AMNet: Memorability Estimation with Attention

Robot Vision Team (RoViT) Kingston University London, UK

Technical demonstration of AMNet for memorability estimation, presented at CVPR2018
(by Jiri Fajtl1, Vasileios Argyriou1, Dorothy Monekosso2, Paolo Remagnino1
1Kingston University, London, 2Leeds Beckett University, Leeds)

Publication: arXiv, Source code: github

Waiting for image(s) to process...


Tap or Drag & Drop images here
or click to select from filesystem
(Up to 100 images, each max 1MB size)


This application lets you to find out how memorable are your images. Memorability
score is a number between 0 and 1 that expresses how likely it is for healthy human
to recall an image after seeing it once. Our model predicts memorability in a sequence
of three "gazes" on image regions that best correlate with its memorability.

This work is co-funded by the NATO within the WITNESS project under grant agreement number G5437.
We gratefully acknowledge the support of NVIDIA Corporation with the donation of the Titan Xp GPU used for this research.

Please cite our work as:
				
@inproceedings{fajtl2018amnet,
  title={AMNet: Memorability Estimation with Attention},
  author={Fajtl, Jiri and Argyriou, Vasileios and Monekosso, Dorothy and Remagnino, Paolo},
  booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
  pages={6363--6372},
  year={2018}
}
  



11/07/2018, Kingston University London, contact: Jiri Fajtl J.Fajtl@kingston.ac.uk