Vehicle-Related Scene Understanding Using Deep Learning

Abstract

Automated driving is an inevitable trend in future transportation, it is also one of the eminent achievements in the matter of artificial intelligence. Deep learning produces a significant contribution to the progression of automatic driving. In this paper, our goal is to primarily deal with the issue of vehicle-related scene understanding using deep learning. To the best of our knowledge, this is the first time that we utilize our traffic environment as an object for scene understanding based on deep learning. Moreover, automatic scene segmentation and object detection are joined for traffic scene understanding. The techniques based on deep learning dramatically decrease human manipulations. Furthermore, the datasets in this paper consist of a large amount of our collected traffic images. Meanwhile, the performance of our algorithms is verified by the experiential results.