Pith. sign in

REVIEW 3 major objections 6 minor 61 references

EyeSim-VQA: A Free-Energy-Guided Eye Simulation Framework for Video Quality Assessment

T0 review · 3 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

Pith's one-line read EyeSim-VQA claims that free-energy-style self-repair before assessment lifts no-reference video quality prediction to state-of-the-art on several benchmarks.

desk verdict A well-ablated but uncontrolled comparison: the architecture is plausible, the SOTA claim needs a rerun on matched splits. read the letter →

arxiv 2506.11549 v1 pith:HEDEFJIN submitted 2025-06-13 cs.CV eess.IV

classification cs.CVeess.IV
keywords videoqualityassessmentno-referenceVQAfree-energyprinciplevisualself-repairdual-brancharchitectureDyTattentionMambascanninguser-generatedcontent
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

EyeSim-VQA applies the free-energy principle of visual self-repair—the idea that the visual system actively reconstructs degraded input—to no-reference video quality assessment, a setting where it has not been used before. The model first enhances the input video in two complementary ways: a global branch cleans resized full frames with a lightweight super-resolution network, while a technical branch iteratively restores patch fragments with a residual network. A biologically inspired prediction head then fuses the two representations through a DyT-based gaze module and a Mamba-inspired scan module. The paper reports that the full model achieves state-of-the-art or competitive SROCC/PLCC on KVQ, KoNViD-1k, YouTube-UGC, LIVE-VQC, and LSVQ, in several cases outperforming LLM-based models. The authors note that the enhancement modules must be pre-trained separately before VQA training, that the framework does not yet use LLM knowledge, and that the restoration backbones are limited to two classical video-super-resolution designs.

What carries the argument

The load-bearing mechanism is the pairing of free-energy-style visual enhancement with a two-branch prediction head. In the aesthetic branch, a lightweight BasicVSR-mini network super-resolves and denoises whole frames; in the technical branch, CleanNet iteratively restores 32-frame patch fragments. Each enhanced output is concatenated with the original input before the backbone, avoiding changes to pre-trained weights. The head then models human perception with a DyT-based gaze branch, using dynamic tanh attention $\mathrm{DyT}(x)=\tanh(\alpha\cdot x)\odot w+b$, and a Mamba-inspired scan branch that aggregates temporal neighborhood features before projection. The combination is what the paper credits for improved global-local fusion and for the reported gains in SROCC/PLCC.

What would settle it

Re-run EyeSim-VQA and every Table I baseline on identical fixed 8:2 splits across several random seeds; if the reported margins (for example 0.870 vs 0.867 SROCC on KVQ, or 0.919 vs 0.922 on KoNViD-1k) flip sign or vanish under any seed, the claimed state-of-the-art status is an artifact of split choice rather than of the model.

Watch

Extended reading notes

Core claim

The central claim is that modeling how the human eye repairs degraded scenes improves automated video quality judgment. Concretely, EyeSim-VQA asserts that enhancing a video before feature extraction—restoring full frames globally and patch fragments locally—produces features that correlate more tightly with human mean opinion scores, and that a dual-branch prediction head which scans frames sequentially (Mamba-style temporal convolution) and gazes selectively (DyT-based attention) fuses global and local evidence better than a single regressor. On Table I, the model reports the best or second-best correlation on each of five benchmarks, including 0.870 SROCC on KVQ (above KSVQE's 0.867), 0.886 on LIVE-VQC, and 0.888 SROCC on LSVQ test, with its largest margin on LSVQ 1080p (0.807 vs KSVQE's 0.790). The paper's abstract states the outcome as 'competitive or superior,' while the contributions section describes it as state-of-the-art.

Load-bearing premise

The load-bearing premise is that the 8:2 random train/test splits used for KVQ, KoNViD-1k, YouTube-UGC, and LIVE-VQC are comparable to the splits on which the cited baseline results in Table I were originally measured; the paper says it follows 'the partitioning protocol from our previous work' without naming that work or giving a seed.

