Deep convolutional neural network architecture for urban traffic flow estimation

Sabbani, Imad (Faculty of sciences and techniques, Hassan II University, Mohammedia, Morocco) ; Perez-Uribe, Andres (School of Management and Engineering Vaud, HES-SO // University of Applied Sciences Western Switzerland) ; Bouattane, Omar (Faculty of sciences and techniques, Hassan II University, Mohammedia, Morocco) ; El Moudni, Abdellah (Faculty of sciences and techniques, Franche-Comté University, Besançon, France)

Road traffic density estimation can be very helpful for the successful deployment of intelligent Transportation systems. In this paper, we introduce a deep convolutional neural network (DCNN) based method that learns traffic density from pre-labeled images in order to estimate the traffic flow density in highways. Our method classifies the traffic flow density into three different states: light, medium and heavy. A standard database of real videos from Seattle roads was used to develop our proposed approach. The cross-validation and the class activation mapping techniques were employed in this work, in order to evaluate the performance of our method. The results show that our model outperformed all the existing conventional methods by reaching the highest accuracy of 99,62%.


Keywords:
Article Type:
scientifique
Faculty:
Ingénierie et Architecture
School:
HEIG-VD
Institute:
IICT - Institut des Technologies de l'Information et de la Communication
Subject(s):
Ingénierie
Date:
2018-07
Pagination:
7 p.
Published in:
IJCSNS International Journal of Computer Science and Network Security
Numeration (vol. no.):
2018, vol. 18, no- 7, pp. 69-75
ISSN:
1738-7906
External resources:
Appears in Collection:



 Record created 2018-12-04, last modified 2018-12-11

Fulltext:
Download fulltext
PDF

Rate this document:

Rate this document:
1
2
3
 
(Not yet reviewed)