Pith. sign in

REVIEW 4 major objections 7 minor 1 cited by

Embodied Image Quality Assessment for Robotic Intelligence

T0 review · 4 major / 7 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read The paper claims that an embodied robot's image quality is defined by how well the image supports its task, not by human visual appeal, and that the two views are nearly uncorrelated.

desk verdict EPD is a genuinely new resource and the low human-robot correlation is worth taking seriously, but the reward-based label definition and missing cross-policy reliability check mean the central divergence claim is not yet nailed down. read the letter →

arxiv 2412.18774 v3 pith:A5QI5IHT submitted 2024-12-25 cs.CV

Jianbo Zhang , Chunyi Li , Jie Hao , Jun Jia , Huiyu Duan , Guoquan Zheng , Liang Yuan , Guangtao Zhai This is my paper · ORCID
classification cs.CV
keywords imagequalityassessmentembodiedAIroboticperceptionreinforcementlearningrewardno-referencebenchmarkdatasethumanvisualsystemmanipulation
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

The paper tries to establish that image quality for an embodied robot is a task-dependent property: a distorted image is 'good' to the extent that a robot acting on it still earns high task reward, and this is not the same as what a human finds attractive. To make this concrete, it builds the Embodied Preference Database (EPD), 12,500 distorted images from robotic-arm episodes, each labeled by the normalized cumulative reward the robot accumulated while performing push and pick tasks with that image as input. A human-rating study on the same images gives a correlation of only 0.2116 with the robot labels, and sixteen human-centered IQA algorithms perform poorly on EPD, which the paper reads as evidence that human-visual-system metrics are not valid proxies for robotic perception. It also proposes MA-EIQA, a lightweight no-reference model with multi-scale feature fusion and channel-spatial attention, and reports it as the best performer on EPD.

What carries the argument

The carrying mechanism is the EPD annotation protocol: a first-frame RGB image from a robotic-arm episode is corrupted by one of 25 distortions at five intensity levels, the episode is then executed by reinforcement-learning policies, and the normalized cumulative reward of the episode is attached to that image as its embodied DMOS. This protocol converts 'quality for the robot' into a measurable scalar without any human opinion. The paper's model, MA-EIQA, is a no-reference network built from a pretrained residual backbone, a bidirectional multi-scale feature encoder that fuses high-level semantics with low-level texture, and an attention module that weights channels and spatial regions; it regresses the reward-derived score with mean squared error loss.

What would settle it

Run many episodes that start from the exact same distorted first frame, changing only the random seed: if the spread in accumulated rewards across seeds is as large as the spread across distortion levels, the EPD scores are not stable properties of the image and the claimed human-robot gap would need to be recomputed from per-image medians.

Watch

Extended reading notes

Core claim

On the paper's terms, the discovery is that quality assessment of embodied images is different from that of humans. Robots and humans assign nearly unrelated quality scores to the same distorted images, with PLCC values of 0.2116 overall, 0.1778 on the push subtask, and 0.1297 on the pick subtask; existing IQA methods designed for the human visual system, even deep Transformer-based ones, plateau below 0.6 SRCC/PLCC on the EPD benchmark. The paper explains the gap by the different feature budgets of the two systems: embodied agents depend on low-level texture, edge, and structural information needed for manipulation, while human judgment keys on semantic content and aesthetics. It then introduces MA-EIQA, a lightweight no-reference network that fuses multi-scale features and applies channel-spatial attention, achieving higher correlation with the reward-derived labels than any of the compared human-centered methods.

Load-bearing premise

The whole benchmark depends on treating the total reward a robot earns during a 50-step episode as a fair quality score for the single starting image; if that reward is driven mainly by chance variation in the robot's behavior or by the simulator rather than by what is visible in that one frame, the labels measure how robust the robot is to damage, not image quality.

Editorial extensions

If this is right

  • Human-centered IQA metrics should not be treated as proxies for robotic perception; any robot system that uses them to filter or select images may be optimizing the wrong quantity.
  • No-reference models are a better starting point for embodied quality than full-reference ones: on EPD they outperform reference-based methods, consistent with the idea that the robot's own task defines quality rather than comparison with a pristine image.
  • Lightweight CNN designs can compete with much larger Transformer models on this quality axis, which matters for onboard, real-time use in robots with limited compute.
  • EPD-style reward labels give a robot a training signal to predict, before acting, which images will hurt its task, enabling selective restoration, reacquisition, or refusal.
  • Because the human-robot correlation is so low, embodied-AI benchmarks that evaluate perception should include task-derived image quality labels rather than relying on human subjective ratings.

