Pith. sign in

REVIEW 5 major objections 5 minor 41 references

Fully Automated Image De-fencing using Conditional Generative Adversarial Networks

T0 review · 5 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read A single conditional GAN, conditioned on a Canny edge map of the fenced photo, can de-fence images with visual quality close to that of a two-network mask-and-inpaint pipeline while needing fewer layers and less time.

desk verdict A plausible extension-level idea for GAN-based de-fencing that is currently unsupported by missing quantitative evidence and an underspecified two-stage baseline; worth a serious referee only as a major-revision candidate. read the letter →

arxiv 1908.06837 v1 pith:LPBZB3KH submitted 2019-08-19 cs.CV

Divyanshu Gupta , Shorya Jain , Utkarsh Tripathi , Pratik Chattopadhyay , Lipo Wang This is my paper · ORCID
classification cs.CV
keywords imagede-fencingconditionalgenerativeadversarialnetworksCannyedgemapfencemaskgenerationinpaintingsingle-stagetranslationedge-conditioned
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

Fences and cage bars in photos block the view of the intended subject, and removing them usually takes two separate steps: find the fence pixels, then fill in what they hide. This paper tries to show that both steps can be replaced by a single conditional generative adversarial network (cGAN) that maps a fenced image directly to a de-fenced image. To keep the one-network version accurate despite having fewer layers, the generator is trained not only on fenced/de-fenced image pairs but also on a Canny edge map of the fenced input, which tells it where the fence lines are. The authors report that this single-stage network produces de-fenced images with visual quality similar to their own two-stage mask-then-inpaint pipeline while being faster, and that both GAN-based approaches beat earlier semi-automated de-fencing techniques. If that holds, fast and fully automatic fence removal becomes practical for ordinary photographs.

What carries the argument

The central mechanism is the edge-conditioned input to a single cGAN generator: the fenced RGB image and its Canny edge map are stacked and passed through an encoder–decoder, while a PatchGAN discriminator—a network that classifies small 16×16 image patches as real or fake—provides the adversarial feedback. The Canny map is a cheap hand-crafted prior that makes the periodic fence lines explicit, so the generator can locate occluding structure even without the extra layers of the two-stage pipeline. Training minimizes a composite objective: adversarial loss plus L1, perceptual, style, and SSIM losses, mirroring the loss of the two-stage image-recovery network.

What would settle it

Take a set of real scenes photographed both through a fence and from the same viewpoint without the fence, run both the single-stage and two-stage networks, and compare the de-fenced output with the unobstructed photo; if the single-stage output is visibly worse or scores clearly lower on structural similarity, the paper's parity claim fails.

Watch

Extended reading notes

Core claim

The paper argues that image de-fencing can be compressed into a single generator. Its two-stage baseline first runs a cGAN that outputs a fence mask from a fenced image, then a second cGAN that inpaints the masked image; the single-stage variant skips the mask step and instead concatenates a Canny edge map of the fenced image with the RGB input, feeding the pair to one encoder-decoder generator and one PatchGAN discriminator. The paper reports that this single-stage network produces de-fenced images visually similar to the two-stage outputs, despite having fewer layers, and that both GAN-based approaches improve on earlier de-fencing methods. It identifies the Canny edge map as the ingredient that preserves quality in the shallower network.

Load-bearing premise

The load-bearing premise is that the synthetic fences overlaid on ordinary dataset images during training are representative enough of real fences that the reported visual-parity results transfer to actual fenced photographs.

Editorial extensions

If this is right

  • A single forward pass through one generator can replace the two-network mask-and-inpaint sequence, so de-fencing is faster and needs less memory.
  • No user interaction is required at test time: the edge map is computed automatically and the same generator handles any input of the trained size.
  • If the quality comparison holds, practitioners can choose the single-stage model whenever response time matters without giving up much visual quality.
  • The two GAN-based formulations give a common learning-based baseline that later de-fencing work can be measured against.

