Pith. sign in

REVIEW 20 cited by

Q-Bench: A Benchmark for General-Purpose Foundation Models on Low-level Vision

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2309.14181 v3 pith:ZTBPJH5G submitted 2023-09-25 cs.CV cs.AIcs.MM

Q-Bench: A Benchmark for General-Purpose Foundation Models on Low-level Vision

classification cs.CV cs.AIcs.MM
keywords mllmslow-levelvisualabilitiesmodelsqualityabilityassessment
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

The rapid evolution of Multi-modality Large Language Models (MLLMs) has catalyzed a shift in computer vision from specialized models to general-purpose foundation models. Nevertheless, there is still an inadequacy in assessing the abilities of MLLMs on low-level visual perception and understanding. To address this gap, we present Q-Bench, a holistic benchmark crafted to systematically evaluate potential abilities of MLLMs on three realms: low-level visual perception, low-level visual description, and overall visual quality assessment. a) To evaluate the low-level perception ability, we construct the LLVisionQA dataset, consisting of 2,990 diverse-sourced images, each equipped with a human-asked question focusing on its low-level attributes. We then measure the correctness of MLLMs on answering these questions. b) To examine the description ability of MLLMs on low-level information, we propose the LLDescribe dataset consisting of long expert-labelled golden low-level text descriptions on 499 images, and a GPT-involved comparison pipeline between outputs of MLLMs and the golden descriptions. c) Besides these two tasks, we further measure their visual quality assessment ability to align with human opinion scores. Specifically, we design a softmax-based strategy that enables MLLMs to predict quantifiable quality scores, and evaluate them on various existing image quality assessment (IQA) datasets. Our evaluation across the three abilities confirms that MLLMs possess preliminary low-level visual skills. However, these skills are still unstable and relatively imprecise, indicating the need for specific enhancements on MLLMs towards these abilities. We hope that our benchmark can encourage the research community to delve deeper to discover and enhance these untapped potentials of MLLMs. Project Page: https://q-future.github.io/Q-Bench.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 20 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. SenseBench: A Benchmark for Remote Sensing Low-Level Visual Perception and Description in Large Vision-Language Models

    cs.CV 2026-05 unverdicted novelty 8.0

    SenseBench is the first physics-based benchmark with 10K+ instances and dual protocols to evaluate VLMs on remote sensing low-level perception and diagnostic description, revealing domain bias and specific failure modes.

  2. PointQ-Bench: Benchmarking Diagnostic and Interpretable Point Cloud Quality Assessment

    cs.CV 2026-05 unverdicted novelty 7.0

    PointQ-Bench is a benchmark with annotated point clouds supporting anomaly sensing, defect diagnosis, usability grading, and open-ended quality reporting, plus the SSFRQ-5D evaluation protocol.

  3. IQA-Spider: Unifying Multi-Granularity Image Quality Assessment with Reasoning, Grounding and Referring

    cs.CV 2026-05 unverdicted novelty 7.0

    IQA-Spider unifies reasoning, grounding, and referring for multi-granularity image quality assessment via a four-task paradigm and two-stage LMM training with training-free text-to-point mapping.

  4. Tunable Polariton Canalization in Natural van der Waals Oxide

    physics.optics 2026-04 unverdicted novelty 7.0

    Natural alpha-V2O5 exhibits continuously tunable polariton canalization with unidirectional Poynting vector propagation controlled by incident infrared frequency.

  5. LLaVA-NeXT-Interleave: Tackling Multi-image, Video, and 3D in Large Multimodal Models

    cs.CV 2024-07 unverdicted novelty 7.0

    LLaVA-NeXT-Interleave unifies multi-image, video, and 3D capabilities in large multimodal models via a new 1.18M-sample interleaved dataset and benchmark, achieving leading results across those tasks while preserving ...

  6. SVGEval: A Vision-Grounded Framework for Perceptual-Quality Benchmarking and Evaluation in Text-to-SVG Generation

    cs.CV 2026-08 conditional novelty 6.0

    SVGEval benchmarks and trains an explainable multimodal scorer for perceptual quality of text-to-SVG generation, showing a consistent gap on spatial and structural judgments.

  7. Latent Denoising Improves Visual Alignment in Large Multimodal Models

    cs.CV 2026-04 unverdicted novelty 6.0

    A latent denoising objective with saliency-aware corruption and contrastive distillation improves visual alignment and corruption robustness in large multimodal models.

