Pith. sign in

REVIEW 3 major objections 5 minor 42 references

Image Inpainting with Learnable Bidirectional Attention Maps

T0 review · 3 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 This paper claims that replacing partial convolution's hand-crafted mask renormalization and mask updating with learnable bidirectional attention maps yields sharper, more coherent, and more visually plausible image inpainting for…

desk verdict Reverse attention is a real idea, but the paper's main claim against partial convolution isn't established because the baseline numbers come from a different training protocol. read the letter →

arxiv 1909.00968 v3 pith:NDPKGLXH submitted 2019-09-03 cs.CV

Chaohao Xie , Shaohui Liu , Chao Li , Ming-Ming Cheng , Wangmeng Zuo , Xiao Liu , Shilei Wen , Errui Ding This is my paper · ORCID
classification cs.CV
keywords imageinpaintinglearnableattentionmapsbidirectionalpartialconvolutionirregularholesU-Netfeaturerenormalizationadversarialtraining
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 proposes learnable bidirectional attention maps (LBAM) for image inpainting with irregular holes. It argues that replacing partial convolution's fixed mask renormalization and mask-updating rules with learned attention maps, and adding reverse attention maps that focus the decoder on the missing region, produces sharper and more coherent results. On Paris StreetView and Places, the authors report higher PSNR and SSIM than existing methods at high mask ratios, along with a 63.2% user preference. If correct, this makes learned attention a practical alternative to the hand-crafted rules in partial convolution networks.

What carries the argument

The central object is the learnable attention map, a module that renormalizes convolution features by multiplying them with an attention map produced by an asymmetric Gaussian-shaped activation, and updates the mask with a ReLU-based function where the exponent is a hyperparameter. The attention map's parameters are layer-specific and learned end-to-end, so the network decides how much to trust intermediate filled regions rather than trusting all of them equally. The reverse attention map applies the same mechanism to the decoder using the complementary mask, restricting the decoder's task to the hole. These maps are inserted into every U-Net layer — forward on the encoder, reverse on the decoder — and this bidirectional learned renormalization is what carries the claimed improvement.

What would settle it

Rerun partial convolution under this paper's exact training protocol — the same Places subset, mask generation, loss weights, and epoch count — and compare PSNR and SSIM on the same test masks. If the gap disappears or reverses on masks with ratio above 0.3, the claimed advantage over partial convolution is not established.

Watch

Extended reading notes

Core claim

The paper claims that the three hand-crafted components of partial convolution — the fixed mask-convolution filter, the hard 0-1 mask updating, and the all-one decoder mask — can be replaced by learned counterparts. Its forward attention map learns feature renormalization and mask propagation through an asymmetric Gaussian activation function and a ReLU-based mask update, while a reverse attention map on the decoder side, driven by the complementary mask, lets the decoder concentrate on filling the hole. The result is a U-Net whose encoder and decoder features are both renormalized by learned masks, trained end-to-end with pixel reconstruction, perceptual, style, and adversarial losses. The paper argues that this yields visually sharper and more coherent inpaintings, particularly for masks covering more than 30% of the image.

Load-bearing premise

The quantitative advantage over partial convolution rests on Table 1's partial-convolution numbers being copied from the original partial-convolution paper instead of rerun under this paper's training data and mask protocol; if the training setups differ, the 0.2 to 0.4 dB PSNR differences may not reflect a real advantage.

Editorial extensions

If this is right

  • On irregular masks covering 30–50% of an image, LBAM reports higher PSNR and SSIM than partial convolution, contextual attention, global-and-local, and PatchMatch on the Places benchmark.
  • Because the decoder is focused on the hole via reverse attention, the model avoids wasting capacity reconstructing known regions, which the paper shows in feature visualizations and in higher-quality results on real-world object removal.
  • The learned mask update makes it feasible to train with an adversarial loss, which the paper reports is difficult with partial convolution alone; the adversarial loss trades a small drop in PSNR and SSIM for better visual quality.
  • The model runs in about 70 ms per 256×256 image, roughly three times faster than the global-and-local method and five times faster than contextual attention.