Reading between the lines

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

  • [Editorial inference] Because the edge map supplies the periodic structure, the same input trick likely transfers to other repeating occluders, such as window grilles, chain-link mesh, or slatted blinds, after retraining.
  • [Editorial inference] The training data are synthetic fence overlays on general photos, so the parity claim is probably strongest for regular, near-textureless fence patterns; real fences with irregular spacing or strong cast shadows may widen the gap unless real paired data are added.
  • [Editorial inference] If the single-stage speed holds up, adding a temporal-consistency term could turn this into a practical video de-fencing method, which the paper lists as future work.
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

5 major / 5 minor

Summary. The paper proposes two cGAN-based image de-fencing pipelines: a two-stage network that first predicts a fence mask and then inpaints the masked regions, and a single-stage network that directly maps a fenced image plus its Canny edge map to a de-fenced image. The authors claim that the single-stage network achieves visual quality comparable to the two-stage network while being faster, and that comparative experiments show superiority over existing de-fencing methods. The manuscript describes the network losses in detail but the experimental section (Section 4) contains only qualitative image examples; no numeric metrics, baseline comparisons, or reproducibility artifacts are provided.

Significance. If supported, the single-stage design would be a practically useful contribution because it targets the latency bottleneck of sequential mask-and-inpaint pipelines and uses a simple edge prior. The related-work survey is thorough and the loss design follows established cGAN inpainting practice. However, as submitted the paper's central empirical claims are unverifiable: the abstract promises quantitative results and comparisons, but Section 4 presents none, the two-stage baseline is not reproducible as written, and the single-stage evaluation rests on a small set of qualitative examples. The contribution therefore cannot be assessed beyond a plausible architectural proposal.

major comments (5)
  1. [Abstract and Section 4] The abstract states that 'qualitative and quantitative experimental results reported in the manuscript' support the claims, and Section 4.3 states a 'comparative performance analysis' with state-of-the-art techniques, but no quantitative results appear anywhere. There are no PSNR, SSIM, or any other numeric metrics, no tables, no error bars, and no comparison to prior de-fencing methods. The central claim of comparable quality and superiority over existing methods is therefore unsupported by any evidence.
  2. [Section 3.1.1, Eq. (1) and Section 3.1.2, Eq. (5)] The two-stage pipeline is not reproducible as written. Eq. (1) defines the mask generator as Ipred = G1(Ifen, Imask), conditioning on the ground-truth fence mask, and Eq. (5) defines the recovery network as \tilde{I}pred = G2(\tilde{I}fen, Idef), conditioning on the ground-truth de-fenced image. Since these ground-truths are unavailable at evaluation, the test-time comparison in Section 4.3 between the single- and two-stage networks cannot be executed as described. The paper must state clearly which inputs are used at train time versus test time and rewrite the equations accordingly; if conditioning on ground truth is intentional, the task becomes trivial for the generator.
  3. [Section 4.1] The synthetic fence generation procedure is not specified. The text says only that 'artificial fence structures' are added to Pascal VOC and COCO images, but gives no details of fence types, densities, orientations, or overlay protocol. This makes the dataset irreproducible and prevents assessment of whether the synthetic distribution is representative of real fences. Moreover, because both training and test sets are generated by the same overlay process, the reported qualitative success partly reflects the network's ability to invert that specific synthetic operation rather than general de-fencing ability; the few real images in the test set have no ground truth and are not evaluated quantitatively.
  4. [Section 4.3 and Section 3.2] The claim that the single-stage network achieves 'similar visual quality' to the two-stage network with 'fewer number of layers' is not supported by any quantitative comparison or by reporting the actual number of layers or inference time of either network. The comparison in Figure 5 is purely visual, on a small selection of images, and no evidence is given that the single-stage network is indeed faster in terms of parameters, FLOPs, or wall-clock time. Without such measurements, the central efficiency claim is unsubstantiated.
  5. [Section 3.2] The contribution of the Canny edge map channel is not ablated. Although Figure 5 shows outputs with and without Canny supervision on a few images, the paper provides no quantitative measure (e.g., PSNR/SSIM on the test set) of how much the edge map improves reconstruction, and no analysis of its robustness to the Canny detector's threshold parameters. Since the paper's central proposal is the use of edge maps to compensate for fewer layers, this missing ablation is a load-bearing gap.
minor comments (5)
  1. [Section 3.1.1, Eq. (2)] There is a mismatched parenthesis in Eq. (2); the expression should be read as min_G1 (α1 max_D1 Ladv,1) + β1 LL1,1. Please correct the typesetting.
  2. [Section 3.1.2, Eq. (11)] The SSIM formula in Eq. (11) is garbled; the terms 'µx2µy 2 +C1' and 'σx2σy 2 +C2' should be rendered with proper superscripts. As written, the definition is unreadable.
  3. [References] The Canny edge filter is used without a citation. Please add the classic reference (Canny, 1986).
  4. [References] In the reference to Zheng and Kambhamettu, 'Digital mmtting' should be 'Digital matting'.
  5. [Contributions and Section 4.1] The contributions state that the codes and dataset are publicly available, but no repository link or dataset URL is provided anywhere in the manuscript. This should be added or the claim removed.

