(Publisher of Peer Reviewed Open Access Journals)

ACCENTS Transactions on Image Processing and Computer Vision (TIPCV)

ISSN (Print):    ISSN (Online):2455-4707
Volume-5 Issue-17 November-2019
Full-Text PDF
Paper Title : Traffic signal timing control using deep learning
Author Name : Vishnu , Vikram , Jayakanthan , Likhita and Paavai Anand
Abstract :

India is the second most populated country with 1.37 billion people so that avoiding traffic is impossible. But with proper traffic signal control method, we can control the amount of time spent in traffic. Our solution for this problem is to control the traffic signal timing and allocate more time length of green light to lanes containing a greater number of vehicles using deep learning-based computer vision approaches such as object detection. In January 2019, more than a million and a half (1,607,315) new vehicles were bought and registered all across the country. In which 74% of the vehicles were two-wheelers and more than 80% of the total vehicles were petrol driven. India has 5.5 million kilometres of road network while now the number of vehicles registered is three times greater. These single statistics should reveal why Indian roads are getting more congested every month. In 2017, a total of 4,64,910 road accidents have been reported in which 1,47,913 deaths occurred and 4,70,975 people were injured. An average of 1274 accidents and 405 deaths every day. By using deep learning for controlling traffic signals, we can clear traffic more effectively and reduce traffic congestion, traffic violations, accidents, fuel consumption, pollution and time in traffic.

Keywords : Deep learning, Computer vision, Object detection, YOLO v3.
Cite this article : Vishnu , Vikram , Jayakanthan , Likhita , Anand P. Traffic signal timing control using deep learning. ACCENTS Transactions on Image Processing and Computer Vision. 2019; 5 (17): 27-32. DOI:10.19101/TIPCV.2019.5150016.