Pith. sign in

REVIEW 3 major objections 6 minor 49 references

Light Field Image Quality Assessment With Auxiliary Learning Based on Depthwise and Anglewise Separable Convolutions

T0 review · 3 major / 6 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's ALAS-DADS metric estimates light field image quality with separable spatial-angular convolutions plus auxiliary hints, reporting 42.86% and 45.95% lower RMSE than the second-best metric on Win5-LID and SMART.

desk verdict A coherent, potentially useful architecture for no-reference LFIQA with strong reported numbers, but the headline gain rests on an unverified split-order relative to the 8x augmentation, and Table VII is internally inconsistent. read the letter →

arxiv 2412.07079 v1 pith:6AVPCG4W submitted 2024-12-10 eess.IV cs.CVcs.LG

Qiang Qu , Xiaoming Chen , Vera Chung , Zhibo Chen This is my paper · ORCID
classification eess.IVcs.CVcs.LG
keywords no-referencelightfieldimagequalityassessmentdepthwiseseparableconvolutionanglewiseauxiliarylearningnaturalscenestatisticsgradientdirectiondistributionofexperienceimaging
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

This paper proposes a no-reference light field image quality assessment (NR-LFIQA) metric called ALAS-DADS, aimed at predicting how viewers would rate distorted light field images without access to a pristine reference. The authors extend depthwise separable convolution to the spatial subviews of a light field (LF-DSC) and further factorize angular processing (LF-ASC), then train the model with two auxiliary tasks that estimate spatial natural-scene-statistics features and angular gradient-direction-distribution features as hints. On the Win5-LID and SMART datasets they report 42.86% and 45.95% smaller root-mean-square error than the second-best benchmarking metric, with SROCC of 0.9260 and 0.8540 respectively. The underlying claim is that explicitly separating spatial and angular feature extraction, together with auxiliary hints, is what drives the large improvement.

What carries the argument

The load-bearing objects are the two separable convolution variants and the auxiliary loss. LF-DSC treats each angular subview as an ordinary 2D image and factorizes a standard convolution into a depthwise convolution plus a pointwise convolution, cutting the per-subview cost from roughly $c_i c_j k_j^2$ to $c_i(c_j+k_j^2)$ multiply-accumulates. LF-ASC factorizes a full 4D angular convolution with kernel $a_j \times a_j$ into horizontal and vertical angular convolutions, reducing the angular factor from $a_j^2$ to $2a_j$. The auxiliary branch estimates spatial (NSS) and angular (GDD) feature vectors, and the total loss is $L = l_p + \lambda(l_s + l_a)$ with $\lambda=0.01$; this machinery is what the paper credits for the accuracy gain.

What would settle it

Check the augmented Win5-LID and SMART sets for overlap: if any test-set light field is a rotation or vertical flip of a training-set light field, the reported RMSE reductions are inflated. Alternatively, retrain ALAS-DADS with the train/test split performed strictly before augmentation and see whether the 42.86% and 45.95% gains over the second-best metric persist.

Watch

Extended reading notes

Core claim

The central discovery is that a light field image quality model can reach strong no-reference predictions by replacing full 4D convolutions with two factorized operations: light field depthwise separable convolution (LF-DSC), which applies the depthwise-plus-pointwise factorization subview-wise, and light field anglewise separable convolution (LF-ASC), which splits the angular kernel into horizontal and vertical passes. On top of these extractors, the model learns two auxiliary tasks—predicting a 36-dimensional BRISQUE natural-scene-statistics vector and an 8-dimensional NR-LFQA gradient-direction-distribution vector—whose losses are added to the quality-score loss with a weighting factor chosen as 0.01. The paper reports that this combination yields the lowest RMSE for every distortion type in both datasets, including error reductions above 60% for the EPICNN and USCD distortions, and that the factorized backbone has far fewer trainable parameters than a 10-layer 4D convolution baseline.

