Saliency Detection of Visual Signals: From 2D to 3D and Beyond - Part 1

The tutorial aims to firstly provide the audience with the basic knowledge, overview and current progress of saliency detection for various forms of visual signals. It introduces saliency detection from 2D visual signals to 3D visual signals, including images, videos, graphics, animation, stereoscopic images/videos, point clouds, and 360-degree images/videos. Specifically, it will describe saliency detection from the in-depth analysis of both the properties of different visual signals and the characteristics of the human visual system. Besides the traditional saliency detection models built on hand-crafted features, this tutorial will introduce the deep learning based models. The problems with existing datasets and performance evaluation methodology of saliency detection methods will be also discussed. Through the analysis of the challenging problems about saliency detection of visual signals, the tutorial also explores the potential research directions in the research community.