Editorial extensions

If this is right

  • Free-energy self-repair, previously confined to image quality assessment, transfers to video quality assessment without requiring retraining of the pre-trained backbone.
  • The enhancement step can be added by concatenating restored frames with originals before feature extraction, so existing VQA backbones do not need adapters or modified weights.
  • A scan-and-gaze prediction head trained from scratch outperforms a single-head regressor, and replacing LayerNorm with DyT in this head improves accuracy.
  • Cross-dataset tests indicate the enhancement mechanism helps generalization; the paper reports strong transfer in both directions between KVQ and other datasets.
  • The composite restoration loss, including a CONTRIQUE-based quality penalty, steers restoration toward perceptual quality rather than pixel fidelity alone.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Training the enhancement networks on synthetic distortions that match each target benchmark would test whether the reported cross-dataset gains come from the self-repair mechanism or from distribution overlap between synthetic and real distortions.
  • The scan branch approximates Mamba with a compact temporal convolution; replacing it with an actual selective state-space scan under the same parameter budget would show whether sequential sweeping is the active ingredient or a stand-in.
  • An even simpler control, swapping the learned enhancement modules for a fixed denoiser or sharpening filter, would isolate how much of the gain requires learned self-repair rather than any pre-processing.
  • The absence of a named split or seed means the Table I comparison can be made airtight only by re-evaluating all baselines under the authors' exact protocol; this is a reproducibility check, not a claim about model performance.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. The paper proposes EyeSim-VQA, a no-reference video quality assessment framework that combines a dual-branch architecture (aesthetic and technical) with free-energy-inspired visual enhancement modules, a biologically motivated prediction head that models gaze and scanning behavior via DyT-based attention and a Mamba-inspired temporal scan, and a composite enhancement loss with pixel, identity, and CONTRIQUE-based quality terms. The method is evaluated on KVQ, KoNViD-1k, YouTube-UGC, LIVE-VQC, and LSVQ, and the paper claims state-of-the-art or competitive performance, supported by extensive ablation studies on each component.

Significance. If the empirical comparison is sound, the paper makes a meaningful contribution by extending free-energy-guided self-repair from IQA to VQA, a direction that has been largely unexplored. The ablation study is unusually thorough, covering the enhancement modules, the dual-branch head, the embedding strategy, the number of enhanced frames, the DyT placement, and the loss components. The biologically inspired prediction head is a plausible architectural novelty, and the cross-dataset experiments, if properly controlled, would strengthen the generalization claim. However, the central SOTA claim currently depends on an evaluation protocol that is not fully specified, so the significance cannot be fully assessed until the comparison is made reproducible and fair.

major comments (3)
  1. [§IV-A, Table I] The SOTA claim rests on Table I, but the evaluation protocol is not specified in a way that supports head-to-head comparison. Section IV-A says that for KVQ, KoNViD-1k, YouTube-UGC, and LIVE-VQC, the authors 'follow the partitioning protocol from our previous work by randomly splitting each dataset into training and testing sets at an 8:2 ratio,' without naming the previous work, giving a seed, or providing split indices. The baseline numbers in Table I appear to be quoted from the original papers, which use their own split conventions. The reported advantages are small (e.g., KVQ SROCC 0.870 vs 0.867 for KSVQE, YouTube-UGC 0.902 vs 0.900, LSVQ-test 0.888 vs 0.886), and random split variance on datasets of these sizes can easily exceed these margins. Please rerun all baselines on the exact same splits, provide the split indices or seed, and report results across multiple random splits with confidence intervals or variance.
  2. [§IV-D-11, Table XII] The loss hyperparameters alpha_1=0.3 and alpha_2=0.01 are selected using Table XII, which is explicitly 'tested on KoNViD-1k.' The same dataset and split are then reported as a headline result in Table I. This constitutes selection on the test set and can inflate the reported KoNViD-1k performance relative to baselines that did not tune on that test set. The paper should either use a held-out validation split for hyperparameter selection, report results for default or non-tuned hyperparameters, or otherwise demonstrate that the reported improvements are not an artifact of test-set tuning.
  3. [§IV-C-2, Tables II and III] The cross-dataset generalization claim is supported by Tables II and III, but the experimental protocol for the baseline entries is not described. It is not stated whether SimpleVQA, FastVQA, and KSVQE were retrained by the authors under the same training splits and preprocessing, or whether the numbers are quoted from the KVQ paper. If the latter, the comparison may not be controlled for the same training data, input sampling, or evaluation code. Please specify the protocol for every entry, or rerun the baselines under identical conditions.
minor comments (6)
  1. [Abstract and throughout] The method name is written inconsistently as 'EyeSimVQA' in the abstract and 'EyeSim-VQA' elsewhere; please use one consistent name.
  2. [§IV-C-1] The text says the comparison includes '13 representative approaches,' but the list and Table I contain 14 methods; please correct the count.
  3. [Table VII] The column headers 'ET' and 'EA' are not defined; spell them out as technical-branch enhancement and aesthetic-branch enhancement.
  4. [Table XI] The caption says 'Loss Function Hyperparameters,' but the table rows are the loss components (L_pixel, L_identity, L_IQA); retitle the caption to match the content.
  5. [References] Reference [59] is cited for AdamW but is the Adam paper; please cite the AdamW source or correct the reference.
  6. [Equation (2)] Equation (2) defines K=Q with a pooled query, but the attention computation between the pooled vector and each spatial/temporal position is terse; a short clarifying sentence or diagram would help readers understand the gaze mechanism.