Circularity Check

1 steps flagged · score 3.0 of 10

Evaluation is closed on the synthetic degradation procedure; no equation-level circularity.

  1. other [Section 4.1 (Data Set and System Description); Section 4.3 (Experiments with Single-Stage Image De-fencing Network)]
    "The data set for training the single-stage end-to-end image de-fencing network is also constructed in a similar manner by applying random fence structures on a set of images from Pascal VOC data and COCO data. The test set consists of a total of 245 images and is formed by selecting images from the above-mentioned public data sets as well as some images captured by our research team."

    Training inputs are created by overlaying random artificial fence structures on clean Pascal VOC/COCO images, so the ground-truth de-fenced image is, by construction, the exact clean image used to synthesize the fenced input. The test set is then said to be formed by selecting images from the same public data sets, i.e., from the same source distribution used for the synthetic overlay. The reported de-fencing success therefore partly measures whether the network learned to invert the synthetic data-construction rule, rather than whether it generalizes to real fence textures. If the public-dataset test images are not themselves overlaid with fences, they cannot even serve as fenced inputs, which makes the evaluation protocol underspecified.

full rationale

The paper contains no first-principles derivation and no fitted-parameter-then-predicted quantity: the single-stage and two-stage architectures are standard cGAN designs with fixed losses, and no self-citations appear, so no self-citation chain is load-bearing. The central comparison between the single- and two-stage networks is, as written, difficult to reproduce because Eq. (1) conditions G1 on the ground-truth mask Imask and Eq. (5) conditions G2 on the ground-truth de-fenced image Idef, while the abstract says that at evaluation the fenced image is passed through the two generators in succession; this is a specification and reproducibility defect rather than circularity. The claim that quantitative results and public code/data are provided is unsupported by any table, metric, repository link, or baseline in the manuscript, which is a missing-evidence problem, not a circularity problem. The only evaluation-circular aspect is the synthetic-data protocol: training pairs are built by overlaying artificial fence structures on clean images, and the test set is selected from the same public datasets, so the ground truth equals the pre-overlay image used to construct the input. That warrants a modest score around 3, but the paper does not reduce to a self-citation chain or to a fitted input renamed as a prediction.

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

The central claim rests on hand-selected loss weights, an unspecified Canny edge configuration, and the assumption that synthetic fence data represents real scenes. No new physical entities are introduced; all components are recombinations of prior cGAN and inpainting components.

