Pith. sign in

REVIEW 4 major objections 6 minor 35 references

A frozen 7B vision-language model can be adapted to score image quality by learning a 155,880-pixel border prompt.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

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.

T0 review reviewed 2026-08-05 challenge →

load-bearing objection A credible, cheap visual-prompting recipe for NR-IQA, but the headline KADID number is built on a test-set selection and possibly content-overlapping split. the 4 major comments →

arxiv 2509.03494 v2 pith:HSKLPDZ5 submitted 2025-09-03 cs.CV

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

Yahya Benmahane , Mohammed El Hassouni This is my paper
classification cs.CV
keywords no-reference image quality assessmentvisual promptingpixel-space promptsmultimodal large language modelsparameter-efficient adaptationmPLUG-Owl2softmax quality scoringKADID-10k
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper seeks to establish that a frozen multimodal large language model can be turned into a no-reference image-quality assessor by learning only a small pattern of pixels added to the input image, leaving every model weight untouched. On the KADID-10k synthetic-distortion benchmark, that pixel prompt (a 30-pixel border) lifts Spearman correlation from 0.550 for the unadapted model to 0.932, matching or exceeding fully fine-tuned 7B models such as LIQE and Q-Align while training around 155,880 parameters, less than 0.01 percent of the base model. The same recipe works, though less strongly, on realistic (KonIQ-10k) and AI-generated (AGIQA-3k) distortions. If correct, the result means expensive fine-tuning and per-task model copies are not necessary: one frozen base model can be redirected to a new low-level vision task by swapping a tiny additive visual prompt. The authors claim this is the first pixel-level visual-prompt adaptation of an MLLM for image quality assessment.

Core claim

The central claim is that NR-IQA can be posed as a regression whose only trainable parameters are pixels. The method feeds the frozen mPLUG-Owl2-7B a fixed textual query, "Rate the technical quality of the image," alongside the input image plus a learned visual prompt; the model's final-token logits for the two positive words {good, fine} and two negative words {poor, bad} are combined into a softmax score in (0,1), trained by mean-squared error against human MOS. The best configuration is a 30-pixel padding around all four sides, with 155,880 parameters, reaching SRCC 0.932 and PLCC 0.929 on KADID-10k. Ten-pixel or small fixed patches fail, while a full overlay is competitive on KADID but w

What carries the argument

The load-bearing object is the additive visual prompt: a trainable tensor of the same shape as the input image (a border, a fixed patch, or a full overlay), bounded by tanh and clamped to [0,1], added to the image before the MLLM's vision encoder. Because the model is frozen, the prompt is the only channel through which training signals can act. The score extraction is equally important: no regression head is learned; the quality score is a softmax over four vocabulary logits (good, fine versus poor, bad) from the model's final token, so the learned pixels must reorganize the frozen model's internal response to that vocabulary. The 30-pixel padding works best, which the paper ties to prompt

Load-bearing premise

The frozen model's internal logits for the words good, fine, poor, and bad must respond monotonically to image quality, and a single fixed pixel pattern must be able to push that response into the right order on all test images.

What would settle it

Train the 30-pixel padding prompt on KADID-10k, freeze it, and apply it to KonIQ-10k: if SRCC drops to near the unfrozen baseline of about 0.60, the prompt encodes KADID-specific shortcuts rather than a general quality axis.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • One frozen mPLUG-Owl2 can be reused across datasets and tasks by swapping a small stored pixel prompt, instead of storing a separate 7B fine-tuned model per dataset.
  • Visual prompt geometry is decisive: a 30-pixel border yields 0.932 SRCC on KADID-10k while a 10-pixel patch yields 0.390, so prompt size and placement are first-class design choices.
  • Because neither the vision encoder nor the language model is updated, adaptation avoids catastrophic forgetting and lowers memory requirements dramatically relative to full fine-tuning.
  • The method still trails fully fine-tuned methods on KonIQ-10k and AGIQA-3k, indicating pixel-only adaptation is not yet a complete substitute for weight updates on realistic and AI-generated distortions.
  • The softmax-over-quality-words scoring trick from Q-Bench is sufficient when combined with pixel prompting to align model outputs with MOS, without architectural changes.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The learned border probably acts as a global attention bias rather than encoding local distortion features; if so, its transfer across datasets should be limited, and ablating quadrants of the border should show where the signal lives.
  • Cross-dataset transfer is untested in the paper. A natural test is to train the padding on KADID-10k and evaluate on KonIQ-10k; because the two distributions differ sharply, the prompt may collapse toward the frozen baseline.
  • Combining a learned pixel prompt with a learned textual prompt is the obvious next step; the paper itself notes the gap to CLIPIQA+ on KonIQ, so co-tuning both modalities could close it.
  • The four-word vocabulary makes the score a coarse ordinal; expanding the positive and negative token sets or using per-level labels could change results and may be needed for a higher dynamic range.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. The paper proposes a parameter-efficient NR-IQA method that learns a pixel-space visual prompt (padding, fixed patch, or full overlay) added to the input image of a frozen mPLUG-Owl2-7B MLLM. A quality score is obtained by a softmax over positive ("good", "fine") and negative ("poor", "bad") token logits, and the prompt is trained with MSE regression against MOS. The authors report that the best configuration, a 30px padding with 155,880 trainable parameters, achieves SRCC 0.932 on KADID-10k, and they claim competitive performance with fully fine-tuned MLLMs and specialized NR-IQA models on KADID-10k, KonIQ-10k, and AGIQA-3k, while training at most ~600K parameters (<0.01% of the base model). The paper also claims to be the first to apply pixel-space visual prompting to NR-IQA.