Circularity Check

1 steps flagged · score 6.0 of 10

KoNViD-1k SOTA result is selected, not predicted: loss hyperparameters and components are chosen on KoNViD-1k and the same dataset is reported as the benchmark; other datasets remain independent.

  1. fitted input called prediction [Section IV-D-11, Table XII; Table I (KoNViD-1k column)]
    "Through extensive experiments, we find that setting α1 = 0.3 and α2 = 0.01 yields the best performance ... (Table XII caption: 'Ablation Study on the Design of Loss Function Hyperparameters. Results are Tested on the KoNViD-1k Dataset.'); Table I then reports EyeSim-VQA KoNViD-1k SROCC 0.919 / PLCC 0.918."

    The final model's loss weights are selected by maximizing SROCC/PLCC on KoNViD-1k (Table XII, α1=0.3, α2=0.01). Tables V, VII, VIII, IX, X, and XI likewise select the enhancement modules, embedding strategy, frame count, DyT placement, and loss composition on KoNViD-1k. The same KoNViD-1k column is then presented in Table I as evidence of SOTA performance (0.919 vs DOVER 0.908). Thus the KoNViD-1k comparison is a model-selection outcome rather than an independent prediction. The claim is only partly circular because KVQ, YouTube-UGC, LIVE-VQC, and LSVQ are not used for hyperparameter tuning (YouTube-UGC appears in one component ablation) and provide independent evidence.

full rationale

The core VQA pipeline is trained on human MOS and is not definitionally circular: the quality score is not constructed from the MOS labels used as inputs, and the free-energy enhancement module is supervised by pixel reconstruction and CONTRIQUE, an external NR-IQA model, not by the VQA target. The split-protocol sentence ('follow the partitioning protocol from our previous work') is a self-reference, but it is a reproducibility/fairness concern rather than a circular derivation. The substantive circularity is the repeated use of KoNViD-1k as both the model-selection/ablation dataset and one of the headline benchmarks. After the architecture, embedding strategy, number of enhanced frames, DyT placement, loss terms, and α1/α2 are all chosen by KoNViD-1k SROCC/PLCC in Tables V-XII, reporting the KoNViD-1k row of Table I as a SOTA result is effectively reporting a fitted quantity. The paper's abstract claim of 'competitive or superior performance' is still supported by the other four datasets, so the circularity is partial and localized, not total.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The paper introduces no new physical entities; its contributions are architectural modules and a training scheme. The load-bearing assumptions are the representativeness of synthetic distortions, the validity of CONTRIQUE as a perceptual supervisor, the comparability of the data splits, and the sufficiency of two-stage training. Hyperparameters alpha1, alpha2, refinement steps, and the enhanced-frame ratio are tuned on the test split, which is a form of post-hoc selection.