Reading between the lines

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

  • If task reward is the quality axis, then 'good image' becomes policy- and task-dependent: the same frame could be high-quality for one robot policy and low-quality for another, so future databases may need per-policy scores rather than a single averaged MOS.
  • A natural extension the paper does not test is to close the loop: have the robot predict embodied quality online and trigger restoration, replanning, or a request for a new viewpoint before executing a manipulation.
  • The divergence claim should be stress-tested beyond two box-manipulation tasks in one simulator; navigation, locomotion, and real-robot experiments with different cameras and policies would show whether the 0.21 correlation is a general property of embodied vision or an artifact of this task family.
  • An explicit prediction follows from the paper's low-level-feature account: distortions that destroy global semantics but preserve local edges and textures should hurt robot rewards less than they hurt human scores, while the opposite should hold for noise that destroys structure; paired human-and-robot rating experiments could test this directly.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 7 minor

Summary. The paper introduces EPD, an 'Embodied Preference Database' for image quality assessment (IQA) in robotic settings, with 12,500 distorted-image annotations obtained by running RL policies (PPO, SAC, TDMPC2) on push and pick tasks in SAPIEN and attaching the episode reward as the quality score. It also proposes MA-EIQA, a no-reference IQA network combining a multi-scale feature encoder with a CBAM-style attention module, and benchmarks 16 existing IQA methods on EPD. The central claim is that embodied-image quality assessment differs from human perception, supported primarily by a low measured correlation (PLCC 0.2116) between robot reward-derived scores and human MOS. The paper also includes a real-world UR5 experiment and an ablation study.

Significance. If the reward-derived labels are accepted as a valid robot-centric quality axis, the paper provides a useful new resource and a plausible demonstration that human-oriented IQA metrics are poor predictors of robotic task-relevant image quality. The EPD dataset, the benchmark of 16 methods, the released code, and the lightweight MA-EIQA model are concrete contributions. It is also to the paper's credit that the low human-robot PLCC is an external benchmark measurement, so the headline correlation is not purely circular. However, the interpretation of that correlation as evidence of a distinct embodied quality axis depends on a load-bearing premise that the RL episode reward is a stable, image-attributable quality label; this premise is not yet established, and the paper's scope is also narrower than the title suggests.

major comments (4)
  1. [Section III-B, Eqs. (1)-(3); Table III] The robot quality label is defined as the cumulative RL episode reward, which is a property of a particular policy interacting with a dynamical environment, not an intrinsic property of the distorted image. A policy trained on clean images and evaluated under a consistent corruption may fail because of OOD policy fragility, exploration noise, or simulator dynamics rather than because the first-frame image loses task-relevant information. This makes the low PLCC of 0.2116 against human MOS (Fig. 5) substantially entailed by the label construction. To support the claim that a distinct embodied quality axis exists, the paper should demonstrate that reward-derived labels are stable across independently trained policies and correlate with task-relevant image information. Table III reports only distortion-category means for six agents; the missing per-image agreement across those agents (e.g., per-image ICC or rank correlations) is the key evidence needed. Without it, the low human-robot correlation could be an attenuated correlation between policy-specific robustness measurements rather than evidence of a robot-specific image quality axis.
  2. [Section III-A; Section V-A] Section III-A states that 'only the image of the initial frame is selected as the image to be evaluated,' while the reward label is accumulated over an episode of 50 action steps (Section V-A). Because the distortion is applied consistently across all frames, the episode reward is a legitimate label for the distortion condition, but it is not a label for the specific initial-frame image that is scored. The dataset is described as containing '12,500 distorted image annotations,' yet the annotation is episode-level. The paper should either validate that reward scores are insensitive to which frame is selected, or reformulate EPD as containing condition-level annotations attached to representative first frames. This distinction matters for the central claim because the correlation with human MOS is computed on single images.
  3. [Table I; Section V-A; Table IV] All benchmark results in Table I are reported without uncertainty, despite the fact that the train/val split is a single 8:2 partition and several reported differences are small (e.g., MA-EIQA exceeds MANIQA by 2.29% SRCC on All Tasks). Table IV reports ablation means over 10 runs but does not report standard deviations or significance tests, and its caption incorrectly says 'Comparison of 14 IQA methods' when it compares four model variants. The state-of-the-art claim for MA-EIQA and the component contributions from the multi-scale encoder and attention module cannot be assessed without confidence intervals, multiple seeds, or paired significance tests. Given that the central claim rests on the divergence between robot and human quality assessment, the benchmark table should also include bootstrap or permutation-based intervals for the low PLCC values in Fig. 5, since the current 95% prediction intervals describe the fitting curve, not the uncertainty of the correlation estimate.
  4. [Section III-C; Fig. 5] The human subjective experiment used to obtain MOS is described only as involving '15 experienced experts.' No information is provided about viewing conditions, display device, viewing distance, rating scale, instruction, outlier rejection, or inter-observer agreement, and no per-image variance of the human MOS is reported. Since the headline result is the low correlation between robot scores and this human MOS, the reliability of the human labels is load-bearing. The paper should report standard IQA subjective-testing details and at least one inter-observer consistency measure (e.g., average SRCC between each expert and the mean) so that an attenuated PLCC can be separated from noisy human labels or an unreliable protocol.
minor comments (7)
  1. [Abstract; Section I] The abstract contains a typo: 'does the the image quality of Robot-Generated Content' should be 'does the image quality.' The term 'Moravec paradox' is invoked but never explained or cited; either define it or remove the reference.
  2. [Section IV heading] The heading 'PROPOSED METHED' should read 'PROPOSED METHOD.'
  3. [Section II-A] The simulator name is misspelled as 'Issac Sim' in 'The Issac Sim [39]'; it should be 'Isaac Sim.'
  4. [Section III-B; Fig. 3; Fig. 5] Terminology is inconsistent: the text introduces 'embodied Differential Mean Opinion Score (DMOS)' and later uses 'EMOS' in Fig. 5, which is not defined. The normalized score should be given one name and one definition throughout.
  5. [Section III-B, Eq. (3)] The TDMPC2 reward in Eq. (3) contains an unspecified weight factor lambda, and no value or sensitivity analysis is given. If the final EPD labels depend on this hyperparameter, its value should be reported.
  6. [Section V-E; Fig. 9] The real-world experiment is reported as a set of illustrative cases with no systematic quantitative evaluation, sample-size information, or comparison against a baseline. If it is intended as evidence for the human-robot divergence, it needs a protocol and results table; otherwise it should be framed explicitly as a qualitative demonstration.
  7. [General scope] The title and abstract claim a general conclusion about 'embodied image quality assessment,' but the dataset covers only two fixed-base robotic arm tasks in a single simulator. The paper acknowledges this in Section II-B, but the abstract and conclusion should carry a qualifier such as 'for the tested manipulation tasks' to avoid overgeneralization.

Circularity Check

1 steps flagged · score 4.0 of 10

Robot 'image quality' is defined as RL episode reward, so the claimed human-robot divergence is partly built into the label; the external PLCC measurement keeps it from being fully circular.

  1. self definitional [Sec. III-B 'Preference Score Collection' (Eq. 1-3); Sec. III-C 'Database Analysis' (Fig. 5); Abstract]
    "'The reward value of each episodes is adopted as the performance score for evaluating the robot performance in the process of perceiving the environment and completing the embodied intelligence task.' ... 'Each episodes receives a reward corresponding to the score of the image, and finally the embodied Differential Mean Opinion Score (DMOS) are normalised to the range (0, 5).'"

    The paper stipulates that the robot's 'image score' is the RL episode reward, so 'embodied preference' is, by construction, a task-performance score rather than an independent perceptual-quality judgment. The abstract's central conclusion that 'quality assessment of embodied images is different from that of humans' is then tested by correlating this reward-derived DMOS with human MOS (PLCC 0.2116). The low correlation is an external empirical measurement, so the divergence is not logically forced; however, the construct being compared is the paper's own definitional choice, making the headline claim substantially an entailment of the label definition rather than a discovery of an independent robot quality axis.

full rationale

The derivation chain is largely self-contained: the EPD labels come from simulator rollouts, the human MOS is an external subjective experiment, and the benchmark comparisons in Table I use standard train/val splits. The single circular burden is conceptual. The paper defines embodied image quality as the RL episode reward (Sec. III-B: 'The reward value of each episodes is adopted as the performance score... Each episodes receives a reward corresponding to the score of the image'), so the abstract's claim that 'quality assessment of embodied images is different from that of humans' is substantially an entailment of that label choice. The measured PLCC of 0.2116 against human MOS is an external, falsifiable quantity, and the MA-EIQA model evaluation is not circular, which keeps the score moderate rather than severe. No self-citation chain, imported uniqueness theorem, or fitted-parameter-as-prediction pattern is present.

Assumptions & free parameters 4 free parameters · 4 assumptions · 1 invented entities

The central claim rests on the equation of task reward with image quality, on the representativeness of a single first frame, and on the generality of two simulated manipulation tasks. These are domain assumptions, not derived results. No new physical entities are introduced; the 'Embodied DMOS' label is a new construct whose validity is assumed.

free parameters (4)
  • Reward weight lambda (TDMPC2) = not reported
    Introduced in Eq. 3 to combine immediate reward with goal distance; its value is not stated and it shapes the label for one of the three labeling policies.
  • DMOS normalization mapping = 0-5 endpoints; mapping unspecified
    Sec III-B normalizes episode rewards into (0,5) without specifying the mapping; all dataset scores and downstream benchmarks inherit this choice.
  • Distortion intensity levels = 5 levels per distortion, parameters unspecified
    Sec III-A sets 25 distortions at 5 intensities but gives no parameter values; the label distribution (Table II) is a function of these hand-chosen levels.
  • Train/val split = 8:2
    Sec V-A: the split ratio is stated but scene-disjointness is not; if the same scenes appear in train and val, reported correlations are inflated.
assumptions (4)
  • domain assumption RL episode reward is a valid proxy for robot image quality
    Sec III-B defines the preference score as the episode reward; if rewards instead measure policy robustness or exploration noise, the labels are not quality scores.
  • ad hoc to paper One first frame stands for the quality of the whole episode input
    Sec III-A: only the initial frame is labeled, while the distortion is applied consistently across the episode and reward accumulates over 50 steps.
  • domain assumption Clean-trained policies in SAPIEN represent embodied robots generally
    Sec III-B and V-A: two box tasks, three RL algorithms, and one simulator ground conclusions stated for embodied AI and the Robot Visual System.
  • domain assumption Human MOS from 15 experts is a reliable reference
    Sec III-C: no rating procedure, number of ratings per image, or expert screening is reported; the PLCC 0.2116 depends on this unverified reference.
invented entities (1)
  • Embodied DMOS (robot preference score)
    purpose: Quality label for the EPD database, defined as normalized RL task reward
    A new label construct introduced in Sec III-B; apart from the eight-case real-world demo (Sec V-E), no external robot-behavior benchmark validates that it measures image quality rather than policy fragility.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Embodied Image Quality Assessment for Robotic Intelligence." pith.science (2026). https://pith.science/paper/A5QI5IHT