Significance. If the reported results hold under a sound evaluation protocol, the contribution is significant: it demonstrates that a frozen 7B MLLM can be adapted to NR-IQA with a tiny additive pixel pattern, which would offer a highly parameter-efficient alternative to full fine-tuning and avoid model forgetting. The paper's strengths include a simple and reproducible pipeline, public code, standard datasets, explicit parameter counts, and a clear comparison with a range of baselines. The novelty of pixel-level visual prompting for NR-IQA is genuine. However, the central quantitative claim depends on the evaluation protocol, and the manuscript currently does not establish that protocol rigorously: the KADID-10k split is not shown to be content-disjoint, the prompt configuration is selected after seeing test-set results, and the reported numbers are single-run with at least one internal inconsistency.

major comments (4)
  1. [§4.1 (Experimental setup)] The KADID-10k evaluation protocol is underspecified and potentially invalid. KADID-10k contains 10,125 distorted images derived from only 81 reference images. The paper states "we select 80% of the dataset for training, 10% for validation, and the remaining 10% for testing" but does not state that this split is reference-content-disjoint. A random image-level split will place distortions of the same content in both train and test, allowing the learned global additive prompt to exploit content identity rather than learn generalizable quality assessment. The comparison baselines (e.g., LIQE, Q-Align) are typically evaluated with content-disjoint splits, so the reported 0.932 SRCC may not be comparable. Please clarify the split and, if it is not content-disjoint, re-run the experiments accordingly.
  2. [§4.2 / Table 2] The headline result appears to be selected post hoc from test-set performance. Table 2 reports SRCC/PLCC for seven prompt configurations (padding 10/30px, center patch 10/30px, top-left patch 10/30px, full overlay) on the same test sets. The text then says "Based on these results, we notice that the 30px padding is the most correlated type of visual prompt with the MOS across all datasets." This is a test-set maximum; no validation-based model selection is described. The 0.932 SRCC is therefore the best of seven configurations chosen after seeing test labels, which inflates the expected performance of the proposed method. Please use the validation split for selection, or report the full selection rule and adjust the significance claims.
  3. [Table 2 vs Table 3] There is an internal inconsistency in the reported results for KonIQ-10k. Table 2 reports PLCC=0.874 for the 30px padding prompt, while Table 3 reports PLCC=0.865 for "Our Proposed Method", which presumably is the same configuration. If they are the same, the numbers must match; if they differ, the caption or text should explain why. This discrepancy, combined with single-run reporting, makes it difficult to verify that the paper's comparison is based on a single pre-specified protocol. Please clarify and, ideally, report results over multiple seeds (mean±std).
  4. [§4.2 (Implementation details) / §3.1] The paper does not provide any analysis of what the learned visual prompt encodes or whether the underlying assumption is valid: that a single global additive pattern can steer the frozen model's softmax over the four chosen quality words into a monotonically quality-responsive score. Since neither the vision encoder nor the language model is updated, if the base model's logits for {good, fine} vs {poor, bad} are not responsive to quality for the test distribution, the prompt cannot compensate. Please include a diagnostic (e.g., prompt visualization, logit analysis, or cross-dataset transfer experiments) to support the mechanism.
minor comments (6)
  1. [§3.2] The choice of mPLUG-Owl2-7B is justified by API compatibility; please provide the exact model version/checkpoint and preprocessing details (resize and center-crop size) for reproducibility.
  2. [§3.4] The positive/negative token sets are fixed to "good", "fine" vs "poor", "bad". Please clarify whether these are the only tokens considered and whether results are sensitive to this choice; a brief ablation would strengthen the paper.
  3. [§4.2] Hyperparameters are unusual (e.g., learning rate 60 for SGD). Please specify momentum, weight decay, learning-rate schedule, and whether the learning rate is per-example or batch-scaled.
  4. [Figure 2] The pixelated regions in Figure 2 are hard to interpret. A clearer indication of the prompt region (e.g., a highlight or zoomed inset) would improve readability.
  5. [References] Reference formatting is inconsistent: some entries lack venues or page numbers (e.g., [1], [18], [34]), and [34] is a bare URL. Please standardize.
  6. [General] Typographical issue: page 2 has the header "APREPRINT- SEPTEMBER12, 2025" with a missing space; please fix.

