Video Object Tracking : Tasks, Datasets, and Methods

This book provides a thorough overview of recent progress in video object tracking, allowing researchers and industrial practitioners to gain a better understanding of the most important problems and developed technologies in the area. Video tracking is a key research area in computer vision and aims to track unique objects in a given video, which are useful for various applications such as video conference, video editing, surveillance, and autonomous driving. This book begins with an introduction to the task of video object tracking, including the most common problem settings. Given the revolution of deep learning in computer vision problems, numerous new tasks, datasets, and methods have been recently proposed in the domain of video tracking. The book includes these recent results as well as benchmarks in large-scale human-centric video analysis in complex events.