Method for distinguishing electric bicycles and motorcycles through videos

A technology for electric bicycles and motorcycles, applied in character and pattern recognition, instruments, computer parts, etc., can solve the problems of not being able to distinguish electric bicycles and motorcycles well, being susceptible to environmental interference, and low recognition rate. , to achieve the effect of improving accuracy, improving accuracy, high detection rate and high recognition rate

Inactive Publication Date: 2018-05-18
CHENGDU REMARK TECH CO LTD +1
View PDF11 Cites 6 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Features are usually extracted manually, such as using external feature factors such as the size of the exhaust pipe to extract features such as hog and sift. However, these features are easily disturbed by the environment, making the recognition rate very low and cannot be distinguished well. Electric Bicycles and Motorcycles

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
  • Method for distinguishing electric bicycles and motorcycles through videos

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0028] A method for distinguishing electric bicycles and motorcycles through video, mainly used to distinguish the license plates of electric bicycles and motorcycles, mainly includes the following steps:

[0029] Step A1: collect real-time video of electric bicycles and motorcycles on the road through the camera, and generate images;

[0030] Step A2: Use the deep learning method to extract the features of the marked electric bicycle and motorcycle license plates, select the last layer as the result of feature extraction, and then connect a softmax function to determine whether it is an electric bicycle or a motorcycle, and select the probability The larger value is used as the output result to get the training model;

[0031] Step A3: Input the image generated in step A1 into the training model, and output an alarm message if the license plate of a motorcycle is detected.

[0032] The present invention is mainly used to distinguish the license plates of electric bicycles an...

Embodiment 2

[0034] This embodiment is further optimized on the basis of Embodiment 1, and the step A2 mainly includes the following steps:

[0035] Step A21: collect the video of the electric bicycle and motorcycle through the camera, and convert it into an image, and mark the type of each image according to the classification of the electric bicycle and motorcycle;

[0036]Step A22: Divide the image into 13×13 rectangular blocks, and use clustering to predict the anchor point frame for each rectangular block. The rectangular block takes 5 anchor point frames, and the size of the anchor point frame matches the size of different detection objects , so as to avoid the situation that only one object is detected when multiple objects are located in a rectangular block;

[0037] Step A23: Send each segmented rectangular block into a multi-layer convolutional neural network, use the convolutional neural network to extract image features, take out the features of the last layer, and input them i...

Embodiment 3

[0041] This embodiment is further optimized on the basis of Embodiment 1 or 2. In the step A1, the collected video is converted into an image, the image is divided into several 13×13 rectangular blocks, and each rectangular block is aggregated. Class prediction anchor point frame, described rectangular block takes 5 anchor point frames, and the size of anchor point frame matches the size of different detected objects;

[0042] The image generated in step A1 in step A3 is input into the training model generated in step A2, and each rectangular block after segmentation is sent into a multi-layer convolutional neural network, and the image is characterized by using the convolutional neural network Lift, and take the features of the last layer and input the softmax function to initially judge whether each rectangular block contains the license plate features of electric bicycles or motorcycles; if the rectangular blocks contain license plate features of electric bicycles or motorcy...

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 method for distinguishing electric bicycles and motorcycle through videos, and mainly aims at distinguishing license plates of the electric bicycles and motorcycles. The method comprises the following steps of: acquiring license plate features of electric bicycles and motorcycles on roads through a camera and generating an image; extracting features of the images througha deep learning-based training model and judging whether the motorcycles exist in the image or not; and if motorcycles are detected, outputting alarm information. According to the method, deep features of objects are extracted by adoption of a deep learning manner, and the extracted features have strong robustness, so that interferences of various environments can be overcome, the recognition correctness is greatly improved and the investment of manual supervision is greatly decreased; and compared with tradition recognition method, the precision of the method is greatly improved.

Description

technical field [0001] The invention belongs to the technical field of traffic video detection, in particular to a method for distinguishing electric bicycles and motorcycles through video. Background technique [0002] Because of its fast speed, small size, poor safety measures and other reasons, motorcycles are easy to cause traffic accidents when driving on the road, so they are the key monitoring objects of the transportation department. Road video supervision can effectively monitor motorcycle violations. However, electric bicycles and motorcycles have very similar characteristics, and sometimes it is difficult to distinguish electric bicycles from motorcycles, which brings certain troubles to road supervision. [0003] In the traditional method, electric bicycles and motorcycles are distinguished through image processing methods, but often due to driving speed, weather and other reasons, the generated images may appear motion blur and other phenomena. Therefore, a lot...

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): G06K9/00G06K9/62
CPCG06V20/54G06V20/46G06V20/625G06F18/214
Inventor 任帅
Owner CHENGDU REMARK TECH CO LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap