東華大學圖書館 |

Language: English

Help

回圖書館首頁

手機版館藏查詢

Back to Search results for [ null ]

Switch To: Labeled | MARC Mode | ISBD

Linked to FindBook

Google Book

Amazon

博客來

Warning System in Smart Vehicles for Detection of Traffic Signs, Lights, and Obstacles.

Record Type:	Electronic resources : Monograph/item
Title/Author:	Warning System in Smart Vehicles for Detection of Traffic Signs, Lights, and Obstacles./
Author:	Madupuri, Suhas.
Published:	Ann Arbor : ProQuest Dissertations & Theses, : 2021,
Description:	69 p.
Notes:	Source: Masters Abstracts International, Volume: 83-07.
Contained By:	Masters Abstracts International83-07.
Subject:	Computer science. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=28866484
ISBN:	9798759979524

Warning System in Smart Vehicles for Detection of Traffic Signs, Lights, and Obstacles.
Madupuri, Suhas.

Warning System in Smart Vehicles for Detection of Traffic Signs, Lights, and Obstacles. - Ann Arbor : ProQuest Dissertations & Theses, 2021 - 69 p.

Source: Masters Abstracts International, Volume: 83-07.

Thesis (M.S.)--Texas A&M University - Kingsville, 2021.

This item must not be sold to any third party vendors.

Vehicle accidents can occur due to driver negligence and failure to follow traffic rules. For safe driving, the driver needs to be aware of different situations occurring on the road while driving. To improve a driver's ability to maintain their focus and reduce accidents, this research proposes to enhance an automated driver early warning system. Such a system would alert the driver whenever a traffic object is detected. This research specifically focuses on improving detection of traffic signs, signals, pedestrians, and obstacles. In this work, a pre-processing enhancement with ESRGAN (Enhanced Super Resolution Generative Adversarial Network) is employed for enhancing the contrast and resolution of the extracted traffic objects. Many advancements have been made in object detection and image processing, but vehicle systems identifying the traffic signs, lights, pedestrians, and obstacles still need improvement. This research also compares the accuracy of two algorithms, You Only Look Once version 3 (YOLOv3) and You Only Look Once version 4 (YOLOv4) with and without the preprocessing ESRGAN enhancement. This research would facilitate advancing automatic systems in autonomous vehicles for sensing traffic objects in real-time. The main objective of this research is to increase the performance of traffic object detection in terms of accuracy, and compare deep learning techniques to see which would result in better performance. The proposed enhancement in this research is applying preprocessing using ESRGAN to the traffic objects, which increases the accuracy of the traffic object detection in different conditions such as nighttime, rain, or extreme weather conditions. The enhanced YOLOv3 model was found to have an average accuracy of 92.83% for traffic object detection, compared to only 87.11% for the unenhanced YOLOv3 model, which is a 5.72% improvement. The enhanced YOLOv4 model was found to have an average accuracy of 94.16% for traffic object detection, compared to only 87.32% for the unenhanced YOLOv4 model, which is a 6.84% improvement.

ISBN: 9798759979524Subjects--Topical Terms:

523869
Computer science.
Subjects--Index Terms:

CNN

Warning System in Smart Vehicles for Detection of Traffic Signs, Lights, and Obstacles.
LDR:03426nmm a2200421 4500 001 2347706
005 20220823142337.5
008 241004s2021 ||||||||||||||||| ||eng d
020 $a 9798759979524
035 $a (MiAaPQ)AAI28866484
035 $a AAI28866484
040 $a MiAaPQ $c MiAaPQ
100 1 $a Madupuri, Suhas. $3 3686993
245 1 0 $a Warning System in Smart Vehicles for Detection of Traffic Signs, Lights, and Obstacles.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2021
300 $a 69 p.
500 $a Source: Masters Abstracts International, Volume: 83-07.
500 $a Advisor: Goyal, Ayush.
502 $a Thesis (M.S.)--Texas A&M University - Kingsville, 2021.
506 $a This item must not be sold to any third party vendors.
520 $a Vehicle accidents can occur due to driver negligence and failure to follow traffic rules. For safe driving, the driver needs to be aware of different situations occurring on the road while driving. To improve a driver's ability to maintain their focus and reduce accidents, this research proposes to enhance an automated driver early warning system. Such a system would alert the driver whenever a traffic object is detected. This research specifically focuses on improving detection of traffic signs, signals, pedestrians, and obstacles. In this work, a pre-processing enhancement with ESRGAN (Enhanced Super Resolution Generative Adversarial Network) is employed for enhancing the contrast and resolution of the extracted traffic objects. Many advancements have been made in object detection and image processing, but vehicle systems identifying the traffic signs, lights, pedestrians, and obstacles still need improvement. This research also compares the accuracy of two algorithms, You Only Look Once version 3 (YOLOv3) and You Only Look Once version 4 (YOLOv4) with and without the preprocessing ESRGAN enhancement. This research would facilitate advancing automatic systems in autonomous vehicles for sensing traffic objects in real-time. The main objective of this research is to increase the performance of traffic object detection in terms of accuracy, and compare deep learning techniques to see which would result in better performance. The proposed enhancement in this research is applying preprocessing using ESRGAN to the traffic objects, which increases the accuracy of the traffic object detection in different conditions such as nighttime, rain, or extreme weather conditions. The enhanced YOLOv3 model was found to have an average accuracy of 92.83% for traffic object detection, compared to only 87.11% for the unenhanced YOLOv3 model, which is a 5.72% improvement. The enhanced YOLOv4 model was found to have an average accuracy of 94.16% for traffic object detection, compared to only 87.32% for the unenhanced YOLOv4 model, which is a 6.84% improvement.
590 $a School code: 1187.
650 4 $a Computer science. $3 523869
650 4 $a Artificial intelligence. $3 516317
650 4 $a Automotive engineering. $3 2181195
650 4 $a Robotics. $3 519753
653 $a CNN
653 $a COCO datasets
653 $a Enhanced Super Resolution Generative Adversarial Network
653 $a You Only Look Once versions 3 and 4
653 $a Automated driver early warning system
653 $a Object detection
653 $a Image processing
653 $a Convolutional neural networks
690 $a 0984
690 $a 0800
690 $a 0540
690 $a 0771
710 2 $a Texas A&M University - Kingsville. $b Computer Science and Electrical Engineering. $3 3686994
773 0 $t Masters Abstracts International $g 83-07.
790 $a 1187
791 $a M.S.
792 $a 2021
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=28866484