  8. Redefining Quality Criteria and Distance-Aware Score Modeling for Image Editing Assessment

    cs.CV 2026-04 unverdicted novelty 6.0

    DS-IEQA jointly learns evaluation criteria via feedback-driven prompt optimization and continuous score modeling via token-decoupled distance regression, ranking 4th in the 2026 NTIRE X-AIGC Quality Assessment Track 2...

  9. Are We on the Right Way for Evaluating Large Vision-Language Models?

    cs.CV 2024-03 conditional novelty 6.0

    Current LVLM benchmarks overestimate capabilities because many questions can be answered without images due to design flaws or data leakage; MMStar is a human-curated set of 1,500 vision-indispensable samples across 6...

  10. ShareGPT4V: Improving Large Multi-Modal Models with Better Captions

    cs.CV 2023-11 conditional novelty 6.0

    A new 1.2M-caption dataset generated via GPT-4V improves LMMs on MME and MMBench by 222.8/22.0/22.3 and 2.7/1.3/1.5 points respectively when used for supervised fine-tuning.

  11. Tunable Polariton Canalization in Natural van der Waals Oxide

    physics.optics 2026-04 unverdicted novelty 5.0

    Untreated alpha-V2O5 exhibits frequency-tunable in-plane polariton canalization with unidirectional Poynting-vector flow, mapped by infrared nano-imaging and a permittivity phase diagram.

  12. Parameter-Efficient Adaptation of mPLUG-Owl2 via Pixel-Level Visual Prompts for NR-IQA

    cs.CV 2025-09 conditional novelty 5.0

    With a learned 30-pixel border prompt added to input images, a frozen mPLUG-Owl2-7B reaches 0.932 SRCC on KADID-10k using about 156K trainable parameters.

  13. Q-Agent: Quality-Driven Chain-of-Thought Image Restoration Agent through Robust Multimodal Large Language Model

    eess.IV 2025-04 unverdicted novelty 5.0

    Q-Agent uses CoT decomposition on a fine-tuned MLLM for multi-degradation perception plus IQA-driven greedy selection of restoration algorithms to claim better performance than All-in-One IR models.

  14. mPLUG-Owl3: Towards Long Image-Sequence Understanding in Multi-Modal Large Language Models

    cs.CV 2024-08 unverdicted novelty 5.0

    mPLUG-Owl3 introduces hyper attention blocks to integrate vision and language for long image-sequence understanding and reports SOTA results on single-image, multi-image, and video benchmarks.

  15. LLaVA-OneVision: Easy Visual Task Transfer

    cs.CV 2024-08 unverdicted novelty 5.0

    LLaVA-OneVision is the first single open LMM to simultaneously achieve strong performance in single-image, multi-image, and video scenarios with cross-scenario transfer capabilities.

  16. InternLM-XComposer2: Mastering Free-form Text-Image Composition and Comprehension in Vision-Language Large Model

    cs.CV 2024-01 unverdicted novelty 5.0

    InternLM-XComposer2 introduces Partial LoRA on InternLM2-7B to enable high-quality free-form text-image composition while matching or exceeding GPT-4V on select vision-language benchmarks.

  17. mPLUG-Owl2: Revolutionizing Multi-modal Large Language Model with Modality Collaboration

    cs.CL 2023-11 unverdicted novelty 5.0

    mPLUG-Owl2 presents a modular MLLM architecture that enables modality collaboration via shared functional modules and modality-adaptive components, achieving SOTA on both text and multi-modal tasks with one generic model.

  18. FakeVLM-R1: Internalizing Physical Laws via CoT for Synthetic Image Detection

    cs.CV 2026-05 unverdicted novelty 4.0

    FakeVLM-R1 combines GRPO reinforcement learning with critical-thinking CoT and a physics-annotated FakeClue++ dataset to reach claimed SOTA synthetic image detection while reducing over-rejection of real images.

  19. iDiff: Interpretable Difference-aware Framework for Pairwise Image Quality Assessment

    cs.CV 2026-05 unverdicted novelty 4.0

    iDiff is a dual-branch framework with an Answer Model for robust pairwise preference prediction via view decomposition and ensembles, and a Thinking Model for structured rationale generation using templates and answer...

  20. InternLM-XComposer: A Vision-Language Large Model for Advanced Text-image Comprehension and Composition

    cs.CV 2023-09 conditional novelty 4.0

    InternLM-XComposer generates articles with seamlessly integrated images and achieves state-of-the-art results on vision-language benchmarks including MME, MMBench, and Seed-Bench.