Reading between the lines

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

  • Editorial inference: the same learnable-attention recipe could be applied to other masked-input tasks such as image extrapolation, deblurring with structured masks, or video inpainting, where mask confidence needs to propagate across time.
  • Editorial inference: because the attention maps are layer-specific and trained end-to-end, the architecture might transfer to free-form editing with user-provided masks without retraining the mask-update rules, though the paper does not test this.
  • Editorial inference: the asymmetric Gaussian activation suggests the network learns a smooth confidence profile around hole boundaries; this could be probed by freezing the learned parameters and visualizing the effective receptive field of the mask branch.
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

3 major / 5 minor

Summary. The paper introduces learnable bidirectional attention maps (LBAM) for image inpainting with irregular holes. The core idea is to generalize partial convolution (PConv) by replacing handcrafted mask renormalization and hard 0-1 mask updating with learned attention maps g_A and mask-updating functions g_M, and by introducing reverse attention maps on the decoder so that the decoder focuses on filling holes rather than reconstructing known regions. The authors show that, when bias is removed, the mask-convolution step of PConv can be replaced by standard convolution because the features at unfilled mask positions are zero by construction (Eqs. 6-7). They adopt a U-Net with forward attention on the encoder and reverse attention on the decoder, trained with L1, WGAN-GP, perceptual, and style losses. Experiments on Paris StreetView and a 10-category subset of Places365 compare against PatchMatch, Global&Local, Context Attention, and PConv, including a user study in which LBAM is preferred 63.2% of the time. Ablations show the effects of the learned attention maps, reverse attention, activation functions, and the adversarial loss. The authors acknowledge failure cases in the supplementary material and promise to release code and models.

Significance. If the quantitative results are reproducible, LBAM is a useful incremental improvement over partial convolution for irregular-hole inpainting. The paper's analytical observation that PConv without bias reduces to standard convolution plus learned renormalization is elegant and likely to be reused. The systematic ablation, including the comparison of activation functions and the reverse attention maps, provides a good picture of what the components contribute. The user study and the explicit failure cases add credibility to the perceptual claims. The main weakness is that the central quantitative comparison with PConv relies on numbers copied from the original paper under a different protocol, so the claimed margins (0.2-0.4 dB PSNR) may not be significant or even real.

major comments (3)
  1. [Table 1] The PConv* comparison is not protocol-matched. The caption states that PConv* results are taken from [17], but the present paper trains on 50,000 images from 10 Places categories and tests on a different 500-image-per-category split, generates 18,000 random masks mixed with 12,000 masks from [17], and optimizes the objective in Eq. (18) with WGAN-GP and loss weights λ1=1, λ2=0.1, λ3=0.05, λ4=120. The original PConv was trained on the full Places2 dataset with different loss weights (e.g., L_hole + 6*L_valid + 0.05*L_perceptual + 120*L_style + 0.1*L_TV) and its own mask sampling. Since the PSNR margins are only 0.19-0.42 dB and the mean l1 is worse for LBAM at the two lowest mask-ratio bins (1.12 vs 1.09 and 1.93 vs 1.88), these protocol differences are large enough to reverse the ranking. The authors should rerun PConv under their exact training and evaluation protocol, or provide evidence that PConv's numbers are insensitive to the dataset subset, mask distribution, and loss changes.
  2. [Section 4.2, Table 2] The ablation studies do not isolate the contribution of the bidirectional attention maps against a same-protocol PConv baseline. The variants Ours(unlearned) and Ours(3×3) use fixed 1/16 or 3×3 mask filters with hard activation functions, which approximate PConv-like behavior, but they share the paper's U-Net backbone (14 layers, 4×4 stride-2 kernels, no bias, batch normalization) and training loss. It is therefore unclear how much of the gain over PConv in Table 1 is due to the attention modules rather than to backbone, data, or loss changes. Please add a direct PConv control trained and evaluated under the same protocol, or argue explicitly why the current variants control for the relevant variables.
  3. [Section 4.1, Tables 1 and 2] The quantitative claims lack error bars and statistical testing. The reported PSNR/SSIM values are single estimates on one test set, with margins as small as 0.002 SSIM or 0.19 dB PSNR. Without standard deviations across multiple runs or bootstrap confidence intervals, it is not possible to determine whether the observed differences are meaningful. This is particularly important after a protocol-matched PConv rerun is obtained, since the central claim of superiority over PConv rests on these small margins.
minor comments (5)
  1. [Section 4.1] The sentence 'we generate 18,000 masks with random shape, and 12,000 masks from [17] for training and testing' is ambiguous about whether the same mask set is used for both training and testing and how the masks are partitioned; please clarify.
  2. [Section 4.2, Figure 11 caption] The caption for Figure 11 lists '(e) 3×3 filter' and then also '(e) our full LBAM model'; the second entry should be labeled '(d)' for the 3×3 variant and '(e)' for the full model.
  3. [Section 3.5] The paper reports λ4=120 for the style loss, which is an order of magnitude larger than typical style-loss weights in inpainting papers; a brief sentence on how these weights were selected or their sensitivity would be useful.
  4. [Section 4.2] The text acknowledges that Ours(w/o Ladv) outperforms Ours(full) on PSNR/SSIM and justifies the adversarial loss only by qualitative inspection of the supplementary figures; given that the user study supports the qualitative claim, this trade-off between distortion metrics and perceptual quality should be stated more explicitly in the main text.
  5. [Related Work and Experiments] The paper cites gated convolution [35] in the related work but does not include it as an experimental baseline; since that method also targets irregular holes and was available at the time of submission, a sentence explaining its omission or adding the comparison would strengthen the state-of-the-art claim.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the LBAM derivation is self-contained and validated against external benchmarks.

