Improved traffic image target detection method
A target detection and traffic technology, applied in the field of image recognition, can solve the problems of low accuracy and good real-time model, and achieve the effect of improving accuracy and high computing efficiency
Pending Publication Date: 2021-12-21
BEIJING UNIV OF TECH
View PDF0 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0005] The technical problem mainly solved by the present invention is the problem that the simple model has good real-time performance but low accuracy. Without affecting the real-time performance of the model, the detection accuracy of the traffic image target detection model is improved, and a method suitable for traffic image recognition is proposed. The traffic image target detection model that can focus on long-range interactive information, Lambda Yolo V3 (L-Yolov3), is based on the improvement of Yolo v3 proposed by Ross Girshick et al. We introduce the long-range interactive information capture layer LambdaLayer into the backbone network Darknet-53 In , a new traffic image target detection model is formed
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0021] The present invention will be described in further detail below in conjunction with specific network model diagrams and with reference to the accompanying drawings.
[0022] The hardware equipment used in the present invention has a PC and a graphics card rtx3070.
[0023] In this part, experiments are carried out on the proposed ideas to verify the feasibility of the proposed invention, which specifically includes the following steps:
[0024] Step 1: build the Lambda-Yolo v3 target detection model proposed by the present invention, as the core part of the present invention, it mainly includes the following steps:
[0025] Step 1.1: Replace the Conv2d layer in Convolution (3x3) in the residual module section in Darknet-53 with the Lambda Conv layer to obtain the L-CCR module.
[0026] Step 1.2: Improve the 8 CCR residual modules in the third stage in Darknet-53, and replace the 7th and 8th CCR modules with the newly generated L-CCR modules in step 1.1.
[0027] Step ...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention discloses an improved traffic image target detection method, which aims at massive traffic image target detection data, and adopts different replacement strategies to improve residual modules of three stages according to different extraction characteristics of three scale feature maps in a Darknet-53 model. A residual structure in Darknet-53 is improved to obtain an L-CCR module, the original 3 * 3 Spatial Conversion is replaced by LambdaLayer, the convolutional network is endowed with the capability of capturing long-distance interaction information, the accuracy of traffic image target detection is improved, meanwhile, the introduction of the LambdaLayer layer is calculation at a linear level, so that the speed is not greatly influenced. The real-time performance of the original model can be ensured; and the constructed traffic image target detection model has the advantage of high calculation efficiency.
Description
technical field [0001] The invention belongs to the field of image recognition, and in particular relates to a method for detecting objects in traffic images. Background technique [0002] With the rapid development of the information age, the application of artificial intelligence in life is becoming more and more common. As a popular direction of computer vision and digital image processing, target detection is widely used in intelligent transportation systems. Target detection is the basic task of the system. Play a decisive role in the overall performance. However, the amount of data that needs to be processed for video-based target detection is quite large. Simple algorithms have good real-time performance but low accuracy, while complex algorithms have high accuracy and poor real-time performance. Therefore, how to balance the accuracy and accuracy of target detection technology Real-time has become a research hotspot in today's academic circles. [0003] The researc...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06N3/045G06F18/214G06F18/241Y02T10/40
Inventor 李永余凤
Owner BEIJING UNIV OF TECH