@misc{pith2026241218774,
  author       = {Pith},
  title        = {Pith review of: Embodied Image Quality Assessment for Robotic Intelligence},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/A5QI5IHT}},
  note         = {Machine review of arXiv:2412.18774}
}
read the original abstract

Image Quality Assessment (IQA) of User-Generated Content (UGC) is a critical technique for human Quality of Experience (QoE). However, does the the image quality of Robot-Generated Content (RGC) demonstrate traits consistent with the Moravec paradox, potentially conflicting with human perceptual norms? Human subjective scoring is more based on the attractiveness of the image. Embodied agent are required to interact and perceive in the environment, and finally perform specific tasks. Visual images as inputs directly influence downstream tasks. In this paper, we explore the perception mechanism of embodied robots for image quality. We propose the first Embodied Preference Database (EPD), which contains 12,500 distorted image annotations. We establish assessment metrics based on the downstream tasks of robot. In addition, there is a gap between UGC and RGC. To address this, we propose a novel Multi-scale Attention Embodied Image Quality Assessment called MA-EIQA. For the proposed EPD dataset, this is the first no-reference IQA model designed for embodied robot. Finally, the performance of mainstream IQA algorithms on EPD dataset is verified. The experiments demonstrate that quality assessment of embodied images is different from that of humans. We sincerely hope that the EPD can contribute to the development of embodied AI by focusing on image quality assessment. The benchmark is available at https://github.com/Jianbo-maker/EPD_benchmark.

Figures

Figures reproduced from arXiv: 2412.18774 by the authors.

Figure 1
Figure 1. Input image quality is assessed based on the performance of the robot [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The construction overview of the proposed EPD Benchmark. The EPD benchmark is oriented towards embodied AI and includes 25 common image [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. DMOS of the EPD, visualized in 25 distortion subsets. Different color denotes distortion strength [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: The dataset analyses of EPD. (a) The correlation matrix for MOS. It [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Scatter plots and polynomial fitted curves of embodied MOS (EMOS) [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 6
Figure 6. Figure 6: Framework of the proposed method. MA-EIQA is the first embodied No-Reference IQA model, which consists of two components: multi-scale feature [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]
Figure 7
Figure 7. Figure 7: The different scenes that the benchmark can be supported. The EPD support push and pick two major types of tasks. A wide range of different objects [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: The first and third columns are the reference images, and the second [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 9
Figure 9. Figure 9: Cases of real-world experiment. The input for the robot is first-person view images with various types of distortion. The third-person perspective [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. RGC-VQA: An Exploration Database for Robotic-Generated Video Quality Assessment

    cs.CV 2025-06 conditional novelty 6.0 of 10

    A 2,100-video database with human opinions shows that current video quality models underperform on robot-generated content, motivating a new VQA subfield.

Reference graph

Works this paper leans on

65 extracted references · 44 canonical work pages · cited by 1 Pith paper

  1. [1]

    A survey on image quality assessment: Insights, analysis, and future outlook,

    C. Ma, Z. Shi, Z. Lu, S. Xie, F. Chao, and Y . Sui, “A survey on image quality assessment: Insights, analysis, and future outlook,”arXiv preprint arXiv:2502.08540, 2025

    arXiv 2025

  2. [2]

    No-reference image quality assessment: Obtain mos from image quality score distribution,

    Y . Gao, X. Min, Y . Cao, X. Liu, and G. Zhai, “No-reference image quality assessment: Obtain mos from image quality score distribution,” IEEE Transactions on Circuits and Systems for Video Technology , vol. 35, no. 2, pp. 1840–1854, 2025

    work page 2025

  3. [3]

    Study of subjective and objective naturalness assessment of ai-generated images,

    Z. Chen, W. Sun, H. Wu, Z. Zhang, J. Jia, R. Huang, X. Min, G. Zhai, and W. Zhang, “Study of subjective and objective naturalness assessment of ai-generated images,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 35, no. 4, pp. 3573–3588, 2025

    work page 2025

  4. [4]

    Single image dehazing using fuzzy region segmentation and haze density decomposition,

    T. Li, Y . Liu, W. Ren, B. Shiri, and W. Lin, “Single image dehazing using fuzzy region segmentation and haze density decomposition,” IEEE Transactions on Circuits and Systems for Video Technology , pp. 1–1, 2025

    work page 2025

  5. [5]

    Dual degradation representation for joint deraining and low-light enhancement in the dark,

