Bayesian Neural Networks for One-to-Many Mapping in Image Enhancement

Guoxi Huang; Qirui Yang; RuiRui Lin; Zipeng Qi; David Bull; Nantheera Anantrasirichai

doi:10.48550/arXiv.2501.14265

Bayesian Neural Networks for One-to-Many Mapping in Image Enhancement

Guoxi Huang^*, Qirui Yang, RuiRui Lin, Zipeng Qi, David Bull, Nantheera Anantrasirichai

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Conference Contribution (Conference Proceeding)

2 Downloads (Pure)

Abstract

In image enhancement tasks, such as low-light and underwater image enhancement, a degraded image can correspond to multiple plausible target images due to dynamic photography conditions. This naturally results in a one-to-many mapping problem. To address this, we propose a Bayesian Enhancement Model (BEM) that incorporates Bayesian Neural Networks (BNNs) to capture data uncertainty and produce diverse outputs. To enable fast inference, we introduce a BNN-DNN framework: a BNN is first employed to model the one-to-many mapping in a low-dimensional space, followed by a Deterministic Neural Network (DNN) that refines fine-grained image details. Extensive experiments on multiple low-light and underwater image enhancement benchmarks demonstrate the effectiveness of our method.

Original language	English
Title of host publication	Proceedings of the 40th Annual AAAI Conference on Artificial Intelligence (AAAI-26)
Publisher	AAAI Press
Number of pages	9
ISBN (Electronic)	978-1-57735-906-7, 1-57735-906-2
DOIs	https://doi.org/10.48550/arXiv.2501.14265 https://doi.org/10.1609/aaai.v40i7.37413
Publication status	Accepted/In press - 8 Nov 2025
Event	The 40th Annual AAAI Conference on Artificial Intelligence - , Singapore Duration: 20 Jan 2026 → 27 Jan 2026 https://aaai.org/conference/aaai/aaai-26/

Publication series

Name	Proceedings of the AAAI Conference on Artificial Intelligence
Publisher	AAAI Press
Number	7
Volume	40
ISSN (Print)	2159-5399
ISSN (Electronic)	2374-3468

Conference

Conference	The 40th Annual AAAI Conference on Artificial Intelligence
Country/Territory	Singapore
Period	20/01/26 → 27/01/26
Internet address	https://aaai.org/conference/aaai/aaai-26/

Bibliographical note

Accepted to The 40th Annual AAAI Conference on Artificial Intelligence, 2026

Keywords

cs.CV

Access to Document

Full-text PDF (accepted manuscript)
This is the accepted author manuscript (AAM) of the article which has been made Open Access under the University of Bristol's Scholarly Works Policy. The final published version (Version of Record) can be found on the publisher's website. The copyright of any third-party content, such as images, remains with the copyright holder.
Accepted author manuscript, 6.28 MBLicence: CC BY

Handle.net

Persistent link

1 Finished
1 Active

MyWorld: Intelligent Post-Production for Challenging Data Acquisition
Anantrasirichai, P. (Principal Investigator)
1/05/21 → 31/03/27
Project: Research
MyUnderwaterWorld: Intelligent Underwater Scene Representation
Anantrasirichai, P. (Principal Investigator)
1/02/24 → 31/03/26
Project: Research

Cite this

Huang, G., Yang, Q., Lin, R., Qi, Z., Bull, D., & Anantrasirichai, N. (Accepted/In press). Bayesian Neural Networks for One-to-Many Mapping in Image Enhancement. In Proceedings of the 40th Annual AAAI Conference on Artificial Intelligence (AAAI-26) (Proceedings of the AAAI Conference on Artificial Intelligence; Vol. 40, No. 7). AAAI Press. https://doi.org/10.48550/arXiv.2501.14265, https://doi.org/10.1609/aaai.v40i7.37413

@inproceedings{9d77de5f61134b0ba803a66f65d00455,

title = "Bayesian Neural Networks for One-to-Many Mapping in Image Enhancement",

abstract = "In image enhancement tasks, such as low-light and underwater image enhancement, a degraded image can correspond to multiple plausible target images due to dynamic photography conditions. This naturally results in a one-to-many mapping problem. To address this, we propose a Bayesian Enhancement Model (BEM) that incorporates Bayesian Neural Networks (BNNs) to capture data uncertainty and produce diverse outputs. To enable fast inference, we introduce a BNN-DNN framework: a BNN is first employed to model the one-to-many mapping in a low-dimensional space, followed by a Deterministic Neural Network (DNN) that refines fine-grained image details. Extensive experiments on multiple low-light and underwater image enhancement benchmarks demonstrate the effectiveness of our method.",

keywords = "cs.CV",

author = "Guoxi Huang and Qirui Yang and RuiRui Lin and Zipeng Qi and David Bull and Nantheera Anantrasirichai",

note = "Accepted to The 40th Annual AAAI Conference on Artificial Intelligence, 2026; The 40th Annual AAAI Conference on Artificial Intelligence ; Conference date: 20-01-2026 Through 27-01-2026",

year = "2025",

month = nov,

day = "8",

doi = "10.48550/arXiv.2501.14265",

language = "English",

series = "Proceedings of the AAAI Conference on Artificial Intelligence",

publisher = "AAAI Press",

number = "7",

booktitle = "Proceedings of the 40th Annual AAAI Conference on Artificial Intelligence (AAAI-26)",

address = "United States",

url = "https://aaai.org/conference/aaai/aaai-26/",

}

Huang, G, Yang, Q, Lin, R, Qi, Z, Bull, D & Anantrasirichai, N 2025, Bayesian Neural Networks for One-to-Many Mapping in Image Enhancement. in Proceedings of the 40th Annual AAAI Conference on Artificial Intelligence (AAAI-26). Proceedings of the AAAI Conference on Artificial Intelligence, no. 7, vol. 40, AAAI Press, The 40th Annual AAAI Conference on Artificial Intelligence, Singapore, 20/01/26. https://doi.org/10.48550/arXiv.2501.14265, https://doi.org/10.1609/aaai.v40i7.37413

Bayesian Neural Networks for One-to-Many Mapping in Image Enhancement. / Huang, Guoxi; Yang, Qirui; Lin, RuiRui et al.
Proceedings of the 40th Annual AAAI Conference on Artificial Intelligence (AAAI-26). AAAI Press, 2025. (Proceedings of the AAAI Conference on Artificial Intelligence; Vol. 40, No. 7).

Research output: Chapter in Book/Report/Conference proceeding › Conference Contribution (Conference Proceeding)

TY - GEN

T1 - Bayesian Neural Networks for One-to-Many Mapping in Image Enhancement

AU - Huang, Guoxi

AU - Yang, Qirui

AU - Lin, RuiRui

AU - Qi, Zipeng

AU - Bull, David

AU - Anantrasirichai, Nantheera

N1 - Accepted to The 40th Annual AAAI Conference on Artificial Intelligence, 2026

PY - 2025/11/8

Y1 - 2025/11/8

N2 - In image enhancement tasks, such as low-light and underwater image enhancement, a degraded image can correspond to multiple plausible target images due to dynamic photography conditions. This naturally results in a one-to-many mapping problem. To address this, we propose a Bayesian Enhancement Model (BEM) that incorporates Bayesian Neural Networks (BNNs) to capture data uncertainty and produce diverse outputs. To enable fast inference, we introduce a BNN-DNN framework: a BNN is first employed to model the one-to-many mapping in a low-dimensional space, followed by a Deterministic Neural Network (DNN) that refines fine-grained image details. Extensive experiments on multiple low-light and underwater image enhancement benchmarks demonstrate the effectiveness of our method.

AB - In image enhancement tasks, such as low-light and underwater image enhancement, a degraded image can correspond to multiple plausible target images due to dynamic photography conditions. This naturally results in a one-to-many mapping problem. To address this, we propose a Bayesian Enhancement Model (BEM) that incorporates Bayesian Neural Networks (BNNs) to capture data uncertainty and produce diverse outputs. To enable fast inference, we introduce a BNN-DNN framework: a BNN is first employed to model the one-to-many mapping in a low-dimensional space, followed by a Deterministic Neural Network (DNN) that refines fine-grained image details. Extensive experiments on multiple low-light and underwater image enhancement benchmarks demonstrate the effectiveness of our method.

KW - cs.CV

U2 - 10.48550/arXiv.2501.14265

DO - 10.48550/arXiv.2501.14265

M3 - Conference Contribution (Conference Proceeding)

T3 - Proceedings of the AAAI Conference on Artificial Intelligence

BT - Proceedings of the 40th Annual AAAI Conference on Artificial Intelligence (AAAI-26)

PB - AAAI Press

T2 - The 40th Annual AAAI Conference on Artificial Intelligence

Y2 - 20 January 2026 through 27 January 2026

ER -

Bayesian Neural Networks for One-to-Many Mapping in Image Enhancement

Abstract

Publication series

Conference

Bibliographical note

Keywords

Access to Document

Handle.net

Fingerprint

Projects

MyWorld: Intelligent Post-Production for Challenging Data Acquisition

MyUnderwaterWorld: Intelligent Underwater Scene Representation

Cite this