Circularity Check

0 steps flagged

No circular derivation; the central result is a held-out empirical measurement, and protocol caveats are not circularity.

full rationale

The paper's chain is: (i) define a scalar quality score s_i as the softmax ratio of frozen-MLLM logits over fixed positive/negative token sets ('good','fine' vs 'poor','bad'); (ii) minimize MSE between s_i and normalized ground-truth MOS over a training split to learn only the pixel prompt p; (iii) evaluate SRCC/PLCC on a held-out test split. No equation is defined in terms of the target metric, and the test metrics are not used as a training loss. There are no load-bearing self-citations: the references to mPLUG-Owl2, Q-Bench, and prior visual-prompting work are external baselines/starting points, not uniqueness theorems or fitted values. The manuscript does contain evaluation-protocol concerns that could affect the validity of the headline number: §4.1 does not state that the 80/10/10 KADID split is content-disjoint, and Table 2 reports all seven prompt configurations on the test set with the 30px padding selected as best, making the 'Our Proposed Method' row a test-set maximum. These are selection and leakage issues, not cases where a prediction is equivalent to an input by construction or where a fitted parameter is renamed a prediction, so they do not raise the circularity score under the provided rubric.

Axiom & Free-Parameter Ledger

6 free parameters · 6 axioms · 0 invented entities

The central claim rests on the frozen model's logit geometry being steerable by an additive pixel pattern (a domain assumption), the choice of the four quality words, the hand-picked learning schedule, and the post-hoc selection of the 30px padding shape. No new physical entities are introduced.

free parameters (6)
  • Visual prompt tensor p (padding 30px variant) = not reported (learned on training split; 155,880 parameters for 30px padding)
    The central trainable object; x' = x + p. Its converged values are never shown or analyzed.
  • Visual prompt geometry (type and size) = Padding, 30px
    Selected as the best performer from Table 2 on the same benchmark, i.e., post hoc model selection.
  • Learning rate = 60 (extended run uses 20)
    Hand-chosen; unusually high for SGD on a frozen backbone; no sensitivity analysis.
  • Batch size = 32 (KADID-10k), 4 (KonIQ-10k, AGIQA-3k)
    Hand-chosen; likely memory-limited; no sensitivity analysis.
  • Training schedule = 25 epochs on KADID-10k, +25 for padding, +10 for AGIQA-3k
    Hand-chosen; the '+10 epochs' phrasing is ambiguous.
  • Quality token sets P and N = P = {good, fine}, N = {poor, bad}
    The softmax score function depends entirely on these token IDs; no ablation of other quality words.
axioms (6)
  • domain assumption Frozen mPLUG-Owl2-7B's final-token logits for quality-related words respond monotonically to image quality after an additive pixel perturbation
    The whole pipeline assumes a single global additive pattern can steer the model's quality judgment; supported only by the observed baseline correlation (>0.5 on two datasets) and the training experiment.
  • domain assumption Softmax ratio over the chosen positive and negative token sets is a valid monotone transform of perceptual quality
    Section 3.1 defines the score this way, following Q-Bench's empirical finding; no analysis of token choice.
  • domain assumption Textual prompt 'Rate the technical quality of the image.' is appropriate and the word 'technical' improves it
    Section 3.4 asserts this without an ablation; the prompt is fixed from Q-Bench.
  • ad hoc to paper Normalization after adding the visual prompt is required for good performance
    Section 4.2: 'we experimentally verified to yield superior performance' but no details are given.
  • domain assumption The public MOS labels on the train splits are accurate ground truth and the dataset splits are unbiased
    Standard supervised assumption; the datasets are widely used.
  • standard math Standard backpropagation through the frozen model yields a nontrivial gradient for the prompt
    The prompt is optimized by MSE; standard gradient-based learning on a frozen backbone.

