Biologically Inspired Deep Learning Model for Efficient Foveal-Peripheral Vision

While abundant in biology, foveated vision is nearly absent from computational models and especially deep learning architectures. Despite considerable hardware improvements, training deep neural networks still presents a challenge and constraints complexity of models. Here we propose an end-to-end n...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Hristofor Lukanov, Peter König, Gordon Pipa
Formato: article
Lenguaje:EN
Publicado: Frontiers Media S.A. 2021
Materias:
Acceso en línea:https://doaj.org/article/dd5d9cfd27fd412e8e6bb0d0b7609983
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
id oai:doaj.org-article:dd5d9cfd27fd412e8e6bb0d0b7609983
record_format dspace
spelling oai:doaj.org-article:dd5d9cfd27fd412e8e6bb0d0b76099832021-11-22T05:13:39ZBiologically Inspired Deep Learning Model for Efficient Foveal-Peripheral Vision1662-518810.3389/fncom.2021.746204https://doaj.org/article/dd5d9cfd27fd412e8e6bb0d0b76099832021-11-01T00:00:00Zhttps://www.frontiersin.org/articles/10.3389/fncom.2021.746204/fullhttps://doaj.org/toc/1662-5188While abundant in biology, foveated vision is nearly absent from computational models and especially deep learning architectures. Despite considerable hardware improvements, training deep neural networks still presents a challenge and constraints complexity of models. Here we propose an end-to-end neural model for foveal-peripheral vision, inspired by retino-cortical mapping in primates and humans. Our model has an efficient sampling technique for compressing the visual signal such that a small portion of the scene is perceived in high resolution while a large field of view is maintained in low resolution. An attention mechanism for performing “eye-movements” assists the agent in collecting detailed information incrementally from the observed scene. Our model achieves comparable results to a similar neural architecture trained on full-resolution data for image classification and outperforms it at video classification tasks. At the same time, because of the smaller size of its input, it can reduce computational effort tenfold and uses several times less memory. Moreover, we present an easy to implement bottom-up and top-down attention mechanism which relies on task-relevant features and is therefore a convenient byproduct of the main architecture. Apart from its computational efficiency, the presented work provides means for exploring active vision for agent training in simulated environments and anthropomorphic robotics.Hristofor LukanovPeter KönigPeter KönigGordon PipaFrontiers Media S.A.articlespace-variant visionactive visionfoveal visionperipheral visiondeep learning-artificial neural network (DL-ANN)bottom-up attentionNeurosciences. Biological psychiatry. NeuropsychiatryRC321-571ENFrontiers in Computational Neuroscience, Vol 15 (2021)
institution DOAJ
collection DOAJ
language EN
topic space-variant vision
active vision
foveal vision
peripheral vision
deep learning-artificial neural network (DL-ANN)
bottom-up attention
Neurosciences. Biological psychiatry. Neuropsychiatry
RC321-571
spellingShingle space-variant vision
active vision
foveal vision
peripheral vision
deep learning-artificial neural network (DL-ANN)
bottom-up attention
Neurosciences. Biological psychiatry. Neuropsychiatry
RC321-571
Hristofor Lukanov
Peter König
Peter König
Gordon Pipa
Biologically Inspired Deep Learning Model for Efficient Foveal-Peripheral Vision
description While abundant in biology, foveated vision is nearly absent from computational models and especially deep learning architectures. Despite considerable hardware improvements, training deep neural networks still presents a challenge and constraints complexity of models. Here we propose an end-to-end neural model for foveal-peripheral vision, inspired by retino-cortical mapping in primates and humans. Our model has an efficient sampling technique for compressing the visual signal such that a small portion of the scene is perceived in high resolution while a large field of view is maintained in low resolution. An attention mechanism for performing “eye-movements” assists the agent in collecting detailed information incrementally from the observed scene. Our model achieves comparable results to a similar neural architecture trained on full-resolution data for image classification and outperforms it at video classification tasks. At the same time, because of the smaller size of its input, it can reduce computational effort tenfold and uses several times less memory. Moreover, we present an easy to implement bottom-up and top-down attention mechanism which relies on task-relevant features and is therefore a convenient byproduct of the main architecture. Apart from its computational efficiency, the presented work provides means for exploring active vision for agent training in simulated environments and anthropomorphic robotics.
format article
author Hristofor Lukanov
Peter König
Peter König
Gordon Pipa
author_facet Hristofor Lukanov
Peter König
Peter König
Gordon Pipa
author_sort Hristofor Lukanov
title Biologically Inspired Deep Learning Model for Efficient Foveal-Peripheral Vision
title_short Biologically Inspired Deep Learning Model for Efficient Foveal-Peripheral Vision
title_full Biologically Inspired Deep Learning Model for Efficient Foveal-Peripheral Vision
title_fullStr Biologically Inspired Deep Learning Model for Efficient Foveal-Peripheral Vision
title_full_unstemmed Biologically Inspired Deep Learning Model for Efficient Foveal-Peripheral Vision
title_sort biologically inspired deep learning model for efficient foveal-peripheral vision
publisher Frontiers Media S.A.
publishDate 2021
url https://doaj.org/article/dd5d9cfd27fd412e8e6bb0d0b7609983
work_keys_str_mv AT hristoforlukanov biologicallyinspireddeeplearningmodelforefficientfovealperipheralvision
AT peterkonig biologicallyinspireddeeplearningmodelforefficientfovealperipheralvision
AT peterkonig biologicallyinspireddeeplearningmodelforefficientfovealperipheralvision
AT gordonpipa biologicallyinspireddeeplearningmodelforefficientfovealperipheralvision
_version_ 1718418192391471104