Load-bearing premise

The reported gains assume the 8-fold augmentation (90/180/270-degree rotations and vertical flipping) is applied only after the train/test split, so no augmented copy of a training light field appears in the test set; the paper describes both steps but does not explicitly state the order, so the metrics would be inflated if the split came after augmentation.

Editorial extensions

If this is right

  • On Win5-LID and SMART, ALAS-DADS reports lower RMSE than PSNR, SSIM, BRISQUE, and NR-LFQA across every distortion type tested.
  • For particularly challenging distortions, EPICNN and USCD, the paper reports RMSE reductions of 76.38% and 62.51% over the best prior metric.
  • The LF-DSC and LF-ASC decomposition produces comparable or better feature extraction with far fewer trainable parameters and shorter training time than a full 4D convolution stack in the ablation study.
  • The auxiliary spatial and angular feature estimation tasks act as hints for the primary quality score, with the weighting factor $\lambda=0.01$ balancing the three losses.
  • The authors argue that LF-DSC, LF-ASC, and the auxiliary learning scheme can be adapted to other light field tasks such as super-resolution, classification, and depth estimation, though these adaptations are not evaluated in the paper.

Reading between the lines

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

  • The auxiliary labels are themselves derived from 2D image statistics (BRISQUE NSS and NR-LFQA GDD), so the model's angular-quality notion is only as rich as those 2D features; pairing it with a true angular-consistency measure could improve accuracy further or reveal a ceiling in the current hints.
  • The LF-DSC and LF-ASC factorization should transfer to other 4D light-field tasks, but the paper only demonstrates it for quality assessment; a direct test on super-resolution or depth estimation would show whether the complexity savings carry over.
  • Because the split-versus-augmentation order is ambiguous, an independent re-run that fixes the split before augmentation would settle whether the 42.86% and 45.95% margins are the true gains.
  • Both datasets lack modern compression formats such as JPEG Pleno, so the metric's robustness on future immersive broadcasting distortions remains untested even if the reported gains are confirmed.
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 ALAS-DADS, a no-reference light field image quality assessment (NR-LFIQA) method built from two new convolution primitives: LF-DSC, which applies depthwise separable convolution in the spatial domain of each light field subview, and LF-ASC, which factorizes the angular dimension into horizontal and vertical 3D convolutions. A primary quality regression head is trained jointly with two auxiliary heads that estimate 36-D BRISQUE NSS features and 8-D NR-LFQA GDD features as spatial and angular hints. On Win5-LID and SMART, the method is reported to reduce RMSE by 42.86% and 45.95% relative to the second-best compared metric. The paper also contributes a computational-complexity comparison between standard 4D convolution and the proposed separable variants.

Significance. If the reported numbers are obtained under a clean evaluation protocol, the paper would be a solid contribution: the separable convolution layers are parameter-efficient, the complexity reductions in Eqs. (2)-(8) are concrete, and the benchmark gains are large. The paper also appears to be the first to use auxiliary spatial/angular feature estimation for NR-LFIQA. However, the evaluation as written has three load-bearing gaps: the split/augmentation order is ambiguous and can cause training/test leakage; the ablation does not isolate the auxiliary-learning component; and the sample-prediction table contains internal inconsistencies. The absence of code further limits reproducibility.

