Interactive image co-segmentation

Lu, Marco (2019) Interactive image co-segmentation. Bachelor's Thesis, Computing Science.

Preview	Text bCS_2019_LuM.pdf Download (19MB) | Preview
	Text Toestemming.pdf Restricted to Registered users only Download (116kB)

Abstract

Image co-segmentation segments common foreground objects simultaneously from a set of images. Multiple implementations are available for image co-segmentation all with varying results yet there is no simple way to test and compare the segmentations. Interactive image co-segmentation provides a framework to facilitate user input, providing an easier way to compare implementations and view varying results by changing parameters. The GUI facilitates a pipeline of algorithms for co-segmentation which consists of super pixel calculation, feature extraction from the superpixels and co-segmentation. The GUI allows selection of a set of images, changing the algorithm used for each pipe, changing the parameters values of an algorithm as well as marking foreground and background for each image to be later used for co-segmentation. The GUI also provides multiple ways to view results of each pipe and comparing the resulting image with a ground truth image, which is a user made result image to compare with the result of the pipeline. The GUI was implemented with one way to calculate superpixels, seven different ways to compute feature extraction which also can be used in combination with each other and two different ways to perform co-segmentation. The finished GUI makes it easier for users to interact and compare these algorithms and gives them a good idea how each parameter for an algorithm exactly impacts the calculations.

Item Type:	Thesis (Bachelor's Thesis)
Supervisor name:	Kosinka, J.
Degree programme:	Computing Science
Thesis type:	Bachelor's Thesis
Language:	English
Date Deposited:	11 Jul 2019
Last Modified:	12 Jul 2019 11:05
URI:	https://fse.studenttheses.ub.rug.nl/id/eprint/20114

Actions (login required)

View Item