point is randomly selected on the new image, and the selected multiple traffic sign images are cropped to the size of the vertical and horizontal lines of the center point;

[0072] The cropped images are placed in a preset order to obtain a preprocessed data set.

[0073] In some embodiments, the pre-processing module 120 is further configured to:

[0074] Arrange the cropped images in the order of upper left, lower left, upper right, and lower right.

[0075] Another aspect of the present invention provides an electronic device, comprising:

[0076] one or more processors;

[0077] The storage unit is used to store one or more programs, which, when executed by the one or more processors, enable the one or more processors to implement the method described above.

[0078] Another aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program can implement the method described above when executed by a processor.

[0079] The computer-readable medium may be included in the apparatus, device, or system of the present invention, or may exist independently.

[0080] Among them, computer-readable storage media can be any tangible medium that contains or stores a program, which can be an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment. More specific examples include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, an optical fiber, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0081] Among them, the computer-readable storage medium may also include a data signal propagated in a baseband or as part of a carrier wave, which carries computer-readable program code, and specific examples thereof include, but are not limited to, electromagnetic signals, optical signals, or any suitable combination thereof.

[0082] It can be understood that the above embodiments are merely exemplary embodiments adopted to illustrate the principles of the present invention. However, the present invention is not limited thereto. For those of ordinary skill in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also regarded as the protection scope of the present invention.

Claims

1. A method for traffic sign recognition that improves YOLOv3, characterized in that, The method includes: Collecting a number of traffic sign pictures to obtain a data set; Preprocessing the data set based on a preset data augmentation method; Dividing the preprocessed data set into a training set and a test set; Introducing a channel attention enhanced deep network into the YOLOv3 network model to obtain an improved YOLOv3 network model; Inputting the training set into the improved YOLOv3 network model for training and using the test set for verification to obtain a trained YOLOv3 network model; Identifying traffic signs using the trained YOLOv3 network model; The preprocessing the data set based on a preset data augmentation method includes: preprocessing the data set based on the mosaic data augmentation method, specifically: randomly selecting multiple traffic sign pictures from the data set; respectively performing flipping and scaling operations on the selected multiple traffic sign pictures; randomly selecting a center point on the new picture and performing size cropping on the selected multiple traffic sign pictures along the vertical and horizontal lines where the center point is located; arranging the cropped pictures in a preset order to obtain a preprocessed data set.

2. The method according to claim 1, wherein The arranging the cropped pictures in a preset order includes: Arranging the cropped pictures in the order of top left, bottom left, top right, and bottom right.

3. An apparatus for traffic sign recognition that improves YOLOv3, characterized in that, The device includes: A collection module for collecting a number of traffic sign pictures to obtain a data set; A preprocessing module for preprocessing the data set based on a preset data augmentation method; A division module for dividing the preprocessed data set into a training set and a test set; A design module for introducing a channel attention enhanced deep network into the YOLOv3 network model to obtain an improved YOLOv3 network model; A training module for inputting the training set into the improved YOLOv3 network model for training and using the test set for verification to obtain a trained YOLOv3 network model; An identification module for identifying traffic signs using the trained YOLOv3 network model; The preprocessing module is specifically further configured to: preprocess the data set based on the mosaic data augmentation method, specifically: randomly selecting multiple traffic sign pictures from the data set; respectively performing flipping and scaling operations on the selected multiple traffic sign pictures; randomly selecting a center point on the new picture and performing size cropping on the selected multiple traffic sign pictures along the vertical and horizontal lines where the center point is located; arranging the cropped pictures in a preset order to obtain a preprocessed data set.

4. The device according to claim 3, characterized in that, The preprocessing module is specifically further configured to: Arrange the cropped pictures in the order of top left, bottom left, top right, and bottom right.

5. An electronic device, characterized in that, It includes: One or more processors; A storage unit for storing one or more programs, which when executed by the one or more processors, can enable the one or more processors to implement the method according to claim 1 or 2.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that, The computer program, when executed by a processor, can implement the method according to claim 1 or 2.

Images