free parameters (5)
  • alpha_1 (identity loss weight) = 0.3
    Selected by ablation on the KoNViD test split (Table XII); balances pixel fidelity against over-modification of high-quality frames.
  • alpha_2 (IQA loss weight) = 0.01
    Selected by ablation on KoNViD (Table XII); kept small because CONTRIQUE is less sensitive to pixel-level details.
  • CleanNet refinement steps = 3
    Empirically fixed as a trade-off between enhancement quality and compute (Section III-B2).
  • Concatenated enhanced frame count = 10 of 32 frames
    Chosen by ablation on KoNViD (Table IX); replacing about one-third of frames with enhanced versions was best.
  • Synthetic distortion parameter ranges = unspecified
    Section IV-B says distortion parameters are 'randomly sampled within appropriate ranges' but the ranges are not given, making the enhancement pretraining dataset an unstated degree of freedom.
assumptions (4)
  • domain assumption The five synthetic degradations (Gaussian noise, motion blur, JPEG, salt-and-pepper, mean blur) are representative of real UGC distortion.
    Section IV-B: used to pretrain the enhancement network; if unrepresentative, the enhancement may not transfer to test videos.
  • domain assumption CONTRIQUE's predicted quality scores are a valid perceptual target for training the enhancement network.
    Section III-D Eq. 8: LIQA uses CONTRIQUE scores; the enhancement is optimized to satisfy an IQA model rather than human labels directly.
  • ad hoc to paper The 8:2 random split from the authors' prior work is comparable to the splits used in the original baseline papers.
    Section IV-A: the split protocol is not described or cited, yet Table I compares against numbers from the original papers.
  • domain assumption A two-stage training procedure (pretrain enhancement, then train VQA with frozen enhancement) is sufficient; joint training is not required.
    Section V states pretraining the enhancement module is required; no experiments compare against end-to-end joint training.

how reviews work

0 comments
Cite this review

Pith. "Pith review of EyeSim-VQA: A Free-Energy-Guided Eye Simulation Framework for Video Quality Assessment." pith.science (2026). https://pith.science/paper/HEDEFJIN

@misc{pith2026250611549,
  author       = {Pith},
  title        = {Pith review of: EyeSim-VQA: A Free-Energy-Guided Eye Simulation Framework for Video Quality Assessment},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HEDEFJIN}},
  note         = {Machine review of arXiv:2506.11549}
}
read the original abstract

Free-energy-guided self-repair mechanisms have shown promising results in image quality assessment (IQA), but remain under-explored in video quality assessment (VQA), where temporal dynamics and model constraints pose unique challenges. Unlike static images, video content exhibits richer spatiotemporal complexity, making perceptual restoration more difficult. Moreover, VQA systems often rely on pre-trained backbones, which limits the direct integration of enhancement modules without affecting model stability. To address these issues, we propose EyeSimVQA, a novel VQA framework that incorporates free-energy-based self-repair. It adopts a dual-branch architecture, with an aesthetic branch for global perceptual evaluation and a technical branch for fine-grained structural and semantic analysis. Each branch integrates specialized enhancement modules tailored to distinct visual inputs-resized full-frame images and patch-based fragments-to simulate adaptive repair behaviors. We also explore a principled strategy for incorporating high-level visual features without disrupting the original backbone. In addition, we design a biologically inspired prediction head that models sweeping gaze dynamics to better fuse global and local representations for quality prediction. Experiments on five public VQA benchmarks demonstrate that EyeSimVQA achieves competitive or superior performance compared to state-of-the-art methods, while offering improved interpretability through its biologically grounded design.

Figures

Figures reproduced from arXiv: 2506.11549 by the authors.

Figure 1
Figure 1. Overview of the proposed EyeSim-VQA framework. (a) View Decomposition module, shared with existing models such as DOVER [21] and KSVQE [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Illustration of the visual enhancement module. (a) The aesthetic branch adopts a lightweight BasicVSR [33] (BasicVSR-mini) architecture with reduced [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. (a) Schematic illustration of the proposed EyeSim-VQA Head, which simulates the human visual mechanisms of gaze and scanning through a [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Overview of the training pipeline for visual enhancement. The opti [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Visualization of clustering results using t-SNE [61] on features extracted by the backbone networks from the aesthetic and technical branches. The [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 7
Figure 7. Figure 7: Visualization of video quality enhancement results from the Aesthetic [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

61 extracted references · 40 canonical work pages

  1. [1]

    Discovqa: Temporal distortion-content transformers for video quality assessment,

    H. Wu, C. Chen, L. Liao, J. Hou, W. Sun, Q. Yan, and W. Lin, “Discovqa: Temporal distortion-content transformers for video quality assessment,”IEEE Transactions on Circuits and Systems for Video Technology, vol. 33, no. 9, pp. 4840–4854, 2023

  2. [2]

    Spatiotemporal representation learning for blind video quality assessment,

    Y . Liu, J. Wu, L. Li, W. Dong, J. Zhang, and G. Shi, “Spatiotemporal representation learning for blind video quality assessment,”IEEE Trans- actions on Circuits and Systems for Video Technology, vol. 32, no. 6, pp. 3500–3513, 2021

  3. [3]

