A Deep Learning Framework for Joint Image Restoration and Recognition

Ruilong Chen, Lyudmila Mihaylova, Hao Zhu, Nidhal Carla Bouaynaya

Research output: Contribution to journalArticle

Abstract

Image restoration and recognition are important computer vision tasks representing an inherent part of autonomous systems. These two tasks are often implemented in a sequential manner, in which the restoration process is followed by a recognition. In contrast, this paper proposes a joint framework that simultaneously performs both tasks within a shared deep neural network architecture. This joint framework integrates the restoration and recognition tasks by incorporating: (i) common layers, (ii) restoration layers and (iii) classification layers. The total loss function combines the restoration and classification losses. The proposed joint framework, based on capsules, provides an efficient solution that can cope with challenges due to noise, image rotations and occlusions. The developed framework has been validated and evaluated on a public vehicle logo dataset under various degradation conditions, including Gaussian noise, rotation and occlusion. The results show that the joint framework improves the accuracy compared with the single task networks.

Original languageEnglish (US)
JournalCircuits, Systems, and Signal Processing
DOIs
StateAccepted/In press - Jan 1 2019

Fingerprint

Image recognition
Image Recognition
Image Restoration
Image reconstruction
Restoration
Occlusion
Network architecture
Computer vision
Gaussian Noise
Network Architecture
Loss Function
Autonomous Systems
Efficient Solution
Computer Vision
Degradation
Integrate
Learning
Framework
Deep learning
Neural Networks

All Science Journal Classification (ASJC) codes

  • Signal Processing
  • Applied Mathematics

Cite this

@article{39544cad011a4aa782ff0daf0ff762d7,
title = "A Deep Learning Framework for Joint Image Restoration and Recognition",
abstract = "Image restoration and recognition are important computer vision tasks representing an inherent part of autonomous systems. These two tasks are often implemented in a sequential manner, in which the restoration process is followed by a recognition. In contrast, this paper proposes a joint framework that simultaneously performs both tasks within a shared deep neural network architecture. This joint framework integrates the restoration and recognition tasks by incorporating: (i) common layers, (ii) restoration layers and (iii) classification layers. The total loss function combines the restoration and classification losses. The proposed joint framework, based on capsules, provides an efficient solution that can cope with challenges due to noise, image rotations and occlusions. The developed framework has been validated and evaluated on a public vehicle logo dataset under various degradation conditions, including Gaussian noise, rotation and occlusion. The results show that the joint framework improves the accuracy compared with the single task networks.",
author = "Ruilong Chen and Lyudmila Mihaylova and Hao Zhu and Bouaynaya, {Nidhal Carla}",
year = "2019",
month = "1",
day = "1",
doi = "10.1007/s00034-019-01222-x",
language = "English (US)",
journal = "Circuits, Systems, and Signal Processing",
issn = "0278-081X",
publisher = "Birkhause Boston",

}

A Deep Learning Framework for Joint Image Restoration and Recognition. / Chen, Ruilong; Mihaylova, Lyudmila; Zhu, Hao; Bouaynaya, Nidhal Carla.

In: Circuits, Systems, and Signal Processing, 01.01.2019.

Research output: Contribution to journalArticle

TY - JOUR

T1 - A Deep Learning Framework for Joint Image Restoration and Recognition

AU - Chen, Ruilong

AU - Mihaylova, Lyudmila

AU - Zhu, Hao

AU - Bouaynaya, Nidhal Carla

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Image restoration and recognition are important computer vision tasks representing an inherent part of autonomous systems. These two tasks are often implemented in a sequential manner, in which the restoration process is followed by a recognition. In contrast, this paper proposes a joint framework that simultaneously performs both tasks within a shared deep neural network architecture. This joint framework integrates the restoration and recognition tasks by incorporating: (i) common layers, (ii) restoration layers and (iii) classification layers. The total loss function combines the restoration and classification losses. The proposed joint framework, based on capsules, provides an efficient solution that can cope with challenges due to noise, image rotations and occlusions. The developed framework has been validated and evaluated on a public vehicle logo dataset under various degradation conditions, including Gaussian noise, rotation and occlusion. The results show that the joint framework improves the accuracy compared with the single task networks.

AB - Image restoration and recognition are important computer vision tasks representing an inherent part of autonomous systems. These two tasks are often implemented in a sequential manner, in which the restoration process is followed by a recognition. In contrast, this paper proposes a joint framework that simultaneously performs both tasks within a shared deep neural network architecture. This joint framework integrates the restoration and recognition tasks by incorporating: (i) common layers, (ii) restoration layers and (iii) classification layers. The total loss function combines the restoration and classification losses. The proposed joint framework, based on capsules, provides an efficient solution that can cope with challenges due to noise, image rotations and occlusions. The developed framework has been validated and evaluated on a public vehicle logo dataset under various degradation conditions, including Gaussian noise, rotation and occlusion. The results show that the joint framework improves the accuracy compared with the single task networks.

UR - http://www.scopus.com/inward/record.url?scp=85070239281&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85070239281&partnerID=8YFLogxK

U2 - 10.1007/s00034-019-01222-x

DO - 10.1007/s00034-019-01222-x

M3 - Article

AN - SCOPUS:85070239281

JO - Circuits, Systems, and Signal Processing

JF - Circuits, Systems, and Signal Processing

SN - 0278-081X

ER -