Recurrent Mixture Density Network for Spatiotemporal Visual Attention
L. Bazzani, H. Larochelle, L. Torresani
International Conference on Learning Representations (ICLR), 2017
OpenReview /
video /
bibtex
@conference{Bazzani:ICLR2017,
title={Recurrent Mixture Density Network for Spatiotemporal Visual Attention},
author={Bazzani, Loris and Larochelle, Hugo and Torresani, Lorenzo},
booktitle={International Conference on Learning Representations (ICLR)},
year={2017}
}
We propose an attentional model that learns where to look in a video directly from human fixation data. The model is a combination of 3D ConvNet, RNN and mixture density network which models the saliency map.