HOMEto discover WORK WITH USto achieve PUBLICATIONSto innovate FUNDINGto establish ACTIVITIESto engage PEOPLEto collaborate TEACHINGto inspire CONTACTSto connect

Self-funded PhD Positions Available

• Deep Learning based Computer Graphics for Virtual Human Synthesis
• Computer Vision with Deep Learning for Human Data Modelling
• Deep Learning and Computer Vision for Understanding the People Context in Artworks
• Machine Learning based Human Motion Analysis for Automatic Diseases Diagnosis

Illumination-based Data Augmentation for Robust Background Subtraction

Abstract

A core challenge in background subtraction (BGS) is handling videos with sudden illumination changes in consecutive frames. In this paper, we tackle the problem from a data point-of-view using data augmentation. Our method performs data augmentation that not only creates endless data on the fly, but also features semantic transformations of illumination which enhance the generalisation of the model. It successfully simulates flashes and shadows by applying the Euclidean distance transform over a binary mask generated randomly. Such data allows us to effectively train an illumination-invariant deep learning model for BGS. Experimental results demonstrate the contribution of the synthetics in the ability of the models to perform BGS even when significant illumination changes take place.

Publication

Dimitrios Sakkos, Hubert P. H. Shum and Edmond S. L. Ho,
"Illumination-based Data Augmentation for Robust Background Subtraction",
Proceedings of the 2019 International Conference on Software Knowledge Information Management and Applications (SKIMA), 2019
Citation=1^## Best Paper Award

## Citation counts from Google Scholar as of 2020

Downloads

References

BibTeX

@inproceedings{sakkos19triplet,
author={Sakkos, Dimitrios and Shum, Hubert P. H. and Ho, Edmond S. L.},
booktitle={Proceedings of the 2018 International Conference on Software Knowledge Information Management and Applications},
series={SKIMA '19},
title={Illumination-based Data Augmentation for Robust Background Subtraction},
year={2019},
numpages={8},
publisher={IEEE},
location={Island of Ukulhas, Maldives},
}

EndNote/RefMan

TY  - CONF
AU  - Sakkos, Dimitrios
AU  - Shum, Hubert P. H.
AU  - Ho, Edmond S. L.
T2  - Proceedings of the 2018 International Conference on Software Knowledge Information Management and Applications
TI  - Illumination-based Data Augmentation for Robust Background Subtraction
PY  - 2019
PB  - IEEE
ER  -

Plain Text

Dimitrios Sakkos, Hubert P. H. Shum and Edmond S. L. Ho, "Illumination-based Data Augmentation for Robust Background Subtraction," in SKIMA '19: Proceedings of the 2018 International Conference on Software Knowledge Information Management and Applications, Island of Ukulhas, Maldives, IEEE, 2019.