full rationale

The paper does not derive its central claim from its own assumptions in a circular way. The learnable bidirectional attention map is defined as a generalization of partial convolution (Eqs. 6-11), with learnable mask-updating and attention activations trained end-to-end using standard reconstruction, perceptual, style, and adversarial losses (Eqs. 14-18). The claimed improvement over prior methods is supported by quantitative comparisons on Paris StreetView and Places and by a user study. Self-citations to [33] and [37] appear only in related work and are not load-bearing for the main claim. The only notable concern is Table 1's caption, 'Results of PConv* are taken from [17],' meaning the PConv baseline was not rerun under this paper's training and mask protocol; this is a comparison-fairness issue, not a circularity issue, because the paper's own method is not defined in terms of the borrowed baseline numbers and no fitted parameter is relabeled as a prediction. The derivation chain is therefore self-contained against external benchmarks, with no reduction of a predicted result to its inputs by construction.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The ledger contains hand-set hyperparameters (alpha, loss weights) and domain assumptions about masks, VGG losses, and the U-Net backbone. No new physical entities are introduced; the learnable attention maps are network modules rather than invented entities.

free parameters (2)
  • alpha in mask update gM = 0.8
    Hand-set hyperparameter in Eqn. 8 controlling softness of mask updating; affects all forward and reverse attention layers.
  • loss weights (lambda1, lambda2, lambda3, lambda4) = 1, 0.1, 0.05, 120
    Hand-tuned weights in Eqn. 18; the large style weight and small adversarial weight shape the final output quality.
assumptions (4)
  • domain assumption Random-shaped binary masks used in training are representative of the irregular holes seen at test time.
    The method is trained with 18,000 random masks plus 12,000 masks from PConv [17]; success depends on this distributional match.
  • domain assumption Pre-trained VGG-16 features provide a perceptually meaningful loss for inpainting.
    Perceptual and style losses in Eqns. 15-16 rely on ImageNet-trained VGG features being good proxies for human quality.
  • domain assumption The U-Net encoder-decoder with skip connections is a sound backbone for hole filling.
    The architecture is borrowed from [11], and the paper modifies it with attention maps rather than deriving it.
  • standard math Standard CNN operations and backpropagation are assumed.
    The derivations in Section 3 assume convolution is linear and gradients are computable; no formal proof is given.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Image Inpainting with Learnable Bidirectional Attention Maps." pith.science (2026). https://pith.science/paper/NDPKGLXH

@misc{pith2026190900968,
  author       = {Pith},
  title        = {Pith review of: Image Inpainting with Learnable Bidirectional Attention Maps},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NDPKGLXH}},
  note         = {Machine review of arXiv:1909.00968}
}
read the original abstract

Most convolutional network (CNN)-based inpainting methods adopt standard convolution to indistinguishably treat valid pixels and holes, making them limited in handling irregular holes and more likely to generate inpainting results with color discrepancy and blurriness. Partial convolution has been suggested to address this issue, but it adopts handcrafted feature re-normalization, and only considers forward mask-updating. In this paper, we present a learnable attention map module for learning feature renormalization and mask-updating in an end-to-end manner, which is effective in adapting to irregular holes and propagation of convolution layers. Furthermore, learnable reverse attention maps are introduced to allow the decoder of U-Net to concentrate on filling in irregular holes instead of reconstructing both holes and known regions, resulting in our learnable bidirectional attention maps. Qualitative and quantitative experiments show that our method performs favorably against state-of-the-arts in generating sharper, more coherent and visually plausible inpainting results. The source code and pre-trained models will be available.

Figures

Figures reproduced from arXiv: 1909.00968 by the authors.

