eduzhai > Applied Sciences > Engineering >

Defining Traffic States using Spatio-temporal Traffic Graphs

  • king
  • (0) Download
  • 20210507
  • Save

... pages left unread,continue reading

Document pages: 6 pages

Abstract: Intersections are one of the main sources of congestion and hence, it isimportant to understand traffic behavior at intersections. Particularly, indeveloping countries with high vehicle density, mixed traffic type, andlane-less driving behavior, it is difficult to distinguish between congestedand normal traffic behavior. In this work, we propose a way to understand thetraffic state of smaller spatial regions at intersections using traffic graphs.The way these traffic graphs evolve over time reveals different traffic states- a) a congestion is forming (clumping), the congestion is dispersing(unclumping), or c) the traffic is flowing normally (neutral). We train aspatio-temporal deep network to identify these changes. Also, we introduce alarge dataset called EyeonTraffic (EoT) containing 3 hours of aerial videoscollected at 3 busy intersections in Ahmedabad, India. Our experiments on theEoT dataset show that the traffic graphs can help in correctly identifyingcongestion-prone behavior in different spatial regions of an intersection.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...