reviewed 2026-08-05 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Parameter-Efficient Adaptation of mPLUG-Owl2 via Pixel-Level Visual Prompts for NR-IQA." pith.science (2026). https://pith.science/paper/HSKLPDZ5

@misc{pith2026250903494,
  author       = {Pith},
  title        = {Pith review of: Parameter-Efficient Adaptation of mPLUG-Owl2 via Pixel-Level Visual Prompts for NR-IQA},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HSKLPDZ5}},
  note         = {Machine review of arXiv:2509.03494}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

In this paper, we propose a novel parameter-efficient adaptation method for No- Reference Image Quality Assessment (NR-IQA) using visual prompts optimized in pixel-space. Unlike full fine-tuning of Multimodal Large Language Models (MLLMs), our approach trains only 600K parameters at most (< 0.01% of the base model), while keeping the underlying model fully frozen. During inference, these visual prompts are combined with images via addition and processed by mPLUG-Owl2 with the textual query "Rate the technical quality of the image." Evaluations across distortion types (synthetic, realistic, AI-generated) on KADID- 10k, KonIQ-10k, and AGIQA-3k demonstrate competitive performance against full finetuned methods and specialized NR-IQA models, achieving 0.93 SRCC on KADID-10k. To our knowledge, this is the first work to leverage pixel-space visual prompts for NR-IQA, enabling efficient MLLM adaptation for low-level vision tasks. The source code is publicly available at https: // github. com/ yahya-ben/ mplug2-vp-for-nriqa.

Figures

Figures reproduced from arXiv: 2509.03494 by Mohammed EL Hassouni, Yahya Benmahane.

Figure 1
Figure 1. Figure 1: Our proposed method for efficiently adapting mPLUG-Owl2 [10] for NR-IQA using visual [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: The following figure illustrates four types of visual prompts applied to sample images from [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

discussion (0)

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

Reference graph

Works this paper leans on

35 extracted references · 17 canonical work pages · 8 internal anchors

  1. [1]

    Completely Blind

    A. Mittal, R. Soundararajan, and A. C. Bovik. Making a “Completely Blind” Image Quality Analyzer. 20(3):209–212. URL: http://ieeexplore.ieee.org/document/6353522/, doi:10.1109/LSP.2012.2227726

    work page arXiv 2012

  2. [2]

    Blind Image Quality Assessment Using A Deep Bilinear Convolutional Neural Network

    Weixia Zhang, Kede Ma, Jia Yan, Dexiang Deng, and Zhou Wang. Blind Image Quality Assessment Using A Deep Bilinear Convolutional Neural Network. 30(1):36–47. URL: http://arxiv.org/abs/1907.02665,doi:10.1109/TCSVT.2018.2886771

    work page internal anchor Pith review Pith/arXiv arXiv 1907

  3. [3]

    MUSIQ: Multi-scale Image Quality Transformer

    Junjie Ke, Qifei Wang, Yilin Wang, Peyman Milanfar, and Feng Yang. MUSIQ: Multi-scale Image Quality Transformer. URL:http://arxiv.org/abs/2108.05997

    Pith/arXiv arXiv

  4. [4]

    Q-Instruct: Improving Low-level Visual Abilities for Multi-modality Foundation Models

    Haoning Wu, Zicheng Zhang, Erli Zhang, Chaofeng Chen, Liang Liao, Annan Wang, Kaixin Xu, Chunyi Li, Jingwen Hou, Guangtao Zhai, Geng Xue, Wenxiu Sun, Qiong Yan, and Weisi Lin. Q-Instruct: Improving Low-level Visual Abilities for Multi-modality Foundation Models. URL:http://arxiv.org/abs/2311.06783

    Pith/arXiv arXiv

  5. [5]