major comments (3)
  1. [Section IV-A2 and IV-B] The manuscript does not establish that the 80/20 data split is performed before the eightfold augmentation. Section IV-A2 states the datasets are enlarged to 1760 and 2048 LFIs, and Section IV-B says the datasets are 'firstly split' into 0.8/0.2 segments, but the order of these operations is ambiguous. Table VII shows test examples with names such as '+rot90' and '+rot180+flip', which indicates augmented copies are being evaluated. If the split is applied after augmentation, each original distorted LFI contributes eight rotation/flip copies sharing the same MOS/BT label, so near-duplicates appear in both training and test sets and the reported RMSE/SROCC in Table IV would be inflated. Please state explicitly the split order and, if the current protocol is ambiguous, re-run the comparison with a split on original LFI identities before augmentation.
  2. [Section IV-C, Table III] The ablation study compares 10-4D-Conv, 10-LF-DSC, 10-LF-ASC, and 10-LF-DSC-ASC and demonstrates that the separable-convolution backbones outperform standard 4D convolution. However, none of these variants includes the auxiliary spatial/angular feature estimation tasks described in Section III-D. The paper therefore provides no evidence that auxiliary learning contributes to the reported performance, even though the method's name and abstract attribute part of the gain to it. An ablation of ALAS-DADS with and without the auxiliary losses, under identical backbone and training protocol, is needed to support the central claim.
  3. [Section IV-D3, Table VII] Table VII contains internally inconsistent entries. For 'LN-museum-10 +rot270+flip', the true MOS is 4.3913 and ALAS-DADS predicts 2.8190, so the absolute error is 1.5723, not the reported 0.0070; for 'JPEG2000-museum-150 +rot90+flip', true is 2.8261, prediction is 4.3850, error 1.5589, not 0.0063. The rows appear to have swapped predictions. In the SMART block, several true BT scores (e.g., 3.1304 and 4.3913) fall outside the stated [-10,1] BT range. These errors contradict the text's claim that all sample errors are below 0.01 and need correction.
minor comments (6)
  1. [Section III-D] Please state explicitly whether the selection of lambda (1, 0.1, 0.01, 0.001) was done on the validation or test segment; tuning on the test set would invalidate the reported results.
  2. [Section IV-D1] The text contains a typo: 'state-or-art' should be 'state-of-the-art'.
  3. [Table I] The filter-shape notation '7 x 7 x 4 x 4 x 3' is not explained; the reader cannot tell which dimensions correspond to angular, spatial, input channel, and output channel. Please add a note.
  4. [Section IV-C] The ablation is performed only on Win5-LID; reporting the same ablation on SMART would strengthen the generality of the conclusion.
  5. [Section III-D and IV-D1] The comparison against NR-LFQA should be discussed in light of the fact that NR-LFQA's GDD features are used to form the angular auxiliary labels; the authors should clarify whether this gives ALAS-DADS access to information from a competing baseline.
  6. [Section IV-A3] The paper reports point estimates of RMSE, SROCC, and PLCC without confidence intervals or significance tests; given the small effective test sizes, such statistics would help assess the improvements.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the quality prediction is trained on human MOS/BT labels, and the auxiliary features borrowed from BRISQUE/NR-LFQA are transparent inputs rather than the evaluation target.

full rationale

The central claim (42.86%/45.95% RMSE reductions on Win5-LID/SMART) is an empirical result of training a CNN to regress subjective MOS/BT scores (Section IV-A1, Eq. 9), not a quantity derived from the model's own definitions. LF-DSC and LF-ASC are concrete layer designs whose only formal content is the FLOP arithmetic in Eqs. (2)-(8); the ablation in Table III provides independent empirical support. The auxiliary tasks use NSS features from BRISQUE [22] and GDD features from NR-LFQA [7] as regression labels (Section III-D), and [7] is a self-citation; however, NR-LFQA is a public method whose code was obtained and retrained under identical conditions, so the citation is independent support, and the primary evaluation target is human labels, not the auxiliary feature values. No uniqueness theorem or prior claim is imported to force the architecture. The one substantive validity concern is not circularity: Section IV-A2 describes an 8-fold geometric augmentation before Section IV-B says the datasets are 'firstly split' into 0.8/0.2, leaving unclear whether augmented near-duplicates of a training LFI appear in the test set; if so, the test RMSE/SROCC figures in Table IV could be inflated. This is a data-leakage risk requiring clarification, not a derivation that reduces to its inputs. The complexity analysis is parameter-free. Score 1 reflects the mild design-level borrowing from a self-authored baseline, with no construction-level circularity.