    Generative latent coding for ultra-low bitrate image and video compression,

    L. Qi, Z. Jia, J. Li, B. Li, H. Li, and Y . Lu, “Generative latent coding for ultra-low bitrate image and video compression,”IEEE Transactions on Circuits and Systems for Video Technology, 2025

  4. [4]

    Temporal wavelet transform- based low-complexity perceptual quality enhancement of compressed video,

    C. Dong, H. Ma, Z. Li, L. Li, and D. Liu, “Temporal wavelet transform- based low-complexity perceptual quality enhancement of compressed video,”IEEE Transactions on Circuits and Systems for Video Technol- ogy, vol. 34, no. 5, pp. 4040–4053, 2023

  5. [5]

    Perceptual quality as- sessment of smartphone photography,

    Y . Fang, H. Zhu, Y . Zeng, K. Ma, and Z. Wang, “Perceptual quality as- sessment of smartphone photography,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 3677– 3686

  6. [6]

    Learning cross- scale weighted prediction for efficient neural video compression,

    Z. Guo, R. Feng, Z. Zhang, X. Jin, and Z. Chen, “Learning cross- scale weighted prediction for efficient neural video compression,”IEEE Transactions on Image Processing, vol. 32, pp. 3567–3579, 2023

  7. [7]

    The konstanz natural video database (konvid-1k),

    V . Hosu, F. Hahn, M. Jenadeleh, H. Lin, H. Men, T. Szir ´anyi, S. Li, and D. Saupe, “The konstanz natural video database (konvid-1k),” in 2017 Ninth international conference on quality of multimedia experience (QoMEX). IEEE, 2017, pp. 1–6

  8. [8]

    Youtube ugc dataset for video compression research,

    Y . Wang, S. Inguva, and B. Adsumilli, “Youtube ugc dataset for video compression research,” in2019 IEEE 21st international workshop on multimedia signal processing (MMSP). IEEE, 2019, pp. 1–5

Show all 61 references
  1. [9]

    Patch-vq:’patching up’the video quality problem,

    Z. Ying, M. Mandal, D. Ghadiyaram, and A. Bovik, “Patch-vq:’patching up’the video quality problem,” inProceedings of the IEEE/CVF confer- ence on computer vision and pattern recognition, 2021, pp. 14 019– 14 029

  2. [10]

    Kvq: Kwai video quality assessment for short-form videos,

    Y . Lu, X. Li, Y . Pei, K. Yuan, Q. Xie, Y . Qu, M. Sun, C. Zhou, and Z. Chen, “Kvq: Kwai video quality assessment for short-form videos,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 25 963–25 973

  3. [11]

    Finevq: Fine-grained user generated content video quality assessment,

    H. Duan, Q. Hu, J. Wang, L. Yang, Z. Xu, L. Liu, X. Min, C. Cai, T. Ye, X. Zhanget al., “Finevq: Fine-grained user generated content video quality assessment,”arXiv preprint arXiv:2412.19238, 2024. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 13

  4. [12]

    Large-scale study of perceptual video quality,

    Z. Sinno and A. C. Bovik, “Large-scale study of perceptual video quality,”IEEE Transactions on Image Processing, vol. 28, no. 2, pp. 612–627, 2018

  5. [13]

    Image sharpness assessment based on local phase coherence,

    R. Hassen, Z. Wang, and M. M. Salama, “Image sharpness assessment based on local phase coherence,”IEEE Transactions on Image Process- ing, vol. 22, no. 7, pp. 2798–2810, 2013

  6. [14]

    No-reference quality assessment of tone-mapped hdr pictures,

    D. Kundu, D. Ghadiyaram, A. C. Bovik, and B. L. Evans, “No-reference quality assessment of tone-mapped hdr pictures,”IEEE Transactions on Image Processing, vol. 26, no. 6, pp. 2957–2971, 2017

  7. [15]

    No-reference image quality assessment in the spatial domain,

    A. Mittal, A. K. Moorthy, and A. C. Bovik, “No-reference image quality assessment in the spatial domain,”IEEE Transactions on image processing, vol. 21, no. 12, pp. 4695–4708, 2012

  8. [16]

    Making a “completely blind

    A. Mittal, R. Soundararajan, and A. C. Bovik, “Making a “completely blind” image quality analyzer,”IEEE Signal processing letters, vol. 20, no. 3, pp. 209–212, 2012

  9. [17]