    Q-Align: Teaching LMMs for Visual Scoring via Discrete Text-Defined Levels

    Haoning Wu, Zicheng Zhang, Weixia Zhang, Chaofeng Chen, Liang Liao, Chunyi Li, Yixuan Gao, Annan Wang, Erli Zhang, Wenxiu Sun, Qiong Yan, Xiongkuo Min, Guangtao Zhai, and 8 APREPRINT- SEPTEMBER12, 2025 Weisi Lin. Q-Align: Teaching LMMs for Visual Scoring via Discrete Text-Defined Levels. URL:http://arxiv.org/abs/2312.17090,doi:10.48550/arXiv.2312.17090

    Pith/arXiv arXiv doi:10.48550/arxiv.2312.17090 2025

  6. [6]

    Jianyi Wang, Kelvin C. K. Chan, and Chen Change Loy. Exploring CLIP for Assessing the Look and Feel of Images. URL: http://arxiv.org/abs/2207.12396, doi:10.48550/ arXiv.2207.12396

    Pith/arXiv arXiv doi:10.48550/arxiv.2207.12396

  7. [7]

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

    Haoning Wu, Zicheng Zhang, Erli Zhang, Chaofeng Chen, Liang Liao, Annan Wang, Chunyi Li, Wenxiu Sun, Qiong Yan, Guangtao Zhai, and Weisi Lin. Q-Bench: A Benchmark for General-Purpose Foundation Models on Low-level Vision. URL: http://arxiv.org/abs/ 2309.14181

    Pith/arXiv arXiv

  8. [8]

    A Comprehensive Study of Multimodal Large Language Models for Image Quality Assessment

    Tianhe Wu, Kede Ma, Jie Liang, Yujiu Yang, and Lei Zhang. A Comprehensive Study of Multimodal Large Language Models for Image Quality Assessment. URL: http://arxiv. org/abs/2403.10854,doi:10.48550/arXiv.2403.10854

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2403.10854

  9. [9]

    Blind Image Quality Assessment via Vision-Language Correspondence: A Multitask Learning Perspective

    Weixia Zhang, Guangtao Zhai, Ying Wei, Xiaokang Yang, and Kede Ma. Blind Image Quality Assessment via Vision-Language Correspondence: A Multitask Learning Perspective. URL: http://arxiv.org/abs/2303.14968

    work page internal anchor Pith review Pith/arXiv arXiv

  10. [10]

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

    Qinghao Ye, Haiyang Xu, Jiabo Ye, Ming Yan, Anwen Hu, Haowei Liu, Qi Qian, Ji Zhang, Fei Huang, and Jingren Zhou. mPLUG-Owl2: Revolutionizing Multi-modal Large Language Model with Modality Collaboration. URL:http://arxiv.org/abs/2311.04257

    Pith/arXiv arXiv

  11. [11]

    Exploring Visual Prompts for Adapting Large-Scale Models

    Hyojin Bahng, Ali Jahanian, Swami Sankaranarayanan, and Phillip Isola. Exploring Visual Prompts for Adapting Large-Scale Models. URL: http://arxiv.org/abs/2203.17274, doi:10.48550/arXiv.2203.17274

    Pith/arXiv arXiv doi:10.48550/arxiv.2203.17274

  12. [12]

    KonIQ-10k: An ecologically valid database for deep learning of blind image quality assessment

    Vlad Hosu, Hanhe Lin, Tamas Sziranyi, and Dietmar Saupe. KonIQ-10k: An ecologically valid database for deep learning of blind image quality assessment. 29:4041–4056. URL: http://arxiv.org/abs/1910.06180,doi:10.1109/TIP.2020.2967829

    work page internal anchor Pith review Pith/arXiv arXiv 1910

  13. [13]

    KADID-10k: A Large-scale Artifi- cially Distorted IQA Database

    Hanhe Lin, Vlad Hosu, and Dietmar Saupe. KADID-10k: A Large-scale Artifi- cially Distorted IQA Database. In 2019 Eleventh International Conference on Quality of Multimedia Experience (QoMEX), pages 1–3. URL: https://ieeexplore.ieee.org/ document/8743252,doi:10.1109/QoMEX.2019.8743252

    work page arXiv 2019

  14. [14]

    AGIQA-3K: An Open Database for AI-Generated Image Quality Assessment

    Chunyi Li, Zicheng Zhang, Haoning Wu, Wei Sun, Xiongkuo Min, Xiaohong Liu, Guang- tao Zhai, and Weisi Lin. AGIQA-3K: An Open Database for AI-Generated Image Quality Assessment. URL:http://arxiv.org/abs/2306.04717

    work page internal anchor Pith review Pith/arXiv arXiv

  15. [15]