Assumptions & free parameters 1 free parameters · 5 assumptions · 0 invented entities

The central claim is empirical and rests on standard LFI representation, the reliability of borrowed NSS/GDD features, the factorizability assumption of LF-ASC, and the correctness of the train/test split. The only explicit fitted parameter is the auxiliary loss weight lambda. No new physical entities are introduced.

free parameters (1)
  • Auxiliary loss balancing factor lambda = 0.01
    Chosen from {1, 0.1, 0.01, 0.001} based on experimental results (Section III-D). It controls the contribution of the spatial and angular auxiliary tasks to the total loss.
assumptions (5)
  • domain assumption An LFI can be represented as a 4D grid (u,v,x,y) of subviews, and each pixel is indexed by spatial and angular coordinates.
    Standard light field representation used throughout Section II-A and III.
  • domain assumption BRISQUE NSS features and NR-LFQA GDD features are meaningful proxies for spatial and angular LFI quality.
    Used as labels for the auxiliary tasks (Section III-D). If these features are poor indicators, the auxiliary hints may not improve the primary task.
  • ad hoc to paper A 4D convolution can be factorized into two 3D convolutions (horizontal and vertical) without significant loss of representational capacity for quality assessment.
    This is the core assumption behind LF-ASC (Section III-C). It is plausible but not proven for LFI quality features.
  • domain assumption The train/test split is performed on original LFIs before the 8x data augmentation.
    Load-bearing procedural assumption; the paper is ambiguous about the ordering (Section IV-A2 vs IV-B). If false, test metrics are inflated by train/test overlap.
  • domain assumption Human MOS/BT scores in Win5-LID and SMART are reliable ground truth for perceived quality.
    Standard assumption in IQA research; used to train and evaluate the model.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Light Field Image Quality Assessment With Auxiliary Learning Based on Depthwise and Anglewise Separable Convolutions." pith.science (2026). https://pith.science/paper/6AVPCG4W

@misc{pith2026241207079,
  author       = {Pith},
  title        = {Pith review of: Light Field Image Quality Assessment With Auxiliary Learning Based on Depthwise and Anglewise Separable Convolutions},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6AVPCG4W}},
  note         = {Machine review of arXiv:2412.07079}
}
read the original abstract

In multimedia broadcasting, no-reference image quality assessment (NR-IQA) is used to indicate the user-perceived quality of experience (QoE) and to support intelligent data transmission while optimizing user experience. This paper proposes an improved no-reference light field image quality assessment (NR-LFIQA) metric for future immersive media broadcasting services. First, we extend the concept of depthwise separable convolution (DSC) to the spatial domain of light field image (LFI) and introduce "light field depthwise separable convolution (LF-DSC)", which can extract the LFI's spatial features efficiently. Second, we further theoretically extend the LF-DSC to the angular space of LFI and introduce the novel concept of "light field anglewise separable convolution (LF-ASC)", which is capable of extracting both the spatial and angular features for comprehensive quality assessment with low complexity. Third, we define the spatial and angular feature estimations as auxiliary tasks in aiding the primary NR-LFIQA task by providing spatial and angular quality features as hints. To the best of our knowledge, this work is the first exploration of deep auxiliary learning with spatial-angular hints on NR-LFIQA. Experiments were conducted in mainstream LFI datasets such as Win5-LID and SMART with comparisons to the mainstream full reference IQA metrics as well as the state-of-the-art NR-LFIQA methods. The experimental results show that the proposed metric yields overall 42.86% and 45.95% smaller prediction errors than the second-best benchmarking metric in Win5-LID and SMART, respectively. In some challenging cases with particular distortion types, the proposed metric can reduce the errors significantly by more than 60%.

Figures

Figures reproduced from arXiv: 2412.07079 by the authors.