free parameters (4)
  • Loss weights in fence mask network (α1, β1) = α1=1, β1=10
    Hand-picked constants in Eq. (2) without sensitivity analysis.
  • Loss weights in recovery or single-stage network (α2, β2, γ, δ, η) = α2=0.1, β2=10, γ=2, δ=1, η=1
    Hand-picked constants in Eq. (12); no ablations show their effect on quality.
  • Early stopping threshold ε = 1e-3
    Convergence threshold in Section 3.1.1; no justification or impact analysis.
  • Canny edge detector parameters
    Section 3.2 names the Canny filter but never gives thresholds or sigma, so the edge map cannot be reproduced exactly.
assumptions (4)
  • domain assumption Training on synthetic fences from Pascal VOC and COCO images transfers to real fenced photographs
    Section 4.1 builds training data by adding artificial fence structures; if the synthetic distribution differs from real fences, claims of effectiveness may not hold.
  • ad hoc to paper Canny edge maps provide useful supervision that compensates for fewer network layers
    Section 3.2 asserts this benefit but provides no ablation or quantitative evidence isolating the edge-map contribution.
  • domain assumption Pretrained VGG-19 features and PatchGAN discriminator are suitable for de-fencing
    Perceptual and style losses (Eq. 7, 8) and adversarial loss rely on these prior components; suitability is assumed rather than tested.
  • domain assumption The objective functions in Eq. (12) adequately capture de-fencing quality
    The paper chooses a sum of adversarial, L1, perceptual, style, and SSIM losses from prior work; convergence on synthetic data is used as the only validation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Fully Automated Image De-fencing using Conditional Generative Adversarial Networks." pith.science (2026). https://pith.science/paper/LPBZB3KH

@misc{pith2026190806837,
  author       = {Pith},
  title        = {Pith review of: Fully Automated Image De-fencing using Conditional Generative Adversarial Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LPBZB3KH}},
  note         = {Machine review of arXiv:1908.06837}
}
read the original abstract

Image de-fencing is one of the important aspects of recreational photography in which the objective is to remove the fence texture present in an image and generate an aesthetically pleasing version of the same image without the fence texture. In this paper, we aim to develop an automated and effective technique for fence removal and image reconstruction using conditional Generative Adversarial Networks (cGANs). These networks have been successfully applied in several domains of Computer Vision focusing on image generation and rendering. Our initial approach is based on a two-stage architecture involving two cGANs that generate the fence mask and the inpainted image, respectively. Training of these networks is carried out independently and, during evaluation, the input image is passed through the two generators in succession to obtain the de-fenced image. The results obtained from this approach are satisfactory, but the response time is long since the image has to pass through two sets of convolution layers. To reduce the response time, we propose a second approach involving only a single cGAN architecture that is trained using the ground-truth of fenced de-fenced image pairs along with the edge map of the fenced image produced by the Canny Filter. Incorporation of the edge map helps the network to precisely detect the edges present in the input image, and also imparts it an ability to carry out high quality de-fencing in an efficient manner, even in the presence of a fewer number of layers as compared to the two-stage network. Qualitative and quantitative experimental results reported in the manuscript reveal that the de-fenced images generated by the single-stage de-fencing network have similar visual quality to those produced by the two-stage network. Comparative performance analysis also emphasizes the effectiveness of our approach over state-of-the-art image de-fencing techniques.

Figures

Figures reproduced from arXiv: 1908.06837 by the authors.

Figure 1
Figure 1. (a) Input image with fence, (b) fence detection, (c) de-fenced image after inpainting. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Two-stage image de-fencing network. 3.1.1 Fence Mask Generator The task of the fence mask generator can be viewed as an image-to-image translation problem where we convert a fenced image to a fence mask image. Let Imask be the ground-truth fence mask image, and If en the input fenced image. During the training phase, the generator takes the input as If en, conditioned on Imask and predicts the fence mask image Ipred… view at source ↗
Figure 3
Figure 3. Single-stage image de-fencing network. Use of two generators makes the image de-fencing process time-intensive. To reduce the response time, we propose to use a single end-to-end architecture with a single generator-discriminator pair for translating the input image with a fence structure directly to its de-fenced version without any fence generation step in between. Since only a single generator is used here, the s… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Qualitative results for the two-stage image de-fencing network. [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Qualitative results by the single stage end-to-end image de-fencing network. [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

41 extracted references · 34 canonical work pages

  1. [1]

    Simultaneous structure and texture image inpainting

    Marcelo Bertalmio, Luminita Vese, Guillermo Sapiro, and Stanley Osher. Simultaneous structure and texture image inpainting. IEEE Transactions on Image Processing, 12 0 (8): 0 882--889, 2003

    work page 2003

  2. [2]

