3DSAL: An Efficient 3D-CNN Architecture for Video Saliency Prediction

Yasser Djilali, Mohamed Sayah, Kevin McGuinness, Noel O’Connor

Abstract

In this paper, we propose a novel 3D CNN architecture that enables us to train an effective video saliency prediction model. The model is designed to capture important motion information using multiple adjacent frames. Our model performs a cubic convolution on a set of consecutive frames to extract spatio-temporal features. This enables us to predict the saliency map for any given frame using past frames. We comprehensively investigate the performance of our model with respect to state-of-the-art video saliency models. Experimental results on three large-scale datasets, DHF1K, UCF-SPORTS and DAVIS, demonstrate the competitiveness of our approach.

Download


Paper Citation


in Harvard Style

Djilali Y., Sayah M., McGuinness K. and O’Connor N. (2020). 3DSAL: An Efficient 3D-CNN Architecture for Video Saliency Prediction.In Proceedings of the 15th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 4: VISAPP, ISBN 978-989-758-402-2, pages 27-36. DOI: 10.5220/0008875600270036


in Bibtex Style

@conference{visapp20,
author={Yasser Djilali and Mohamed Sayah and Kevin McGuinness and Noel O’Connor},
title={3DSAL: An Efficient 3D-CNN Architecture for Video Saliency Prediction},
booktitle={Proceedings of the 15th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 4: VISAPP,},
year={2020},
pages={27-36},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0008875600270036},
isbn={978-989-758-402-2},
}


in EndNote Style

TY - CONF

JO - Proceedings of the 15th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 4: VISAPP,
TI - 3DSAL: An Efficient 3D-CNN Architecture for Video Saliency Prediction
SN - 978-989-758-402-2
AU - Djilali Y.
AU - Sayah M.
AU - McGuinness K.
AU - O’Connor N.
PY - 2020
SP - 27
EP - 36
DO - 10.5220/0008875600270036