Towards Unsupervised Image Harmonisation

Dolhasz, Alan and Harvey, Carlo and Williams, Ian (2020) Towards Unsupervised Image Harmonisation. In: 15th International Conference on Computer Vision Theory and Applications, 27.02.2020-29.02.2020, Valetta, Malta.

[img] Text
VisAPP_Camera_Ready.pdf - Accepted Version
Restricted to Repository staff only until 28 February 2021.

Download (7MB) | Request a copy

Abstract

The field of image synthesis intrinsically relies on the process of image compositing. This process can be automatic or manual, and depends upon artistic intent. Compositing can introduce errors, due to human-detectable differences in the general pixel level transforms of component elements of an image composite. We report on a pilot study evaluating a proof-of-concept automatic image composite harmonisation system consisting of a state-of-the-art deep harmonisation model and a perceptually-based composite luminance artifact detector. We evaluate the performance of both systems on a large data-set of 68128 automatically generated image composites and find that without any task-specific adaptations, the end-to-end system achieves comparable results to the baseline harmoniser fed with ground truth composite masks. We discuss these findings in the context of extending this to an end-to-end, multi-task system.

Item Type: Conference or Workshop Item (Paper)
Date: 28 February 2020
Uncontrolled Keywords: image compositing, harmonisation, artifact detection, end-to-end compositing, deep learning
Subjects: G400 Computer Science
G700 Artificial Intelligence
Divisions: Faculty of Computing, Engineering and the Built Environment
Faculty of Computing, Engineering and the Built Environment > School of Computing and Digital Technology
Faculty of Computing, Engineering and the Built Environment > School of Computing and Digital Technology > Digital Media Technology
Depositing User: Alan Dolhasz
Date Deposited: 01 Apr 2020 07:45
Last Modified: 02 Jul 2020 08:31
URI: http://www.open-access.bcu.ac.uk/id/eprint/9000

Actions (login required)

View Item View Item

Research

In this section...