A Study on Classification of Traffic Accident Injury Grade Using CNN and NASS-CDS Data

Abstract

In this paper, we propose a new occupant injury prediction algorithm using real car accident data base, NASS-CDS DB. Field crash data which are collected by IIHS is used as input of convolutional neural network to estimate occupant injury. And in order to applying CNN, we are encoding crash information data as two dimensional image. Our experiment results can be shown the validity of our proposed algorithm.