    Blind image quality assessment using joint statistics of gradient magnitude and laplacian features,

    W. Xue, X. Mou, L. Zhang, A. C. Bovik, and X. Feng, “Blind image quality assessment using joint statistics of gradient magnitude and laplacian features,”IEEE Transactions on Image Processing, vol. 23, no. 11, pp. 4850–4862, 2014

  10. [18]

    Quality assessment of in-the-wild videos,

    D. Li, T. Jiang, and M. Jiang, “Quality assessment of in-the-wild videos,” inProceedings of the 27th ACM international conference on multimedia, 2019, pp. 2351–2359

  11. [19]

    Fast-vqa: Efficient end-to-end video quality assessment with fragment sampling,

    H. Wu, C. Chen, J. Hou, L. Liao, A. Wang, W. Sun, Q. Yan, and W. Lin, “Fast-vqa: Efficient end-to-end video quality assessment with fragment sampling,” inEuropean conference on computer vision. Springer, 2022, pp. 538–554

  12. [20]

    Neighbourhood representative sampling for efficient end-to- end video quality assessment,

    H. Wu, C. Chen, L. Liao, J. Hou, W. Sun, Q. Yan, J. Gu, and W. Lin, “Neighbourhood representative sampling for efficient end-to- end video quality assessment,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 12, pp. 15 185–15 202, 2023

  13. [21]

    Disentangling aesthetic and technical effects for video quality assessment of user generated content,

    H. Wu, L. Liao, C. Chen, J. Hou, A. Wang, W. Sun, Q. Yan, and W. Lin, “Disentangling aesthetic and technical effects for video quality assessment of user generated content,”arXiv preprint arXiv:2211.04894, vol. 2, no. 5, p. 6, 2022

  14. [22]

    Clif-vqa: Enhancing video quality assessment by incorporating high-level semantic infor- mation related to human feelings,

    Y . Mi, Y . Shu, Y . Li, C. Hui, P. Zhou, and S. Liu, “Clif-vqa: Enhancing video quality assessment by incorporating high-level semantic infor- mation related to human feelings,” inProceedings of the 32nd ACM International Conference on Multimedia, 2024, pp. 9989–9998

  15. [23]

    Learning spatio-temporal features with 3d residual networks for action recognition,

    K. Hara, H. Kataoka, and Y . Satoh, “Learning spatio-temporal features with 3d residual networks for action recognition,” inProceedings of the IEEE international conference on computer vision workshops, 2017, pp. 3154–3160

  16. [24]

    Can spatiotemporal 3d cnns retrace the history of 2d cnns and imagenet?

    ——, “Can spatiotemporal 3d cnns retrace the history of 2d cnns and imagenet?” inProceedings of the IEEE conference on Computer Vision and Pattern Recognition, 2018, pp. 6546–6555

  17. [25]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” inProceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 770–778

  18. [26]

    Identity mappings in deep residual networks,

    ——, “Identity mappings in deep residual networks,” inComputer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11–14, 2016, Proceedings, Part IV 14. Springer, 2016, pp. 630–645

  19. [27]

    Efficientnet: Rethinking model scaling for con- volutional neural networks,

    M. Tan and Q. Le, “Efficientnet: Rethinking model scaling for con- volutional neural networks,” inInternational conference on machine learning. PMLR, 2019, pp. 6105–6114

  20. [28]

    Efficientnetv2: Smaller models and faster training,

    ——, “Efficientnetv2: Smaller models and faster training,” inInterna- tional conference on machine learning. PMLR, 2021, pp. 10 096– 10 106

  21. [29]

    Vcrnet: Visual compensation restoration network for no-reference image quality assessment,

    Z. Pan, F. Yuan, J. Lei, Y . Fang, X. Shao, and S. Kwong, “Vcrnet: Visual compensation restoration network for no-reference image quality assessment,”IEEE Transactions on Image Processing, vol. 31, pp. 1613– 1627, 2022

  22. [30]

    Dif- fusion model-based visual compensation guidance and visual difference analysis for no-reference image quality assessment,

    Z. Wang, B. Hu, M. Zhang, J. Li, L. Li, M. Gong, and X. Gao, “Dif- fusion model-based visual compensation guidance and visual difference analysis for no-reference image quality assessment,”IEEE Transactions on Image Processing, 2025

  23. [31]