    I know that person: Generative full body and face de-identification of people in images

    Karla Brkic, Ivan Sikiric, Tomislav Hrkac, and Zoran Kalafatic. I know that person: Generative full body and face de-identification of people in images. In 2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), pages 1319--1328. IEEE, 2017

    work page 2017

  3. [3]

    Infogan: Interpretable representation learning by information maximizing generative adversarial nets

    Xi Chen, Yan Duan, Rein Houthooft, John Schulman, Ilya Sutskever, and Pieter Abbeel. Infogan: Interpretable representation learning by information maximizing generative adversarial nets. In Advances in Neural Information Processing Systems, pages 2172--2180, 2016

    work page 2016

  4. [4]

    Region filling and object removal by exemplar-based image inpainting

    Antonio Criminisi, Patrick P \'e rez, and Kentaro Toyama. Region filling and object removal by exemplar-based image inpainting. IEEE Transactions on Image Processing, 13 0 (9): 0 1200--1212, 2004

    work page 2004

  5. [5]

    Image melding: Combining inconsistent images using patch-based synthesis

    Soheil Darabi, Eli Shechtman, Connelly Barnes, Dan B Goldman, and Pradeep Sen. Image melding: Combining inconsistent images using patch-based synthesis. ACM Transactions on Graphics, 31 0 (4): 0 82--1, 2012

    work page 2012

  6. [6]

    Accurate and efficient video de-fencing using convolutional neural networks and temporal information

    Chen Du, Byeongkeun Kang, Zheng Xu, Ji Dai, and Truong Nguyen. Accurate and efficient video de-fencing using convolutional neural networks and temporal information. In Proceeding of the IEEE International Conference on Multimedia and Expo, pages 1--6. IEEE, 2018

    work page 2018

  7. [7]

    The pascal visual object classes (voc) challenge

    Mark Everingham, Luc Van Gool, Christopher KI Williams, John Winn, and Andrew Zisserman. The pascal visual object classes (voc) challenge. International Journal of Computer Vision, 88 0 (2): 0 303--338, 2010

    work page 2010

  8. [8]

    Image de-fencing framework with hybrid inpainting algorithm

    Muhammad Shahid Farid, Arif Mahmood, and Marco Grangetto. Image de-fencing framework with hybrid inpainting algorithm. Signal, Image and Video Processing, 10 0 (7): 0 1193--1201, 2016

    work page 2016

Show all 41 references
  1. [9]

    Efficient belief propagation for early vision

    Pedro F Felzenszwalb and Daniel P Huttenlocher. Efficient belief propagation for early vision. International Journal of Computer Vision, 70 0 (1): 0 41--54, 2006

    2006

  2. [10]

    Image style transfer using convolutional neural networks

    Leon A Gatys, Alexander S Ecker, and Matthias Bethge. Image style transfer using convolutional neural networks. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 2414--2423, 2016

    2016

  3. [11]

    Discovering texture regularity as a higher-order correspondence problem

    James Hays, Marius Leordeanu, Alexei A Efros, and Yanxi Liu. Discovering texture regularity as a higher-order correspondence problem. In European Conference on Computer Vision, pages 522--535. Springer, 2006

    2006

  4. [12]

    Yu, and Changhu Wang

    He Huang, Philip S. Yu, and Changhu Wang. An introduction to image synthesis with generative adversarial nets. CoRR, abs/1803.04469, 2018

    2018 arXiv

  5. [13]

    Image completion using planar structure guidance

    Jia-Bin Huang, Sing Bing Kang, Narendra Ahuja, and Johannes Kopf. Image completion using planar structure guidance. ACM Transactions on Graphics (TOG), 33 0 (4): 0 129, 2014

    2014

  6. [14]

    Image-to-image translation with conditional adversarial networks

    Phillip Isola, Jun-Yan Zhu, Tinghui Zhou, and Alexei A Efros. Image-to-image translation with conditional adversarial networks. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 1125--1134, 2017

    2017

  7. [15]