Figure 1
Figure 1. The Architecture of the ALAS-DADS Framework [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Data Augmentation for a Sample LFI from Win5-LID [37] [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Comparison of Regression and Distortion Scatter Plots Between ALAS-DADS and Other FR-IQA Metrics [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Comparison of Regression and Distortion Scatter Plots Between ALAS-DADS and Other NR-IQA Metrics [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]
Figure 5
Figure 5. Figure 5: Scatter Plots Showing Predicted Values of ALAS-DADS vs Ground Truth Values [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]
Figure 6
Figure 6. Figure 6: Sample LFIs from Win5-LID [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]
Figure 7
Figure 7. Figure 7: Sample LFIs from SMART TABLE VII: Example Image Results for [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

49 extracted references · 39 canonical work pages

  1. [1]

    6-DoF image local- ization from massive geo-tagged reference images,

    Y . Song, X. Chen, X. Wang, Y . Zhang, and J. Li, “6-DoF image local- ization from massive geo-tagged reference images,” IEEE Transactions on Multimedia , vol. 18, no. 8, pp. 1542–1554, 2016

    work page 2016

  2. [2]

    Reduced reference quality assess- ment of light field images,

    P. Paudyal, F. Battisti, and M. Carli, “Reduced reference quality assess- ment of light field images,” IEEE Transactions on Broadcasting, vol. 65, no. 1, pp. 152–165, 2019. IEEE TRANSACTIONS ON BROADCASTING, VOL. 67, NO. 4, 2021 13

    work page 2019

  3. [3]

    Immersive light field video with a layered mesh representation,

    M. Broxton, J. Flynn, R. Overbeck, D. Erickson, P. Hedman, M. DuVall, J. Dourgarian, J. Busch, M. Whalen, and P. Debevec, “Immersive light field video with a layered mesh representation,” vol. 39, no. 4, pp. 86:1– 86:15, 2020

    work page 2020

  4. [4]

    Apple invents a light field panorama camera system for idevices and hmd that will create immersive scenes with 6 degrees of freedom,

    “Apple invents a light field panorama camera system for idevices and hmd that will create immersive scenes with 6 degrees of freedom,” https://www.patentlyapple.com/patently-apple/2020/04/apple-invents-a- light-field-panorama-camera-system-for-idevices-hmd-that-will-create- immersive-scenes-with-6-degrees-of-fre .html, accessed: 2021-03-21

    work page 2020

  5. [5]

    Eye-sensing light field display: Delivering 3D creators’ visions to customers the way they intended,

    “Eye-sensing light field display: Delivering 3D creators’ visions to customers the way they intended,” https://www .sony.net/SonyInfo/ technology/stories/LFD/, accessed: 2021-03-21

    work page 2021

  6. [6]

    BELIF: Blind quality evaluator of light field image with tensor structure variation index,

    L. Shi, S. Zhao, and Z. Chen, “BELIF: Blind quality evaluator of light field image with tensor structure variation index,” in 2019 IEEE International Conference on Image Processing (ICIP) . IEEE, 2019, pp. 3781–3785

    work page 2019

  7. [7]

    No-reference light field image quality assessment based on spatial-angular measurement,

    L. Shi, W. Zhou, Z. Chen, and J. Zhang, “No-reference light field image quality assessment based on spatial-angular measurement,” IEEE Transactions on Circuits and Systems for Video Technology , vol. 30, no. 11, pp. 4114–4128, 2019

    work page 2019

  8. [8]

    Image quality assessment: from error visibility to structural similarity,

    Z. Wang, A. C. Bovik, H. R. Sheikh, and E. P. Simoncelli, “Image quality assessment: from error visibility to structural similarity,” IEEE transactions on image processing , vol. 13, no. 4, pp. 600–612, 2004

    2004

Show all 49 references
  1. [9]

    Light field image processing: An overview,

    G. Wu, B. Masia, A. Jarabo, Y . Zhang, L. Wang, Q. Dai, T. Chai, and Y . Liu, “Light field image processing: An overview,” IEEE Journal of Selected Topics in Signal Processing , vol. 11, no. 7, pp. 926–954, 2017

    2017

  2. [10]

