A Novel Domain Transfer-Based Approach for Unsupervised Thermal Image Super-Resolution
This paper presents a transfer domain strategy to tackle the limitations of low-resolution thermal sensors and generate higher-resolution images of reasonable quality. The proposed technique employs a CycleGAN architecture and uses a ResNet as an encoder in the generator along with an attention modu...
Main Authors: | , , , |
---|---|
Other Authors: | |
Format: | Article |
Published: |
Multidisciplinary Digital Publishing Institute
2022
|
Online Access: | https://hdl.handle.net/1721.1/141368 |
_version_ | 1811084380784820224 |
---|---|
author | Rivadeneira, Rafael E. Sappa, Angel D. Vintimilla, Boris X. Hammoud, Riad |
author2 | Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory |
author_facet | Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory Rivadeneira, Rafael E. Sappa, Angel D. Vintimilla, Boris X. Hammoud, Riad |
author_sort | Rivadeneira, Rafael E. |
collection | MIT |
description | This paper presents a transfer domain strategy to tackle the limitations of low-resolution thermal sensors and generate higher-resolution images of reasonable quality. The proposed technique employs a CycleGAN architecture and uses a ResNet as an encoder in the generator along with an attention module and a novel loss function. The network is trained on a multi-resolution thermal image dataset acquired with three different thermal sensors. Results report better performance benchmarking results on the 2nd CVPR-PBVS-2021 thermal image super-resolution challenge than state-of-the-art methods. The code of this work is available online. |
first_indexed | 2024-09-23T12:49:31Z |
format | Article |
id | mit-1721.1/141368 |
institution | Massachusetts Institute of Technology |
last_indexed | 2024-09-23T12:49:31Z |
publishDate | 2022 |
publisher | Multidisciplinary Digital Publishing Institute |
record_format | dspace |
spelling | mit-1721.1/1413682024-01-05T21:09:46Z A Novel Domain Transfer-Based Approach for Unsupervised Thermal Image Super-Resolution Rivadeneira, Rafael E. Sappa, Angel D. Vintimilla, Boris X. Hammoud, Riad Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory This paper presents a transfer domain strategy to tackle the limitations of low-resolution thermal sensors and generate higher-resolution images of reasonable quality. The proposed technique employs a CycleGAN architecture and uses a ResNet as an encoder in the generator along with an attention module and a novel loss function. The network is trained on a multi-resolution thermal image dataset acquired with three different thermal sensors. Results report better performance benchmarking results on the 2nd CVPR-PBVS-2021 thermal image super-resolution challenge than state-of-the-art methods. The code of this work is available online. 2022-03-24T19:01:58Z 2022-03-24T19:01:58Z 2022-03-14 2022-03-24T14:46:53Z Article http://purl.org/eprint/type/JournalArticle https://hdl.handle.net/1721.1/141368 Sensors 22 (6): 2254 (2022) PUBLISHER_CC http://dx.doi.org/10.3390/s22062254 Creative Commons Attribution https://creativecommons.org/licenses/by/4.0/ application/pdf Multidisciplinary Digital Publishing Institute Multidisciplinary Digital Publishing Institute |
spellingShingle | Rivadeneira, Rafael E. Sappa, Angel D. Vintimilla, Boris X. Hammoud, Riad A Novel Domain Transfer-Based Approach for Unsupervised Thermal Image Super-Resolution |
title | A Novel Domain Transfer-Based Approach for Unsupervised Thermal Image Super-Resolution |
title_full | A Novel Domain Transfer-Based Approach for Unsupervised Thermal Image Super-Resolution |
title_fullStr | A Novel Domain Transfer-Based Approach for Unsupervised Thermal Image Super-Resolution |
title_full_unstemmed | A Novel Domain Transfer-Based Approach for Unsupervised Thermal Image Super-Resolution |
title_short | A Novel Domain Transfer-Based Approach for Unsupervised Thermal Image Super-Resolution |
title_sort | novel domain transfer based approach for unsupervised thermal image super resolution |
url | https://hdl.handle.net/1721.1/141368 |
work_keys_str_mv | AT rivadeneirarafaele anoveldomaintransferbasedapproachforunsupervisedthermalimagesuperresolution AT sappaangeld anoveldomaintransferbasedapproachforunsupervisedthermalimagesuperresolution AT vintimillaborisx anoveldomaintransferbasedapproachforunsupervisedthermalimagesuperresolution AT hammoudriad anoveldomaintransferbasedapproachforunsupervisedthermalimagesuperresolution AT rivadeneirarafaele noveldomaintransferbasedapproachforunsupervisedthermalimagesuperresolution AT sappaangeld noveldomaintransferbasedapproachforunsupervisedthermalimagesuperresolution AT vintimillaborisx noveldomaintransferbasedapproachforunsupervisedthermalimagesuperresolution AT hammoudriad noveldomaintransferbasedapproachforunsupervisedthermalimagesuperresolution |