Unlock instant, AI-driven research and patent intelligence for your innovation.

Traffic target detection method, device and equipment and readable storage medium

A target detection and traffic technology, applied in the field of image processing, can solve the problems of low target detection accuracy and difference in detection performance.

Pending Publication Date: 2021-06-04
济南博观智能科技有限公司
View PDF6 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] However, due to the differences in the number of various targets in the images used for model training, the trained traffic target detection model has different detection performances for various targets, and the accuracy of target detection is low.

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 more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Traffic target detection method, device and equipment and readable storage medium
  • Traffic target detection method, device and equipment and readable storage medium
  • Traffic target detection method, device and equipment and readable storage medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0052] see figure 1 , figure 1 It is an implementation flowchart of the traffic target detection method in the embodiment of the present invention, and the method may include the following steps:

[0053] S101: Analyzing the received traffic target detection request to obtain traffic scene images and traffic targets to be detected.

[0054] When a specific traffic target in a certain traffic scene needs to be detected, a traffic target detection request is sent to the traffic target detection center, and the traffic target detection request includes the traffic scene image and the traffic target to be detected. The traffic object detection center receives the traffic object detection request, and analyzes the received traffic object detection request to obtain the traffic scene image and the traffic object to be detected.

[0055] Traffic objects can include motor vehicles, tricycles, motorcycles, bicycles, pedestrians, and so on.

[0056] The image of the traffic scene to ...

Embodiment 2

[0064] see figure 2 , figure 2 It is another implementation flowchart of the traffic target detection method in the embodiment of the present invention, the method may include:

[0065] S201: Obtain sample balance degrees corresponding to each original batch of samples.

[0066] Acquire multiple traffic scene images in advance, divide each traffic scene image into batches, and perform quantitative statistics on various traffic objects contained in each batch of traffic scene images. Assume that a certain batch of samples contains a total of N sample categories, The number of samples contained in each sample category is class1_num, class2_num, class3_num, ..., classN_num. After counting the number of sample categories contained in each original batch of samples, the sample balance degree corresponding to each original batch of samples can be obtained, and the balance degree can be calculated by the following formula:

[0067] Balance degree = 1 / number of sample categories....

Embodiment 3

[0111] see image 3 , image 3 It is another implementation flowchart of the traffic object detection method in the embodiment of the present invention, and the method may include the following steps:

[0112] S301: Obtain sample balance degrees corresponding to each original batch of samples.

[0113] S302: Determine weak sample categories corresponding to each original batch of samples according to the balance degree of each sample.

[0114] S303: Obtain the number of weak samples corresponding to each weak sample category in each original batch of samples.

[0115] After the weak sample categories corresponding to each original batch of samples are determined, the number of weak samples corresponding to each weak sample category in each original batch of samples is obtained.

[0116] S304: According to the quantity of each weak sample, perform sample strength sorting for each weak sample category in each original batch of samples, and obtain a sorting result.

[0117] A...

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

No PUM Login to View More

Abstract

The invention discloses a traffic target detection method, and the method comprises the following steps: analyzing a received traffic target detection request, and obtaining a to-be-detected traffic scene image and a traffic target; calling a traffic target detection model obtained by training processed batch samples obtained by using the sample equalization processing; and detecting a traffic target in the traffic scene image by using the traffic target detection model. By applying the traffic target detection method provided by the invention, the performance difference of the traffic target detection model obtained by training on various sample detection is reduced, and the accuracy of target detection is improved. The invention further discloses a traffic target detection device and equipment and a storage medium, which have corresponding technical effects.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to a traffic target detection method, device, equipment and computer-readable storage medium. Background technique [0002] With the rapid development of deep learning object detection, the detection scenarios faced are becoming more and more complex. For example, when detecting targets such as motor vehicles, tricycles, motorcycles, bicycles, and pedestrians in common traffic scenes, the images of the collected traffic scenes are usually used in advance, and each image is used to train the traffic target detection model. . [0003] However, due to the differences in the number of various targets in the images used for model training, the trained traffic target detection model has different detection performance for various targets, and the accuracy of target detection is low. [0004] How to effectively solve the problems of the existing traffic object detection ...

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
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/00G06K9/62G08G1/01
CPCG08G1/0104G06V20/54G06F18/24G06F18/214
Inventor 路通谢会斌李聪廷
Owner 济南博观智能科技有限公司