Loading...
Loading...

Go to the content (press return)

Recurrent instance segmentation using sequences of referring expressions

Author
Herrera, A.; Ventura, C.; Silberer, C.; Sorodoc, I.; Boleda, G.; Giro, X.
Type of activity
Presentation of work at congresses
Name of edition
NeurIPS 2019 Workshop
Date of publication
2019
Presentation's date
2019-12-13
Book of congress proceedings
NeurIPS 2019 Workshop on Visually Grounded Interaction and Language (ViGIL)
First page
1
Last page
6
Project funding
MALEGRA, TEC2016-75976-R
Repository
https://imatge.upc.edu/web/publications/recurrent-instance-segmentation-using-sequences-referring-expressions Open in new window
URL
https://vigilworkshop.github.io/static/papers/30.pdf Open in new window
Abstract
The goal of this work is segmenting the objects in an image which are referred to by a sequence of linguistic descriptions (referring expressions). We propose a deep neural network with recurrent layers that output a sequence of binary masks, one for each referring expression provided by the user. The recurrent layers in the architecture allow the model to condition each predicted mask on the previous ones, from a spatial perspective within the same image. Our multimodal approach uses off-the-sh...
Keywords
Deep learning, Instance segmentation, Referring expressions
Group of research
GPI - Image and Video Processing Group
IDEAI-UPC - Intelligent Data Science and Artificial Intelligence Research Center

Participants

  • Herrera Palacio, Alba  (author and speaker )
  • Ventura Royo, Carles  (author and speaker )
  • Silberer, Carina  (author and speaker )
  • Sorodoc, Ionut-Teodor  (author and speaker )
  • Boleda, Gemma  (author and speaker )
  • Giro Nieto, Xavier  (author and speaker )