    Unleashing the Power of Visual Prompting At the Pixel Level

    Junyang Wu, Xianhang Li, Chen Wei, Huiyu Wang, Alan Yuille, Yuyin Zhou, and Cihang Xie. Unleashing the Power of Visual Prompting At the Pixel Level. URL: http://arxiv.org/ abs/2212.10556,doi:10.48550/arXiv.2212.10556

    Pith/arXiv arXiv doi:10.48550/arxiv.2212.10556

  16. [16]

    AutoVP: An Automated Visual Prompting Framework and Benchmark

    Hsi-Ai Tsao, Lei Hsiung, Pin-Yu Chen, Sijia Liu, and Tsung-Yi Ho. AutoVP: An Automated Visual Prompting Framework and Benchmark. URL: http://arxiv.org/abs/2310.08381, doi:10.48550/arXiv.2310.08381

    Pith/arXiv arXiv doi:10.48550/arxiv.2310.08381

  17. [17]

    Understanding and Improving Visual Prompting: A Label-Mapping Perspective

    Aochuan Chen, Yuguang Yao, Pin-Yu Chen, Yihua Zhang, and Sijia Liu. Understanding and Improving Visual Prompting: A Label-Mapping Perspective. URL: http://arxiv.org/abs/ 2211.11635,doi:10.48550/arXiv.2211.11635

    Pith/arXiv arXiv doi:10.48550/arxiv.2211.11635

  18. [18]

    End-to-End Blind Image Quality Assessment Using Deep Neural Networks

    Kede Ma, Wentao Liu, Kai Zhang, Zhengfang Duanmu, Zhou Wang, and Wangmeng Zuo. End-to-End Blind Image Quality Assessment Using Deep Neural Networks. 27(3):1202–

    work page

  19. [19]

    Uncertainty-Aware Blind Image Quality Assessment in the Laboratory and Wild

    Weixia Zhang, Kede Ma, Guangtao Zhai, and Xiaokang Yang. Uncertainty-Aware Blind Image Quality Assessment in the Laboratory and Wild. 30:3474–3486. URL: http://arxiv.org/ abs/2005.13983,doi:10.1109/TIP.2021.3061932

    work page internal anchor Pith review Pith/arXiv arXiv 2005

  20. [20]

    MANIQA: Multi-dimension Attention Network for No-Reference Image Quality Assessment

    Sidi Yang, Tianhe Wu, Shuwei Shi, Shanshan Lao, Yuan Gong, Mingdeng Cao, Jiahao Wang, and Yujiu Yang. MANIQA: Multi-dimension Attention Network for No-Reference Image Quality Assessment. In In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), pages 1190–1199. IEEE. URL: https:// ieeexplore.ieee.org/docume...

    work page arXiv 2022

  21. [21]

    Madhusudana, Neil Birkbeck, Yilin Wang, Balu Adsumilli, and Alan C

    Pavan C. Madhusudana, Neil Birkbeck, Yilin Wang, Balu Adsumilli, and Alan C. Bovik. Image Quality Assessment using Contrastive Learning. URL: http://arxiv.org/abs/2110. 13266

    work page

  22. [22]

    Learning Transferable Visual Models From Natural Language Supervision

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agar- wal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, and Ilya Sutskever. Learning Transferable Visual Models From Natural Language Supervision. URL: http://arxiv.org/abs/2103.00020

    Pith/arXiv arXiv

  23. [23]

    Learning to Prompt for Vision-Language Models

    Kaiyang Zhou, Jingkang Yang, Chen Change Loy, and Ziwei Liu. Learning to Prompt for Vision-Language Models. 130(9):2337–2348.doi:10.1007/s11263-022-01653-1

    doi:10.1007/s11263-022-01653-1

  24. [24]

    A Survey of Automatic Prompt Engineering: An Optimization Perspective

    Wenwu Li, Xiangfeng Wang, Wenhao Li, and Bo Jin. A Survey of Automatic Prompt Engineering: An Optimization Perspective. URL: http://arxiv.org/abs/2502.11560, doi:10.48550/arXiv.2502.11560

    Pith/arXiv arXiv doi:10.48550/arxiv.2502.11560

  25. [25]