    X. Lin, J. Yue, S. Ding, C. Ren, L. Qi, and M.-H. Yang, “Dual degradation representation for joint deraining and low-light enhancement in the dark,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 35, no. 3, pp. 2461–2473, 2025

    work page 2025

  6. [6]

    Omni-deblurring: Capturing omni-range context for image deblurring,

    Y . Li, H. An, T. Zhang, X. Chen, B. Jiang, and J. Pan, “Omni-deblurring: Capturing omni-range context for image deblurring,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 35, no. 8, pp. 7543– 7553, 2025

    work page 2025

  7. [7]

    A reference-free quality enhancement framework for low-quality fundus images,

    Q. Hou, Y . Wang, L. Lan, P. Cao, J. Yang, X. Liu, M. Wang, Y . C. Tham, and O. R. Zaiane, “A reference-free quality enhancement framework for low-quality fundus images,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 35, no. 4, pp. 3541–3556, 2025. 12

    work page 2025

  8. [8]

    Synergetic assess- ment of quality and aesthetic: Approach and comprehensive benchmark dataset,

    K. Zhang, D. Zhu, X. Min, Z. Gao, and G. Zhai, “Synergetic assess- ment of quality and aesthetic: Approach and comprehensive benchmark dataset,” IEEE Transactions on Circuits and Systems for Video Technol- ogy, vol. 34, no. 4, pp. 2536–2549, 2024

    work page 2024

Show all 65 references
  1. [9]

    Image quality assess- ment: from error visibility to structural similarity,

    Z. Wang, A. Bovik, H. Sheikh, and E. Simoncelli, “Image quality assess- ment: from error visibility to structural similarity,” IEEE Transactions on Image Processing , vol. 13, no. 4, pp. 600–612, 2004

    2004

  2. [10]

    Visual importance pooling for image quality assessment,

    A. K. Moorthy and A. C. Bovik, “Visual importance pooling for image quality assessment,” IEEE Journal of Selected Topics in Signal Processing, vol. 3, no. 2, pp. 193–201, 2009

    2009

  3. [11]

    Information content weighting for perceptual image quality assessment,

    Z. Wang and Q. Li, “Information content weighting for perceptual image quality assessment,” IEEE Transactions on Image Processing , vol. 20, no. 5, pp. 1185–1198, 2011

    2011

  4. [12]

    Fsim: A feature similarity index for image quality assessment,

    L. Zhang, L. Zhang, X. Mou, and D. Zhang, “Fsim: A feature similarity index for image quality assessment,” IEEE Transactions on Image Processing, vol. 20, no. 8, pp. 2378–2386, 2011

    2011

  5. [13]

    Gradient magnitude similarity deviation: A highly efficient perceptual image quality index,

    W. Xue, L. Zhang, X. Mou, and A. C. Bovik, “Gradient magnitude similarity deviation: A highly efficient perceptual image quality index,” IEEE Transactions on Image Processing , vol. 23, no. 2, pp. 684–695, 2014

    2014

  6. [14]

    Deep learning of human visual sensitivity in image quality assessment framework,

    J. Kim and S. Lee, “Deep learning of human visual sensitivity in image quality assessment framework,” in 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , 2017, pp. 1969–1977

    2017

  7. [15]

    Deep neural networks for no-reference and full-reference image quality assessment,

    S. Bosse, D. Maniry, K.-R. M ¨uller, T. Wiegand, and W. Samek, “Deep neural networks for no-reference and full-reference image quality assessment,” IEEE Transactions on Image Processing, vol. 27, no. 1, pp. 206–219, 2018

    2018

  8. [16]

    A psychovisual quality metric in free-energy principle,

    G. Zhai, X. Wu, X. Yang, W. Lin, and W. Zhang, “A psychovisual quality metric in free-energy principle,” IEEE Transactions on Image Processing, vol. 21, no. 1, pp. 41–52, 2012

    2012

  9. [17]

    Blind image quality assessment: From natural scene statistics to perceptual quality,

    A. K. Moorthy and A. C. Bovik, “Blind image quality assessment: From natural scene statistics to perceptual quality,” IEEE Transactions on Image Processing , vol. 20, no. 12, pp. 3350–3364, 2011

    2011

  10. [18]

    Blind image quality assessment based on high order statistics aggregation,

    J. Xu, P. Ye, Q. Li, H. Du, Y . Liu, and D. Doermann, “Blind image quality assessment based on high order statistics aggregation,” IEEE Transactions on Image Processing, vol. 25, no. 9, pp. 4444–4457, 2016

    2016

  11. [19]

    Metaiqa: Deep meta- learning for no-reference image quality assessment,

    H. Zhu, L. Li, J. Wu, W. Dong, and G. Shi, “Metaiqa: Deep meta- learning for no-reference image quality assessment,” in 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020

    2020

  12. [20]

    Convolutional neural networks for no-reference image quality assessment,

    L. Kang, P. Ye, Y . Li, and D. Doermann, “Convolutional neural networks for no-reference image quality assessment,” in 2014 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , 2014

    2014

  13. [21]

    A statistical evaluation of recent full reference image quality assessment algorithms,