    Perceptual losses for real-time style transfer and super-resolution

    Justin Johnson, Alexandre Alahi, and Li Fei-Fei. Perceptual losses for real-time style transfer and super-resolution. In European Conference on Computer Vision, pages 694--711. Springer, 2016

    2016

  8. [16]

    My camera can see through fences: A deep learning approach for image de-fencing

    Sankaraganesh Jonna, Krishna K Nakka, and Rajiv R Sahay. My camera can see through fences: A deep learning approach for image de-fencing. In Proceedings of the 3^ rd IAPR Asian Conference on Pattern Recognition (ACPR) , pages 261--265. IEEE, 2015 a

    2015

  9. [17]

    A multimodal approach for image de-fencing and depth inpainting

    Sankaraganesh Jonna, Vikram S Voleti, Rajiv R Sahay, and Mohan S Kankanhalli. A multimodal approach for image de-fencing and depth inpainting. In Proceedings of the 8^ th International Conference on Advances in Pattern Recognition (ICAPR) , pages 1--6. IEEE, 2015 b

    2015

  10. [18]

    Deep learning based fence segmentation and removal from an image using a video sequence

    Sankaraganesh Jonna, Krishna K Nakka, and Rajiv R Sahay. Deep learning based fence segmentation and removal from an image using a video sequence. In European Conference on Computer Vision, pages 836--851. Springer, 2016

    2016

  11. [19]

    Seeing through the fence: Image de-fencing using a video sequence

    Vrushali S Khasare, Rajiv R Sahay, and Mohan S Kankanhalli. Seeing through the fence: Image de-fencing using a video sequence. In Proceedings of the IEEE International Conference on Image Processing, pages 1351--1355. IEEE, 2013

    2013

  12. [20]

    Image defencing via signal demixing

    Veepin Kumar, Jayanta Mukherjee, and Shyamal Kumar Das Mandal. Image defencing via signal demixing. In Proceedings of the 10^ th Indian Conference on Computer Vision, Graphics and Image Processing , page 11. ACM, 2016

    2016

  13. [21]

    A closed-form solution to natural image matting

    Anat Levin, Dani Lischinski, and Yair Weiss. A closed-form solution to natural image matting. IEEE Transactions on Pattern Analysis and Machine Intelligence, 30 0 (2): 0 228--242, 2008

    2008

  14. [22]

    Microsoft coco: Common objects in context

    Tsung-Yi Lin, Michael Maire, Serge Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Doll \'a r, and C Lawrence Zitnick. Microsoft coco: Common objects in context. In European Conference on Computer Vision, pages 740--755. Springer, 2014

    2014

  15. [23]

    A lattice-based mrf model for dynamic near-regular texture tracking

    Wen-Chieh Lin and Yanxi Liu. A lattice-based mrf model for dynamic near-regular texture tracking. IEEE Transactions on Pattern Analysis and Machine Intelligence, 29 0 (5): 0 777--792, 2007

    2007

  16. [24]

    Image de-fencing

    Yanxi Liu, Tamara Belkina, James Hays, and Roberto Lublinerman. Image de-fencing. 2008 IEEE Conference on Computer Vision and Pattern Recognition, pages 1--8, 2008

    2008

  17. [25]

    Fully convolutional networks for semantic segmentation

    Jonathan Long, Evan Shelhamer, and Trevor Darrell. Fully convolutional networks for semantic segmentation. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 3431--3440, 2015

    2015

  18. [26]

    Video de-fencing

    Yadong Mu, Wei Liu, and Shuicheng Yan. Video de-fencing. IEEE Transactions on Circuits and Systems for Video Technology, 24 0 (7): 0 1111--1121, 2014

    2014

  19. [27]

    Qureshi, and Mehran Ebrahimi

    Kamyar Nazeri, Eric Ng, Tony Joseph, Faisal Z. Qureshi, and Mehran Ebrahimi. Edgeconnect: Generative image inpainting with adversarial edge learning. CoRR, abs/1901.00212, 2019

    1901 arXiv

  20. [28]

    Deformed lattice discovery via efficient mean-shift belief propagation

    Minwoo Park, Robert T Collins, and Yanxi Liu. Deformed lattice discovery via efficient mean-shift belief propagation. In European Conference on Computer Vision, pages 474--485. Springer, 2008

