Discovering physics and design trends from visual temporal structures

Abstract

Living in a constantly changing world, we cannot help but notice the temporal regularities of visual changes around us. These changes can be irreversible governed by physical laws, such as glass bottles broken into pieces, or influenced by design trends, such as web pages adopting templates with larger background images. In this dissertation, we build computational models to discover and apply the knowledge of the physics for arrow of time, and the design trends for web pages from image sequences. In the first part of the thesis, I train models to learn the visual cues that are indicative of the arrow of time from large real world video datasets. In the second part of the thesis, I investigate the evolution of visual cues and layout in web page design through screenshots over time. The knowledge of these visual temporal structures are not only of scientific interest by themselves, but also of practical uses demonstrated in this thesis.