    Robust Adaptation of Foundation Models with Black-Box Visual Prompting

    Changdae Oh, Gyeongdeok Seo, Geunyoung Jung, Zhi-Qi Cheng, Hosik Choi, Jiyoung Jung, and Kyungwoo Song. Robust Adaptation of Foundation Models with Black-Box Visual Prompt- ing. URL:http://arxiv.org/abs/2407.17491,doi:10.48550/arXiv.2407.17491

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2407.17491

  26. [26]

    Multi-Layer Cross-Modal Prompt Fusion for No-Reference Image Quality Assess- ment

    Yang Lu, Zilu Zhou, Zifan Yang, Shuangyao Han, Xiaoheng Jiang, and Mingliang Xu. Multi-Layer Cross-Modal Prompt Fusion for No-Reference Image Quality Assess- ment. 88:103045. URL: https://www.sciencedirect.com/science/article/pii/ S0141938225000824,doi:10.1016/j.displa.2025.103045

    arXiv 2025

  27. [27]

    Multi-Modal Prompt Learning on Blind Image Quality Assessment

    Wensheng Pan, Timin Gao, Yan Zhang, Runze Hu, Xiawu Zheng, Enwei Zhang, Yuting Gao, Yutao Liu, Yunhang Shen, Ke Li, Shengchuan Zhang, Liujuan Cao, and Rongrong Ji. Multi- Modal Prompt Learning on Blind Image Quality Assessment. URL: http://arxiv.org/ abs/2404.14949,doi:10.48550/arXiv.2404.14949

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2404.14949

  28. [28]

    Q-Adapt: Adapting LMM for Visual Quality Assessment with Progressive Instruction Tuning

    Yiting Lu, Xin Li, Haoning Wu, Bingchen Li, Weisi Lin, and Zhibo Chen. Q-Adapt: Adapting LMM for Visual Quality Assessment with Progressive Instruction Tuning. URL: http:// arxiv.org/abs/2504.01655,doi:10.48550/arXiv.2504.01655

    Pith/arXiv arXiv doi:10.48550/arxiv.2504.01655

  29. [29]

    A Survey on Multimodal Large Language Models

    Shukang Yin, Chaoyou Fu, Sirui Zhao, Ke Li, Xing Sun, Tong Xu, and Enhong Chen. A Survey on Multimodal Large Language Models. URL:http://arxiv.org/abs/2306.13549

    Pith/arXiv arXiv

  30. [31]

    Alireza Golestaneh, Saba Dadsetan, and Kris M

    S. Alireza Golestaneh, Saba Dadsetan, and Kris M. Kitani. No-Reference Image Quality Assessment via Transformers, Relative Ranking, and Self-Consistency. In In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (W ACV), pages 3989–3999. IEEE. URL: https://ieeexplore.ieee.org/document/9706735/, doi:10. 1109/WACV51458.2022.00404

    work page arXiv 2022

  31. [32]

    Perceptual Quality As- sessment of Smartphone Photography

    Yuming Fang, Hanwei Zhu, Yan Zeng, Kede Ma, and Zhou Wang. Perceptual Quality As- sessment of Smartphone Photography. In In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 3674–3683, Seattle, W A, USA, June

    work page

  32. [33]

    From Patches to Pictures (PaQ-2-PiQ): Mapping the Perceptual Space of Picture Quality, December 2019.doi:10.48550/arXiv.1912.10088

    Zhenqiang Ying, Haoran Niu, Praful Gupta, Dhruv Mahajan, Deepti Ghadiyaram, and Alan Bovik. From Patches to Pictures (PaQ-2-PiQ): Mapping the Perceptual Space of Picture Quality, December 2019.doi:10.48550/arXiv.1912.10088

    Pith/arXiv arXiv doi:10.48550/arxiv.1912.10088 2019

  33. [34]

    https://ieeexplore.ieee.org/document/7327186

    Massive Online Crowdsourced Study of Subjective and Objective Picture Quality | IEEE Journals & Magazine | IEEE Xplore. https://ieeexplore.ieee.org/document/7327186. 10

    work page arXiv

  34. [1213]

    2017.2774045

    URL: http://ieeexplore.ieee.org/document/8110690/, doi:10.1109/TIP. 2017.2774045

    arXiv 2017

  35. [2020]

    IEEE.doi:10.1109/cvpr42600.2020.00373

    work page arXiv 2020

This paper was first reviewed by deepseek-v4-flash on August 5, 2026.