    2008

  21. [29]

    Deformed lattice detection in real-world images using mean-shift belief propagation

    Minwoo Park, Kyle Brocklehurst, Robert T Collins, and Yanxi Liu. Deformed lattice detection in real-world images using mean-shift belief propagation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 31 0 (10): 0 1804--1816, 2009

    2009

  22. [30]

    Image de-fencing revisited

    Minwoo Park, Kyle Brocklehurst, Robert T Collins, and Yanxi Liu. Image de-fencing revisited. In Asian Conference on Computer Vision, pages 422--434. Springer, 2010

    2010

  23. [31]

    Context encoders: Feature learning by inpainting

    Deepak Pathak, Philipp Krahenbuhl, Jeff Donahue, Trevor Darrell, and Alexei A Efros. Context encoders: Feature learning by inpainting. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 2536--2544, 2016

    2016

  24. [32]

    Unsupervised representation learning with deep convolutional generative adversarial networks

    Alec Radford, Luke Metz, and Soumith Chintala. Unsupervised representation learning with deep convolutional generative adversarial networks. arXiv preprint arXiv:1511.06434, 2015

    2015 arXiv

  25. [33]

    Generative adversarial text to image synthesis

    Scott Reed, Zeynep Akata, Xinchen Yan, Lajanugen Logeswaran, Bernt Schiele, and Honglak Lee. Generative adversarial text to image synthesis. arXiv preprint arXiv:1605.05396, 2016

    2016 arXiv

  26. [34]

    Image quality assessment: from error visibility to structural similarity

    Zhou Wang, Alan C Bovik, Hamid R Sheikh, Eero P Simoncelli, et al. Image quality assessment: from error visibility to structural similarity. IEEE Transactions on Image Processing, 13 0 (4): 0 600--612, 2004

    2004

  27. [35]

    Image inpainting by patch propagation using patch sparsity

    Zongben Xu and Jian Sun. Image inpainting by patch propagation using patch sparsity. IEEE transactions on image processing, 19 0 (5): 0 1153--1165, 2010

    2010

  28. [36]

    High-resolution image inpainting using multi-scale neural patch synthesis

    Chao Yang, Xin Lu, Zhe Lin, Eli Shechtman, Oliver Wang, and Hao Li. High-resolution image inpainting using multi-scale neural patch synthesis. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 6721--6729, 2017

    2017

  29. [37]

    Semantic image inpainting with deep generative models

    Raymond A Yeh, Chen Chen, Teck Yian Lim, Alexander G Schwing, Mark Hasegawa-Johnson, and Minh N Do. Semantic image inpainting with deep generative models. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 5485--5493, 2017

    2017

  30. [38]

    Lin, Jimei Yang, Xiaohui Shen, Xin Lu, and Thomas S

    Jiahui Yu, Zhe L. Lin, Jimei Yang, Xiaohui Shen, Xin Lu, and Thomas S. Huang. Generative image inpainting with contextual attention. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 5505--5514, 2018

    2018

  31. [39]

    Stackgan: Text to photo-realistic image synthesis with stacked generative adversarial networks

    Han Zhang, Tao Xu, Hongsheng Li, Shaoting Zhang, Xiaogang Wang, Xiaolei Huang, and Dimitris N Metaxas. Stackgan: Text to photo-realistic image synthesis with stacked generative adversarial networks. In Proceedings of the IEEE International Conference on Computer Vision, pages ...

    2017

  32. [40]

    Loss functions for image restoration with neural networks

    Hang Zhao, Orazio Gallo, Iuri Frosio, and Jan Kautz. Loss functions for image restoration with neural networks. IEEE Transactions on Computational Imaging, 3 0 (1): 0 47--57, 2017

    2017

  33. [41]

    Learning based digital mmtting

    Yuanjie Zheng and Chandra Kambhamettu. Learning based digital mmtting. In Proceedings of the IEEE 12^ th International Conference on Computer Vision , pages 889--896. IEEE, 2009

    2009

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.