The authors survey a recent technique in computer vision called Interactive Co-segmentation, which is the task of simultaneously extracting common foreground objects from multiple related images. They survey several of the algorithms, present underlying common ideas, and give an overview of applications of object co-segmentation.
We survey a recent technique in computer vision called Co-segmentation, which is
the task of simultaneously extracting (or co-segmenting) common foreground objects
from multiple related images. The notion of co-segmentation was originally
introduced several years ago, and has since been extended in various ways.
In this manuscript, we review a number of these recent techniques in Chapter 1.
We present one technique for interactive co-segmentation, our recent work called
iCoseg in detail in Chapter 2. We provide an overview of applications of object
co-segmentation in Chapter 3 and finally discuss some exciting future directions in
Chapter 4.
We hope this manuscript gives the reader a combination of breadth and depth
on this topic, and will serve as a good introductory reference for the problem of
co-segmentation.