Object Tracking for Automatic Driving

Abstract

With the development of automatic driving technology, the object tracking based on computer vision is being widely used nowadays. In this paper an overview of object tracking methods in automatic driving are presented. Kalman filtering, LSTM CNN, correlation filtering and Deep Affinity Network will be introduced. Kalman filtering and Kalman filtering extension algorithms and Correlation filtering have been combined with deep learning algorithms about object detection. Learning goals in end-to-end way the appearance of the object characteristics and correlation in several frame, including its appearance modeling study on the hierarchical characteristics of the object and its surrounding. Finally, we conclude the object tracking in automatic driving.