    Learning-based view synthesis for light field cameras,

    N. K. Kalantari, T.-C. Wang, and R. Ramamoorthi, “Learning-based view synthesis for light field cameras,” ACM Transactions on Graphics (TOG), vol. 35, no. 6, pp. 1–10, 2016

    2016

  3. [11]

    Mobilenets: Efficient convo- lutional neural networks for mobile vision applications,

    A. G. Howard, M. Zhu, B. Chen, D. Kalenichenko, W. Wang, T. Weyand, M. Andreetto, and H. Adam, “Mobilenets: Efficient convo- lutional neural networks for mobile vision applications,” arXiv preprint arXiv:1704.04861, 2017

    2017 arXiv

  4. [12]

    Principles of light field imaging: Briefly revisiting 25 years of research,

    I. Ihrke, J. Restrepo, and L. Mignard-Debise, “Principles of light field imaging: Briefly revisiting 25 years of research,”IEEE Signal Processing Magazine, vol. 33, no. 5, pp. 59–69, 2016

    2016

  5. [13]

    Multiscale structural similarity for image quality assessment,

    Z. Wang, E. P. Simoncelli, and A. C. Bovik, “Multiscale structural similarity for image quality assessment,” in The Thrity-Seventh Asilomar Conference on Signals, Systems & Computers , vol. 2. IEEE, 2003, pp. 1398–1402

    2003

  6. [14]

    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

  7. [15]

    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, 2010

    2010

  8. [16]

    VSI: A visual saliency-induced index for perceptual image quality assessment,

    L. Zhang, Y . Shen, and H. Li, “VSI: A visual saliency-induced index for perceptual image quality assessment,” IEEE Transactions on Image processing, vol. 23, no. 10, pp. 4270–4281, 2014

    2014

  9. [17]

    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, 2013

    2013

  10. [18]

    Efficient no-reference quality as- sessment and classification model for contrast distorted images,

    H. Ziaei Nafchi and M. Cheriet, “Efficient no-reference quality as- sessment and classification model for contrast distorted images,” IEEE Transactions on Broadcasting , vol. 64, no. 2, pp. 518–523, 2018

    2018

  11. [19]

    Blind image quality esti- mation via distortion aggravation,

    X. Min, G. Zhai, K. Gu, Y . Liu, and X. Yang, “Blind image quality esti- mation via distortion aggravation,” IEEE Transactions on Broadcasting , vol. 64, no. 2, pp. 508–517, 2018

    2018

  12. [20]

    Modeling the screen content image quality via multiscale edge attention similarity,

    Q. Yang, Z. Ma, Y . Xu, L. Yang, W. Zhang, and J. Sun, “Modeling the screen content image quality via multiscale edge attention similarity,” IEEE Transactions on Broadcasting , vol. 66, no. 2, pp. 310–321, 2020

    2020

  13. [21]

    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

  14. [22]

    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

    2012

  15. [23]

    A no-reference objective image sharpness metric based on the notion of just noticeable blur (jnb),

    R. Ferzli and L. J. Karam, “A no-reference objective image sharpness metric based on the notion of just noticeable blur (jnb),” IEEE transac- tions on image processing , vol. 18, no. 4, pp. 717–728, 2009

    2009

  16. [24]

    No-reference visually significant blocking artifact metric for natural scene images,

    S. Suthaharan, “No-reference visually significant blocking artifact metric for natural scene images,” Signal Processing, vol. 89, no. 8, pp. 1647– 1652, 2009

    2009

  17. [25]

    An improved perception-based no- reference objective image sharpness metric using iterative edge refine- ment,

    S. Varadarajan and L. J. Karam, “An improved perception-based no- reference objective image sharpness metric using iterative edge refine- ment,” in 2008 15th IEEE international conference on image processing . IEEE, 2008, pp. 401–404

    2008

  18. [26]

