Cargando…
Deep saliency models learn low-, mid-, and high-level features to predict scene attention
Deep saliency models represent the current state-of-the-art for predicting where humans look in real-world scenes. However, for deep saliency models to inform cognitive theories of attention, we need to know how deep saliency models prioritize different scene features to predict where people look. H...
Autores principales: | , |
---|---|
Formato: | Online Artículo Texto |
Lenguaje: | English |
Publicado: |
Nature Publishing Group UK
2021
|
Materias: | |
Acceso en línea: | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8445969/ https://www.ncbi.nlm.nih.gov/pubmed/34531484 http://dx.doi.org/10.1038/s41598-021-97879-z |
_version_ | 1784568772563042304 |
---|---|
author | Hayes, Taylor R. Henderson, John M. |
author_facet | Hayes, Taylor R. Henderson, John M. |
author_sort | Hayes, Taylor R. |
collection | PubMed |
description | Deep saliency models represent the current state-of-the-art for predicting where humans look in real-world scenes. However, for deep saliency models to inform cognitive theories of attention, we need to know how deep saliency models prioritize different scene features to predict where people look. Here we open the black box of three prominent deep saliency models (MSI-Net, DeepGaze II, and SAM-ResNet) using an approach that models the association between attention, deep saliency model output, and low-, mid-, and high-level scene features. Specifically, we measured the association between each deep saliency model and low-level image saliency, mid-level contour symmetry and junctions, and high-level meaning by applying a mixed effects modeling approach to a large eye movement dataset. We found that all three deep saliency models were most strongly associated with high-level and low-level features, but exhibited qualitatively different feature weightings and interaction patterns. These findings suggest that prominent deep saliency models are primarily learning image features associated with high-level scene meaning and low-level image saliency and highlight the importance of moving beyond simply benchmarking performance. |
format | Online Article Text |
id | pubmed-8445969 |
institution | National Center for Biotechnology Information |
language | English |
publishDate | 2021 |
publisher | Nature Publishing Group UK |
record_format | MEDLINE/PubMed |
spelling | pubmed-84459692021-09-20 Deep saliency models learn low-, mid-, and high-level features to predict scene attention Hayes, Taylor R. Henderson, John M. Sci Rep Article Deep saliency models represent the current state-of-the-art for predicting where humans look in real-world scenes. However, for deep saliency models to inform cognitive theories of attention, we need to know how deep saliency models prioritize different scene features to predict where people look. Here we open the black box of three prominent deep saliency models (MSI-Net, DeepGaze II, and SAM-ResNet) using an approach that models the association between attention, deep saliency model output, and low-, mid-, and high-level scene features. Specifically, we measured the association between each deep saliency model and low-level image saliency, mid-level contour symmetry and junctions, and high-level meaning by applying a mixed effects modeling approach to a large eye movement dataset. We found that all three deep saliency models were most strongly associated with high-level and low-level features, but exhibited qualitatively different feature weightings and interaction patterns. These findings suggest that prominent deep saliency models are primarily learning image features associated with high-level scene meaning and low-level image saliency and highlight the importance of moving beyond simply benchmarking performance. Nature Publishing Group UK 2021-09-16 /pmc/articles/PMC8445969/ /pubmed/34531484 http://dx.doi.org/10.1038/s41598-021-97879-z Text en © The Author(s) 2021 https://creativecommons.org/licenses/by/4.0/Open AccessThis article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ (https://creativecommons.org/licenses/by/4.0/) . |
spellingShingle | Article Hayes, Taylor R. Henderson, John M. Deep saliency models learn low-, mid-, and high-level features to predict scene attention |
title | Deep saliency models learn low-, mid-, and high-level features to predict scene attention |
title_full | Deep saliency models learn low-, mid-, and high-level features to predict scene attention |
title_fullStr | Deep saliency models learn low-, mid-, and high-level features to predict scene attention |
title_full_unstemmed | Deep saliency models learn low-, mid-, and high-level features to predict scene attention |
title_short | Deep saliency models learn low-, mid-, and high-level features to predict scene attention |
title_sort | deep saliency models learn low-, mid-, and high-level features to predict scene attention |
topic | Article |
url | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8445969/ https://www.ncbi.nlm.nih.gov/pubmed/34531484 http://dx.doi.org/10.1038/s41598-021-97879-z |
work_keys_str_mv | AT hayestaylorr deepsaliencymodelslearnlowmidandhighlevelfeaturestopredictsceneattention AT hendersonjohnm deepsaliencymodelslearnlowmidandhighlevelfeaturestopredictsceneattention |