    Ran4iqa: Restorative adversarial nets for no-reference image quality assessment,

    H. Ren, D. Chen, and Y . Wang, “Ran4iqa: Restorative adversarial nets for no-reference image quality assessment,” inProceedings of the AAAI conference on artificial intelligence, vol. 32, no. 1, 2018

  24. [32]

    Hallucinated-iqa: No-reference image quality assessment via adversarial learning,

    K.-Y . Lin and G. Wang, “Hallucinated-iqa: No-reference image quality assessment via adversarial learning,” inProceedings of the IEEE con- ference on computer vision and pattern recognition, 2018, pp. 732–741

  25. [33]

    Basicvsr: The search for essential components in video super-resolution and beyond,

    K. C. Chan, X. Wang, K. Yu, C. Dong, and C. C. Loy, “Basicvsr: The search for essential components in video super-resolution and beyond,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2021, pp. 4947–4956

  26. [34]

    Basicvsr++: Improving video super-resolution with enhanced propagation and alignment,

    K. C. Chan, S. Zhou, X. Xu, and C. C. Loy, “Basicvsr++: Improving video super-resolution with enhanced propagation and alignment,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 5972–5981

  27. [35]

    Learning generalized spatial-temporal deep feature representation for no-reference video quality assessment,

    B. Chen, L. Zhu, G. Li, F. Lu, H. Fan, and S. Wang, “Learning generalized spatial-temporal deep feature representation for no-reference video quality assessment,”IEEE Transactions on Circuits and Systems for Video Technology, vol. 32, no. 4, pp. 1903–1916, 2021

  28. [36]

    End-to-end blind quality assessment of compressed videos using deep neural networks

    W. Liu, Z. Duanmu, and Z. Wang, “End-to-end blind quality assessment of compressed videos using deep neural networks.” inACM Multimedia, 2018, pp. 546–554

  29. [37]

    Transformers without normalization,

    J. Zhu, X. Chen, K. He, Y . LeCun, and Z. Liu, “Transformers without normalization,”arXiv preprint arXiv:2503.10622, 2025

  30. [38]

    Mamba: Linear-time sequence modeling with selective state spaces,

    A. Gu and T. Dao, “Mamba: Linear-time sequence modeling with selective state spaces,”arXiv preprint arXiv:2312.00752, 2023

  31. [39]

    No-reference video quality assessment via feature learning,

    J. Xu, P. Ye, Y . Liu, and D. Doermann, “No-reference video quality assessment via feature learning,” in2014 IEEE international conference on image processing (ICIP). IEEE, 2014, pp. 491–495

  32. [40]

    Unsupervised feature learning framework for no-reference image quality assessment,

    P. Ye, J. Kumar, L. Kang, and D. Doermann, “Unsupervised feature learning framework for no-reference image quality assessment,” in2012 IEEE conference on computer vision and pattern recognition. IEEE, 2012, pp. 1098–1105

  33. [41]

    Blind prediction of natural video quality,

    M. A. Saad, A. C. Bovik, and C. Charrier, “Blind prediction of natural video quality,”IEEE Transactions on image Processing, vol. 23, no. 3, pp. 1352–1365, 2014

  34. [42]

    Two-level approach for no-reference consumer video quality assessment,

    J. Korhonen, “Two-level approach for no-reference consumer video quality assessment,”IEEE Transactions on Image Processing, vol. 28, no. 12, pp. 5923–5938, 2019

  35. [43]

    Ugc- vqa: Benchmarking blind video quality assessment for user generated content,

    Z. Tu, Y . Wang, N. Birkbeck, B. Adsumilli, and A. C. Bovik, “Ugc- vqa: Benchmarking blind video quality assessment for user generated content,”IEEE Transactions on Image Processing, vol. 30, pp. 4449– 4464, 2021

  36. [44]

    Imagenet: A large-scale hierarchical image database,

    J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, and L. Fei-Fei, “Imagenet: A large-scale hierarchical image database,” in2009 IEEE conference on computer vision and pattern recognition. Ieee, 2009, pp. 248–255

  37. [45]

    Learning phrase representations using rnn encoder-decoder for statistical machine translation,

    K. Cho, B. Van Merri ¨enboer, C. Gulcehre, D. Bahdanau, F. Bougares, H. Schwenk, and Y . Bengio, “Learning phrase representations using rnn encoder-decoder for statistical machine translation,”arXiv preprint arXiv:1406.1078, 2014

  38. [46]

