Development and Evaluation of Technologies for Constructing Driving Video Database

抄録

The safety of automated driving systems, which are regarded as a promising means of helping to reduce traffic accidents, is greatly dependent on functions for recognizing pedestrians and objects, and on making early and accurate judgments of potential collision risks. The collection of large volumes of driving data and the development of technologies and tools for using this data efficiently are recognized as common issues for strengthening competitiveness in the development of such recognition and judgment functions. This article describes efforts undertaken to research and evaluate the utility of a driving video database and simulation evaluation technology needed for developing camera image recognition technology.