    H. Sheikh, M. Sabir, and A. Bovik, “A statistical evaluation of recent full reference image quality assessment algorithms,” IEEE Transactions on Image Processing , vol. 15, no. 11, pp. 3440–3451, 2006

    2006

  14. [22]

    Kadid-10k: A large-scale artificially distorted iqa database,

    H. Lin, V . Hosu, and D. Saupe, “Kadid-10k: A large-scale artificially distorted iqa database,” in 2019 Eleventh International Conference on Quality of Multimedia Experience (QoMEX) , 2019, pp. 1–3

    2019

  15. [23]

    Massive online crowdsourced study of subjective and objective picture quality,

    D. Ghadiyaram and A. C. Bovik, “Massive online crowdsourced study of subjective and objective picture quality,”IEEE Transactions on Image Processing, vol. 25, no. 1, pp. 372–387, 2016

    2016

  16. [24]

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

    V . Hosu, H. Lin, T. Sziranyi, and D. Saupe, “Koniq-10k: An ecologically valid database for deep learning of blind image quality assessment,” IEEE Transactions on Image Processing, vol. 29, pp. 4041–4056, 2020

    2020

  17. [25]

    Q-bench ++: A benchmark for multi-modal foundation models on low-level vision from single images to pairs,

    Z. Zhang, H. Wu, E. Zhang, G. Zhai, and W. Lin, “Q-bench ++: A benchmark for multi-modal foundation models on low-level vision from single images to pairs,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 46, no. 12, pp. 10 404–10 418, 2024

    2024

  18. [26]

    Agiqa-3k: An open database for ai-generated image quality assessment,

    C. Li, Z. Zhang, H. Wu, W. Sun, X. Min, X. Liu, G. Zhai, and W. Lin, “Agiqa-3k: An open database for ai-generated image quality assessment,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 34, no. 8, pp. 6833–6846, 2024

    2024

  19. [27]

    A-bench: Are lmms masters at evaluating ai- generated images?

    Z. Zhang, H. Wu, C. Li, Y . Zhou, W. Sun, X. Min, Z. Chen, X. Liu, W. Lin, and G. Zhai, “A-bench: Are lmms masters at evaluating ai- generated images?” arXiv preprint arXiv:2406.03070 , 2024

    2024 arXiv

  20. [28]

    Cmc-bench: Towards a new paradigm of visual signal compression,

    C. Li, X. Wu, H. Wu, D. Feng, Z. Zhang, G. Lu, X. Min, X. Liu, G. Zhai, and W. Lin, “Cmc-bench: Towards a new paradigm of visual signal compression,” arXiv preprint arXiv:2406.09356 , 2024

    2024 arXiv

  21. [29]

    R-bench: Are your large multimodal model robust to real-world corruptions?

    C. Li, J. Zhang, Z. Zhang, H. Wu, Y . Tian, W. Sun, G. Lu, X. Min, X. Liu, W. Lin, X.-P. Zhang, and G. Zhai, “R-bench: Are your large multimodal model robust to real-world corruptions?” IEEE Journal of Selected Topics in Signal Processing , pp. 1–16, 2025

    2025

  22. [30]

    Aligning cyber space with physical world: A comprehensive survey on embodied ai,

    Y . Liu, W. Chen, Y . Bai, X. Liang, G. Li, W. Gao, and L. Lin, “Aligning cyber space with physical world: A comprehensive survey on embodied ai,” IEEE/ASME Transactions on Mechatronics , pp. 1–22, 2025

    2025

  23. [31]

    Real-time recognition of team be- haviors by multisensory graph-embedded robot learning,

    B. Reily, P. Gao, and W. H. Zhang, “Real-time recognition of team be- haviors by multisensory graph-embedded robot learning,” International Journal of Robotics Research , vol. 41, no. 8, pp. 798–811, 2022

    2022

  24. [32]

    Analysis of an all- terrain tracked robot with innovative suspension system,

    A. Ugenti, R. Galati, G. Mantriota, and G. Reina, “Analysis of an all- terrain tracked robot with innovative suspension system,” Mechanism and Machine Theory , vol. 182, pp. 105 237–, 2023

    2023

  25. [33]

    Swarm of micro flying robots in the wild,

    X. Zhou, X. Wen, Z. Wang, Y . Gao, H. Li, Q. Wang, T. Yang, H. Lu, Y . Cao, and C. Xu, “Swarm of micro flying robots in the wild,” Science robotics, vol. 7, no. 66, p. eabm5954, 2022

    2022

  26. [34]

    Scientific exploration of chal- lenging planetary analog environments with a team of legged robots,

    P. Arm, G. Waibel, J. Preisig, T. Tuna, R. Zhou, V . Bickel, G. Ligeza, T. Miki, F. Kehl, and H. Kolvenbach, “Scientific exploration of chal- lenging planetary analog environments with a team of legged robots,” Science Robotics, vol. 8, no. 80, 2023

    2023

  27. [35]

    Advancements in humanoid robots: A comprehensive review and future prospects,

    Y . Tong, H. Liu, and Z. Zhang, “Advancements in humanoid robots: A comprehensive review and future prospects,” IEEE/CAA Journal of Automatica Sinica, vol. 11, no. 2, pp. 301–328, 2024

