Beyond Aesthetics: Quantifying Information Loss in Turbid Scenes
Quick Answer
The study introduces the Turbid Underwater Baseline (TUB) dataset with 1,320 images and over 16,000 segmentation masks to quantify information loss in turbid underwater scenes.
Quick Take
The study introduces the Turbid Underwater Baseline (TUB) dataset with 1,320 images and over 16,000 segmentation masks to quantify information loss in turbid underwater scenes. A new metric, PCD, shows a strong correlation with instance segmentation model performance, outperforming traditional metrics in assessing real-world turbidity effects.
Key Points
- TUB dataset includes 1,320 images under extreme turbidity conditions.
- Over 16,000 high-confidence ground-truth segmentation masks are provided.
- PCD metric effectively captures structural information loss in turbid scenes.
- PCD correlates strongly with instance segmentation performance on turbid images.
- Common metrics show weak correlation with model performance in this context.
Paper Resources
📖 Reader Mode
~2 min readAbstract:Visibility in underwater environments degrades rapidly under turbid conditions, yet the effects on computer-vision models remain unclear. This issue is compounded by reliance on synthetic turbidity datasets, which may misrepresent real-world information loss. To address this gap, we introduce the Turbid Underwater Baseline (TUB) dataset, comprising 1,320 images captured under extreme turbidity and over 16,000 high-confidence ground-truth segmentation masks. We additionally propose PCD, a metric derived from phase congruency maps that is invariant to contrast and aims to capture the loss of structural information in real turbidity. We show that PCD correlates strongly with the performance of instance segmentation models on both real and synthetic turbid images, whereas common metrics in the field show weak to no correlation at all. The dataset and relevant code can be found on the project page: this https URL
| Subjects: | Computer Vision and Pattern Recognition (cs.CV) |
| Cite as: | arXiv:2606.26295 [cs.CV] |
| (or arXiv:2606.26295v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2606.26295 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Vasiliki Ismiroglou [view email]
[v1]
Wed, 24 Jun 2026 18:40:45 UTC (12,425 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.CV
See more →LLM-Guided ANN Index Optimization for Human-Object Interaction Retrieval
A phase-aware LLM agent optimizes human-object interaction retrieval, outperforming Optuna TPE by 33.3% and VDTuner by 34.2% on the HICO-DET benchmark. This method enhances throughput by 15.3x over UniIR and demonstrates strong transferability across vector database management systems.