Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
A METHOD FOR DETECTING SEATBELT VIOLATION BY USING IMAGES OF LICENSE PLATE RECOGNITION SYSTEM
Document Type and Number:
WIPO Patent Application WO/2020/076264
Kind Code:
A3
Abstract:
The inventive product is a method for ensuring to learn by way of forming multiple deep learning models by means of processing image received from License Plate Recognition System (LPRS) camera, which is not of standard size, dimension and resolution with various process steps, and informing security units on seatbelt violations by way of said learning technique. Accordingly, the camera data is received from a server in the police station over Internet connection and a model is formed in these camera images to detect windshield; the windshield is detected by using this model. It is formed a model for detecting passenger and driver in the windshield area; the detection of driver and passenger is carried out within the windshield area on said model. It is formed a model for detecting seatbelt on the driver and passenger area and the seatbelt detection result is formed on the model depending on score values of the seatbelt detection on the example passenger images.

Inventors:
BASESKI EMRE (TR)
BALCI BURAK (TR)
ARTAN YUSUF OĞUZHAN (TR)
ELIHOS ALPEREN (TR)
Application Number:
PCT/TR2019/050500
Publication Date:
June 25, 2020
Filing Date:
June 25, 2019
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
HAVELSAN HAVA ELEKTRONIK SANAYI VE TICARET ANONIM SIRKETI (TR)
International Classes:
G06K9/00; B60R22/00; G06N20/00
Foreign References:
US20160078306A12016-03-17
US20130336538A12013-12-19
US20150286884A12015-10-08
Attorney, Agent or Firm:
BEKTAŞ, Ekrem Erdal (TR)
Download PDF: