Saved in:
Bibliographic Details
Main Authors: Mentzelopoulos, Andreas, Ellenbogen, Keith
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2602.05163
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908814139195392
author Mentzelopoulos, Andreas
Ellenbogen, Keith
author_facet Mentzelopoulos, Andreas
Ellenbogen, Keith
contents Underwater photography presents significant inherent challenges including reduced contrast, spatial blur, and wavelength-dependent color distortions. These effects can obscure the vibrancy of marine life and awareness photographers in particular are often challenged with heavy post-processing pipelines to correct for these distortions. We develop an image-to-image pipeline that learns to reverse underwater degradations by introducing a synthetic corruption pipeline and learning to reverse its effects with diffusion-based generation. Training and evaluation are performed on a small high-quality dataset of awareness photography images by Keith Ellenbogen. The proposed methodology achieves high perceptual consistency and strong generalization in synthesizing 512x768 images using a model of ~11M parameters after training from scratch on ~2.5k images.
format Preprint
id arxiv_https___arxiv_org_abs_2602_05163
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle LOBSTgER-enhance: an underwater image enhancement pipeline
Mentzelopoulos, Andreas
Ellenbogen, Keith
Computer Vision and Pattern Recognition
Underwater photography presents significant inherent challenges including reduced contrast, spatial blur, and wavelength-dependent color distortions. These effects can obscure the vibrancy of marine life and awareness photographers in particular are often challenged with heavy post-processing pipelines to correct for these distortions. We develop an image-to-image pipeline that learns to reverse underwater degradations by introducing a synthetic corruption pipeline and learning to reverse its effects with diffusion-based generation. Training and evaluation are performed on a small high-quality dataset of awareness photography images by Keith Ellenbogen. The proposed methodology achieves high perceptual consistency and strong generalization in synthesizing 512x768 images using a model of ~11M parameters after training from scratch on ~2.5k images.
title LOBSTgER-enhance: an underwater image enhancement pipeline
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2602.05163