Research map / Vision-language and multimodal / Multimodal LLMs
UQ-Bench: A Benchmark for Evaluating Multimodal LLMs on Underwater Image Quality Assessment
Jingchao Cao, Guo An, Feng Gao, Ke Gu, Yutao Liu
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Introduces UQ-Bench, a benchmark for evaluating multimodal LLMs on low-level perception and quality assessment of underwater images.
Problem
The low-level visual perception ability of multimodal LLMs in underwater environments is underexplored, with no systematic benchmark for underwater image quality assessment at the low-level attribute level.
Approach
A systematically designed benchmark that tests MLLMs on perceiving and assessing underwater image quality at the low-level visual attribute level, the first of its kind for this domain.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| compare closed-source GPT-4o with open-source MLLMs on QA | MLLMs · UW-Perception QA overall | accuracy (%): GPT-4o 79.50% vs best open-source VisualGLM-6B 76.07% | Improved |
| switch from enhanced (UID) to real-world degraded (UIQD) images for quality scoring | mPLUG-Owl · UID vs UIQD | SRCC: 0.2162 -> 0.8579 | Improved |
| switch from UID to UIQD | Qwen-VL · UID vs UIQD | SRCC: -0.1067 -> 0.0068 | No clear effect |
| GPT-4o vs open-source models on description | MLLMs · UW-Describe | sum score: 4.56 vs 2.72-3.41 (open-source) | Improved |
Experimental setup
Benchmarks: UQ-Bench
Closest papers by meaning
- Q-Bench: A Benchmark for General-Purpose Foundation Models on Low-level Vision (ICLR 2024 · similarity 0.88)
- NAUTILUS: A Large Multimodal Model for Underwater Scene Understanding (NeurIPS 2025 · similarity 0.88)
- UVEB: A Large-scale Benchmark and Baseline Towards Real-World Underwater Video Enhancement (CVPR 2024 · similarity 0.87)
- WebUOT-1M: Advancing Deep Underwater Object Tracking with A Million-Scale Benchmark (NeurIPS 2024 · similarity 0.86)
- Diving into Underwater: Segment Anything Model Guided Underwater Salient Instance Segmentation and A Large-scale Dataset (ICML 2024 · similarity 0.86)
- LL-Bench: Rethinking Low-Level Vision Evaluation in the Era of Large-Scale Generative Models (NeurIPS 2026 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