    2024

  28. [36]

    Implicit coordination for 3d underwater collective behaviors in a fish-inspired robot swarm,

    F. Berlinger, M. Gauci, and R. Nagpal, “Implicit coordination for 3d underwater collective behaviors in a fish-inspired robot swarm,” Science Robotics, vol. 6, no. 50, p. eabd8668, 2021

    2021

  29. [37]

    The franka emika robot: A reference platform for robotics research and education,

    S. Haddadin, S. Parusel, L. Johannsmeier, S. Golz, S. Gabl, F. Walch, M. Sabaghian, C. J ¨ahne, L. Hausperger, and S. Haddadin, “The franka emika robot: A reference platform for robotics research and education,” IEEE Robotics & Automation Magazine, vol. 29, no. 2, pp. 46–64, 2022

    2022

  30. [38]

    Bas optimized elm for kuka iiwa robot learning,

    C. Li, S. Zhu, Z. Sun, and J. Rogers, “Bas optimized elm for kuka iiwa robot learning,” IEEE Transactions on Circuits and Systems II: Express Briefs, vol. 68, no. 6, pp. 1987–1991, 2021

    1987

  31. [39]

    Isaac gym: High performance gpu-based physics simulation for robot learning,

    V . Makoviychuk, L. Wawrzyniak, Y . Guo, M. Lu, K. Storey, M. Macklin, D. Hoeller, N. Rudin, A. Allshire, A. Handa, and G. State, “Isaac gym: High performance gpu-based physics simulation for robot learning,” arXiv preprint arXiv:2108.10470 , 2021

    2021 arXiv

  32. [40]

    Design and use paradigms for gazebo, an open-source multi-robot simulator,

    N. Koenig and A. Howard, “Design and use paradigms for gazebo, an open-source multi-robot simulator,” in 2004 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) (IEEE Cat. No.04CH37566), vol. 3, 2004, pp. 2149–2154 vol.3

    2004

  33. [41]

    Pybullet, a python module for physics simulation for games, robotics and machine learning,

    E. Coumans and Y . Bai, “Pybullet, a python module for physics simulation for games, robotics and machine learning,” 2016. [Online]. Available: http://pybullet.org

    2016

  34. [42]

    Ai2-thor: An interactive 3d environment for visual ai,

    E. Kolve, R. Mottaghi, W. Han, E. VanderBilt, L. Weihs, A. Herrasti, M. Deitke, K. Ehsani, D. Gordon, Y . Zhu, A. Kembhavi, A. Gupta, and A. Farhadi, “Ai2-thor: An interactive 3d environment for visual ai,” arXiv preprint arXiv:1712.05474 , 2022

    2022 arXiv

  35. [43]

    Habitat: A platform for embodied ai research,

    M. Savva, A. Kadian, O. Maksymets, Y . Zhao, E. Wijmans, B. Jain, J. Straub, J. Liu, V . Koltun, J. Malik, D. Parikh, and D. Batra, “Habitat: A platform for embodied ai research,” in 2019 IEEE/CVF International Conference on Computer Vision (ICCV) , 2019, pp. 9338–9346

    2019

  36. [44]

    Sapien: A simulated part-based interactive environment,

    F. Xiang, Y . Qin, K. Mo, Y . Xia, H. Zhu, F. Liu, M. Liu, H. Jiang, Y . Yuan, H. Wang, L. Yi, A. X. Chang, L. J. Guibas, and H. Su, “Sapien: A simulated part-based interactive environment,” in 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020, p...

    2020

  37. [45]

    Prox- imal policy optimization algorithms,

    J. Schulman, F. Wolski, P. Dhariwal, A. Radford, and O. Klimov, “Prox- imal policy optimization algorithms,” arXiv preprint arXiv:1707.06347, 2017

    2017 arXiv

  38. [46]

    Soft actor-critic: Off- policy maximum entropy deep reinforcement learning with a stochastic actor,

    T. Haarnoja, A. Zhou, P. Abbeel, and S. Levine, “Soft actor-critic: Off- policy maximum entropy deep reinforcement learning with a stochastic actor,” in Proceedings of the 35th International Conference on Machine Learning, ser. Proceedings of Machine Learning Research, J. Dy a...

    2018

  39. [47]

    Td-mpc2: Scalable, robust world models for continuous control,

    N. Hansen, H. Su, and X. Wang, “Td-mpc2: Scalable, robust world models for continuous control,” arXiv preprint arXiv:2310.16828, 2023

    2023 arXiv

  40. [48]

    Maniskill3: Gpu parallelized robotics simulation and rendering for generalizable embodied ai,

    S. Tao, F. Xiang, A. Shukla, Y . Qin, X. Hinrichsen, X. Yuan, C. Bao, X. Lin, Y . Liu, T. kai Chan, Y . Gao, X. Li, T. Mu, N. Xiao, A. Gurha, Z. Huang, R. Calandra, R. Chen, S. Luo, and H. Su, “Maniskill3: Gpu parallelized robotics simulation and rendering for generalizable em...

