Radar Based Traffic Incident Detection using Support Vector Classification for Road Safety

Main Article Content

Pruk Ambavamata
Phongsak Keeratiwintakorn


This research presents the usage of Support Vector Classification (SVC) to classify incident of vehicles driving in the opposite traffic lane by using data collected from traffic radar. First, we develop a data collection program for gathering traffic data from a traffic radar sensor remotely located in Prachinburi province, Thailand. The gathered data are used in training and verifying our proposed SVC model. Secondly, we develop an Automatic Incident Detection (AID) program with SVC algorithm to detect desired incidents such as driving in an opposite lane. Our research is tested in 6 different environments such as the time of day and the average speed. In the results, we found that it can classify thepattern of driving in the opposite traffic lane, TTD is 1.616 seconds, DR is 99.85 percent and FAR is 1.74 percent. From the result, our proposed work with SVC algorithm can be efficiently applied to classify road incidentsfrom the data collected from traffic radar.

Article Details

Research Paper