    Video swin transformer,

    Z. Liu, J. Ning, Y . Cao, Y . Wei, Z. Zhang, S. Lin, and H. Hu, “Video swin transformer,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 3202–3211

  39. [47]

    Vision mamba: A comprehensive survey and taxonomy,

    X. Liu, C. Zhang, and L. Zhang, “Vision mamba: A comprehensive survey and taxonomy,”arXiv preprint arXiv:2405.04404, 2024

  40. [48]

    A survey on visual mamba,

    H. Zhang, Y . Zhu, D. Wang, L. Zhang, T. Chen, Z. Wang, and Z. Ye, “A survey on visual mamba,”Applied Sciences, vol. 14, no. 13, p. 5683, 2024

  41. [49]

    Investigating tradeoffs in real-world video super-resolution,

    K. C. Chan, S. Zhou, X. Xu, and C. C. Loy, “Investigating tradeoffs in real-world video super-resolution,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 5962–5971

  42. [50]

    Understanding and im- proving layer normalization,

    J. Xu, X. Sun, Z. Zhang, G. Zhao, and J. Lin, “Understanding and im- proving layer normalization,”Advances in neural information processing systems, vol. 32, 2019

  43. [51]

    Two de- terministic half-quadratic regularization algorithms for computed imag- ing,

    P. Charbonnier, L. Blanc-Feraud, G. Aubert, and M. Barlaud, “Two de- terministic half-quadratic regularization algorithms for computed imag- ing,” inProceedings of 1st international conference on image processing, vol. 2. IEEE, 1994, pp. 168–172

  44. [52]

    Image quality assessment using contrastive learning,

    P. C. Madhusudana, N. Birkbeck, Y . Wang, B. Adsumilli, and A. C. Bovik, “Image quality assessment using contrastive learning,”IEEE Transactions on Image Processing, vol. 31, pp. 4149–4161, 2022

  45. [53]

    A completely blind video quality evaluator,

    Q. Zheng, Z. Tu, X. Zeng, A. C. Bovik, and Y . Fan, “A completely blind video quality evaluator,”IEEE Signal Processing Letters, vol. 29, pp. 2228–2232, 2022

  46. [54]

    Rapique: Rapid and accurate video quality prediction of user generated content,

    Z. Tu, X. Yu, Y . Wang, N. Birkbeck, B. Adsumilli, and A. C. Bovik, “Rapique: Rapid and accurate video quality prediction of user generated content,”IEEE Open Journal of Signal Processing, vol. 2, pp. 425–440, 2021

  47. [55]

    A deep learning based no- reference quality assessment model for ugc videos,

    W. Sun, X. Min, W. Lu, and G. Zhai, “A deep learning based no- reference quality assessment model for ugc videos,” inProceedings of the 30th ACM International Conference on Multimedia, 2022, pp. 856– 865

  48. [56]

    Ptm-vqa: efficient video quality assessment leveraging diverse pretrained models from the wild,

    K. Yuan, H. Liu, M. Li, M. Sun, M. Sun, J. Gong, J. Hao, C. Zhou, and Y . Tang, “Ptm-vqa: efficient video quality assessment leveraging diverse pretrained models from the wild,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 2835...

  49. [57]

    A convnet for the 2020s,

    Z. Liu, H. Mao, C.-Y . Wu, C. Feichtenhofer, T. Darrell, and S. Xie, “A convnet for the 2020s,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 11 976–11 986

  50. [58]

    Ava: A large-scale database for aesthetic visual analysis,

    N. Murray, L. Marchesotti, and F. Perronnin, “Ava: A large-scale database for aesthetic visual analysis,” in2012 IEEE conference on computer vision and pattern recognition. IEEE, 2012, pp. 2408–2415

  51. [59]

    Adam: A method for stochastic optimization,

    D. P. Kingma, “Adam: A method for stochastic optimization,”arXiv preprint arXiv:1412.6980, 2014

  52. [60]

    Towards understanding con- vergence and generalization of adamw,

    P. Zhou, X. Xie, Z. Lin, and S. Yan, “Towards understanding con- vergence and generalization of adamw,”IEEE transactions on pattern analysis and machine intelligence, 2024

  53. [61]

    Learning a parametric embedding by preserving local structure,

    L. Van Der Maaten, “Learning a parametric embedding by preserving local structure,” inArtificial intelligence and statistics. PMLR, 2009, pp. 384–391

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.