Loading...
Loading...

Go to the content (press return)

Simple vs complex temporal recurrences for video saliency prediction

Author
Linardos, P.; Mohedano, E.; Nieto, J.; O'Connor, N.; Giro, X.; McGuinness, K.
Type of activity
Presentation of work at congresses
Name of edition
30th British Machine Vision Conference
Date of publication
2019
Presentation's date
2019-09-11
Book of congress proceedings
Proceedings of the 30th British Machine Vision Conference
First page
1
Last page
12
Repository
http://hdl.handle.net/2117/179382 Open in new window
URL
https://bmvc2019.org/wp-content/uploads/papers/0952-paper.pdf Open in new window
Abstract
This paper investigates modifying an existing neural network architecture for static saliency prediction using two types of recurrences that integrate information from the temporal domain. The first modification is the addition of a ConvLSTM within the architecture, while the second is a conceptually simple exponential moving average of an internal convolutional state. We use weights pre-trained on the SALICON dataset and fine-tune our model on DHF1K. Our results show that both modifications ach...
Citation
Linardos, P. [et al.]. Simple vs complex temporal recurrences for video saliency prediction. A: British Machine Vision Conference. "Proceedings of the 30th British Machine Vision Conference". 2019, p. 1-12.
Keywords
Computer vision, Deep learning, Saliency, Video
Group of research
GPI - Image and Video Processing Group
IDEAI-UPC - Intelligent Data Science and Artificial Intelligence Research Center

Participants

  • Linardos, Panagiotis  (author and speaker )
  • Mohedano, Eva  (author and speaker )
  • Nieto, Juan Jose  (author and speaker )
  • O'Connor, Noel  (author and speaker )
  • Giro Nieto, Xavier  (author and speaker )
  • McGuinness, Kevin  (author and speaker )