    Tensor oriented no-reference light field image quality assessment,

    W. Zhou, L. Shi, Z. Chen, and J. Zhang, “Tensor oriented no-reference light field image quality assessment,” IEEE Transactions on Image Processing, vol. 29, pp. 4070–4084, 2020

    2020

  19. [27]

    Light field image quality assessment via the light field coherence,

    Y . Tian, H. Zeng, J. Hou, J. Chen, and K.-K. Ma, “Light field image quality assessment via the light field coherence,” IEEE Transactions on Image Processing, vol. 29, pp. 7945–7956, 2020

    2020

  20. [28]

    Multi-residual networks: Improv- ing the speed and accuracy of residual networks,

    M. Abdi and S. Nahavandi, “Multi-residual networks: Improv- ing the speed and accuracy of residual networks,” arXiv preprint arXiv:1609.05672, 2016

    2016 arXiv

  21. [29]

    Resnet in resnet: Generalizing residual architectures,

    S. Targ, D. Almeida, and K. Lyman, “Resnet in resnet: Generalizing residual architectures,” arXiv preprint arXiv:1603.08029 , 2016

    2016 arXiv

  22. [30]

    An overview of multi-task learning in deep neural networks,

    S. Ruder, “An overview of multi-task learning in deep neural networks,” arXiv preprint arXiv:1706.05098 , 2017

    2017 arXiv

  23. [31]

    Multi- task rank learning for image quality assessment,

    L. Xu, J. Li, W. Lin, Y . Zhang, L. Ma, Y . Fang, and Y . Yan, “Multi- task rank learning for image quality assessment,” IEEE Transactions on Circuits and Systems for Video Technology , vol. 27, no. 9, pp. 1833– 1843, 2016

    2016

  24. [32]

    End-to- end blind image quality assessment using deep neural networks,

    K. Ma, W. Liu, K. Zhang, Z. Duanmu, Z. Wang, and W. Zuo, “End-to- end blind image quality assessment using deep neural networks,” IEEE Transactions on Image Processing , vol. 27, no. 3, pp. 1202–1213, 2017

    2017

  25. [33]

    Naturalness-aware deep no-reference image quality assessment,

    B. Yan, B. Bare, and W. Tan, “Naturalness-aware deep no-reference image quality assessment,” IEEE Transactions on Multimedia , vol. 21, no. 10, pp. 2603–2615, 2019

    2019

  26. [34]

    Personality-assisted multi- task learning for generic and personalized image aesthetics assessment,

    L. Li, H. Zhu, S. Zhao, G. Ding, and W. Lin, “Personality-assisted multi- task learning for generic and personalized image aesthetics assessment,” IEEE Transactions on Image Processing , vol. 29, pp. 3898–3910, 2020

    2020

  27. [35]

    Reinforcement learning with unsupervised auxiliary tasks,

    M. Jaderberg, V . Mnih, W. M. Czarnecki, T. Schaul, J. Z. Leibo, D. Sil- ver, and K. Kavukcuoglu, “Reinforcement learning with unsupervised auxiliary tasks,” arXiv preprint arXiv:1611.05397 , 2016

    2016 arXiv

  28. [36]

    Multi-task learning using uncer- tainty to weigh losses for scene geometry and semantics,

    A. Kendall, Y . Gal, and R. Cipolla, “Multi-task learning using uncer- tainty to weigh losses for scene geometry and semantics,” inProceedings of the IEEE conference on computer vision and pattern recognition , 2018, pp. 7482–7491

    2018

  29. [37]

    Perceptual evaluation of light field image,

    L. Shi, S. Zhao, W. Zhou, and Z. Chen, “Perceptual evaluation of light field image,” in 2018 25th IEEE International Conference on Image Processing (ICIP). IEEE, 2018, pp. 41–45

    2018

  30. [38]

    Depth-wise separable convolutions and multi-level pooling for an efficient spatial CNN-based steganalysis,