    arXiv

  41. [49]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016, pp. 770–778

    2016

  42. [50]

    Path aggregation network for instance segmentation,

    S. Liu, L. Qi, H. Qin, J. Shi, and J. Jia, “Path aggregation network for instance segmentation,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , June 2018

    2018

  43. [51]

    Cbam: Convolutional block attention module,

    S. Woo, J. Park, J.-Y . Lee, and I. S. Kweon, “Cbam: Convolutional block attention module,” in Proceedings of the European Conference on Computer Vision (ECCV) , September 2018

    2018

  44. [52]

    Attentions help cnns see better: Attention-based hybrid image quality 13 assessment network,

    S. Lao, Y . Gong, S. Shi, S. Yang, T. Wu, J. Wang, W. Xia, and Y . Yang, “Attentions help cnns see better: Attention-based hybrid image quality 13 assessment network,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (CVPR) , 2022, pp. 1140–1149

    2022

  45. [53]

    Pieapp: Perceptual image- error assessment through pairwise preference,

    E. Prashnani, H. Cai, Y . Mostofi, and P. Sen, “Pieapp: Perceptual image- error assessment through pairwise preference,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , 2018, pp. 1808–1817

    2018

  46. [54]

    Learning conditional knowledge distillation for degraded-reference image quality assessment,

    H. Zheng, H. Yang, J. Fu, Z.-J. Zha, and J. Luo, “Learning conditional knowledge distillation for degraded-reference image quality assessment,” in Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2021, pp. 10 242–10 251

    2021

  47. [55]

    Image quality assess- ment: Unifying structure and texture similarity,

    K. Ding, K. Ma, S. Wang, and E. P. Simoncelli, “Image quality assess- ment: Unifying structure and texture similarity,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 44, no. 5, pp. 2567– 2581, 2022

    2022

  48. [56]

    The unreasonable effectiveness of deep features as a perceptual metric,

    R. Zhang, P. Isola, A. A. Efros, E. Shechtman, and O. Wang, “The unreasonable effectiveness of deep features as a perceptual metric,” in 2018 IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion (CVPR), 2018, pp. 586–595

    2018

  49. [57]

    Topiq: A top-down approach from semantics to distortions for image quality assessment,

    C. Chen, J. Mo, J. Hou, H. Wu, L. Liao, W. Sun, Q. Yan, and W. Lin, “Topiq: A top-down approach from semantics to distortions for image quality assessment,” IEEE Transactions on Image Processing , vol. 33, pp. 2404–2418, 2024

    2024

  50. [58]

    Perceptual image quality assessment with transformers,

    M. Cheon, S.-J. Yoon, B. Kang, and J. Lee, “Perceptual image quality assessment with transformers,” in 2021 IEEE/CVF Conference on Com- puter Vision and Pattern Recognition Workshops (CVPRW) , 2021, pp. 433–442

    2021

  51. [59]

    Exploring clip for assessing the look and feel of images,

    J. Wang, K. C. Chan, and C. C. Loy, “Exploring clip for assessing the look and feel of images,” in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 37, no. 2, 2023, pp. 2555–2563

    2023

  52. [60]

    Blind image quality assessment using a deep bilinear convolutional neural network,

    W. Zhang, K. Ma, J. Yan, D. Deng, and Z. Wang, “Blind image quality assessment using a deep bilinear convolutional neural network,” IEEE Transactions on Circuits and Systems for Video Technology , vol. 30, no. 1, pp. 36–47, 2020

    2020

  53. [61]

    Maniqa: Multi-dimension attention network for no-reference image quality assessment,

    S. Yang, T. Wu, S. Shi, S. Lao, Y . Gong, M. Cao, J. Wang, and Y . Yang, “Maniqa: Multi-dimension attention network for no-reference image quality assessment,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , 2022, pp. 1191– 1200

    2022

  54. [62]

    Blind image quality assessment via transformer predicted error map and perceptual quality token,

    J. Shi, P. Gao, and A. Smolic, “Blind image quality assessment via transformer predicted error map and perceptual quality token,” IEEE Transactions on Multimedia , vol. 26, pp. 4641–4651, 2024

    2024

  55. [63]

    Blindly assess image quality in the wild guided by a self-adaptive hyper network,

    S. Su, Q. Yan, Y . Zhu, C. Zhang, X. Ge, J. Sun, and Y . Zhang, “Blindly assess image quality in the wild guided by a self-adaptive hyper network,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (CVPR) , 2020, pp. 3667–3676

    2020

  56. [64]

    Quality-aware image- text alignment for opinion-unaware image quality assessment,

    L. Agnolucci, L. Galteri, and M. Bertini, “Quality-aware image- text alignment for opinion-unaware image quality assessment,” arXiv preprint arXiv:2403.11176, 2025

    2025 arXiv

  57. [65]

    IQA-PyTorch: Pytorch toolbox for image quality assessment,

    C. Chen and J. Mo, “IQA-PyTorch: Pytorch toolbox for image quality assessment,” [Online]. Available: https://github.com/chaofengc/ IQA-PyTorch, 2022

    2022

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.