    R. Zhang, F. Zhu, J. Liu, and G. Liu, “Depth-wise separable convolutions and multi-level pooling for an efficient spatial CNN-based steganalysis,” IEEE Transactions on Information F orensics and Security , vol. 15, pp. 1138–1150, 2019

    2019

  31. [39]

    Lightweight deep residual CNN for fault diagnosis of rotating machinery based on depthwise separable convolutions,

    S. Ma, W. Liu, W. Cai, Z. Shang, and G. Liu, “Lightweight deep residual CNN for fault diagnosis of rotating machinery based on depthwise separable convolutions,” IEEE Access , vol. 7, pp. 57 023–57 036, 2019

    2019

  32. [40]

    Xceptiontime: A novel deep architecture based on depthwise separable convolutions for hand gesture classification,

    E. Rahimian, S. Zabihi, S. F. Atashzar, A. Asif, and A. Moham- madi, “Xceptiontime: A novel deep architecture based on depthwise separable convolutions for hand gesture classification,” arXiv preprint arXiv:1911.03803, 2019

    1911 arXiv

  33. [41]

    Improved image classification with 4D light-field and interleaved con- volutional neural network,

    Z. Lu, H. W. Yeung, Q. Qu, Y . Y . Chung, X. Chen, and Z. Chen, “Improved image classification with 4D light-field and interleaved con- volutional neural network,” Multimedia Tools and Applications , vol. 78, no. 20, pp. 29 211–29 227, 2019

    2019

  34. [42]

    Towards the perceptual quality evaluation of compressed light field images,

    P. Paudyal, F. Battisti, M. Sj ¨ostr¨om, R. Olsson, and M. Carli, “Towards the perceptual quality evaluation of compressed light field images,”IEEE Transactions on Broadcasting , vol. 63, no. 3, pp. 507–522, 2017

    2017

  35. [43]

    Fast light field reconstruction with deep coarse-to-fine modeling of spatial-angular clues,

    H. W. F. Yeung, J. Hou, J. Chen, Y . Y . Chung, and X. Chen, “Fast light field reconstruction with deep coarse-to-fine modeling of spatial-angular clues,” in Proceedings of the European Conference on Computer Vision (ECCV), 2018, pp. 137–152

    2018

  36. [44]

    Light field saliency detection with deep convolutional networks,

    J. Zhang, Y . Liu, S. Zhang, R. Poppe, and M. Wang, “Light field saliency detection with deep convolutional networks,” IEEE Transactions on Image Processing, vol. 29, pp. 4421–4434, 2020

    2020

  37. [45]

    F. M. Dekking, C. Kraaikamp, H. P. Lopuha ¨a, and L. E. Meester, A Modern Introduction to Probability and Statistics: Understanding why and how . Springer Science & Business Media, 2005

    2005

  38. [46]

    Zwillinger and S

    D. Zwillinger and S. Kokoska, CRC standard probability and statistics tables and formulae . CRC Press, 1999

    1999

  39. [47]

    On the convergence of adam and beyond,

    S. J. Reddi, S. Kale, and S. Kumar, “On the convergence of adam and beyond,” arXiv preprint arXiv:1904.09237 , 2019

    1904 arXiv

  40. [48]

    No reference quality assessment of stereo video based on saliency and sparsity,

    J. Yang, C. Ji, B. Jiang, W. Lu, and Q. Meng, “No reference quality assessment of stereo video based on saliency and sparsity,” IEEE Transactions on Broadcasting , vol. 64, no. 2, pp. 341–353, 2018

    2018

  41. [49]

    JPEG Pleno: Standardizing a coding framework and tools for plenoptic imaging modalities,

    P. Astola, L. A. da Silva Cruz, E. A. da Silva, T. Ebrahimi, P. G. Freitas, A. Gilles, K.-J. Oh, C. Pagliari, F. Pereira, C. Perra et al. , “JPEG Pleno: Standardizing a coding framework and tools for plenoptic imaging modalities,” ITU Journal: ICT Discoveries , 2020

    2020

Pith tools

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