Figure 1
Figure 1. Qualitative comparison of inpainting results by PatchMatch (PM) [ [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Interplay models between mask and intermediate feature for PConv and our learnable bidirectional attention maps. Here, the [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. The network architecture of our model. The circle with triangle inside denotes operation form of Eqn.( [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (13 more)
Figure 4
Figure 4. Figure 4: Qualitative comparison on Paris StreetView dataset. Comparison with PatchMatch (PM) [ [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Qualitative comparison on Places dataset. Comparison with PatchMatch (PM) [ [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Results on real-world images. From left to right are: original image, input with objects masked (white area), Context Attention [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Visualization of features from the first encoder layer and [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: Visualization of updated masks after activation function [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 9
Figure 9. Figure 9: Visual quality comparison of the effect on the learnable [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]
Figure 10
Figure 10. Figure 10: Visual comparison of variants (i) to (iii) of our LBAM model. From left to right are: Input, (a) Ours with unlearned model, (b) [PITH_FULL_IMAGE:figures/full_fig_p012_10.png]
Figure 11
Figure 11. Figure 11: Visual comparison of different activation functions or [PITH_FULL_IMAGE:figures/full_fig_p012_11.png]
Figure 12
Figure 12. Figure 12: Failure cases of our LBAM model. Each group is ordered as input image, our result and ground truth. All images are scaled to [PITH_FULL_IMAGE:figures/full_fig_p013_12.png]
Figure 13
Figure 13. Figure 13: Qualitative comparison on Paris StreetView dataset. Comparison with PatchMatch (PM) [ [PITH_FULL_IMAGE:figures/full_fig_p015_13.png]
Figure 14
Figure 14. Figure 14: Qualitative comparison on Paris StreetView dataset. Comparison with PatchMatch (PM) [ [PITH_FULL_IMAGE:figures/full_fig_p016_14.png]
Figure 15
Figure 15. Figure 15: Qualitative comparison on Places dataset. Comparison with PatchMatch (PM) [ [PITH_FULL_IMAGE:figures/full_fig_p017_15.png]
Figure 16
Figure 16. Figure 16: Results of our LBAM on object removal task of real world images. All images are scaled to [PITH_FULL_IMAGE:figures/full_fig_p018_16.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 37 canonical work pages

  1. [17]

    Reda, Kevin Shih, Ting-Chun Wang, Andrew Tao, and Bryan Catanzaro

    Guilin Liu, Fitsum A. Reda, Kevin Shih, Ting-Chun Wang, Andrew Tao, and Bryan Catanzaro. Image inpainting for ir- regular holes using partial convolutions. In The European Conference on Computer Vision (ECCV) , volume 11215, pages 89–105, 2018. 1, 2, 3, 4, 5, 6, 7, 11, 15, 16, 17

    work page 2018

  2. [1]

    Wasserstein generative adversarial networks

    Martin Arjovsky, Soumith Chintala, and L ´eon Bottou. Wasserstein generative adversarial networks. In Interna- tional Conference on Machine Learning (ICML), pages 214– 223, 2017. 6

    work page 2017

  3. [2]

    PatchMatch: A randomized correspon- dence algorithm for structural image editing

    Connelly Barnes, Eli Shechtman, Adam Finkelstein, and Dan B Goldman. PatchMatch: A randomized correspon- dence algorithm for structural image editing. ACM Trans- actions on Graphics (TOG) , pages 24:1–24:11, 2009. 1, 2, 5, 6, 7, 11, 15, 16, 17

    work page 2009

  4. [3]

    Image inpainting

    Marcelo Bertalmio, Guillermo Sapiro, Vincent Caselles, and Coloma Ballester. Image inpainting. In Annual Conference on Computer Graphics and Interactive Techniques (SIG- GRAPH), pages 417–424, 2000. 1

    work page 2000

  5. [4]

    Re- gion filling and object removal by exemplar-based image in- painting

    Antonio Criminisi, Patrick Perez, and Kentaro Toyama. Re- gion filling and object removal by exemplar-based image in- painting. IEEE Transactions on Image Processing (TIP) , pages 1200–1212, 2004. 2

    work page 2004

  6. [5]

    A note on the gradient of a multi-image

    Silvano Di Zenzo. A note on the gradient of a multi-image. Computer Vision, Graphics, and Image Processing , pages 116–125, 1986. 2

    work page 1986

  7. [6]

    What makes paris look like paris?Com- munications of the ACM, pages 103–110, 2015

    Carl Doersch, Saurabh Singh, Abhinav Gupta, Josef Sivic, and Alexei A Efros. What makes paris look like paris?Com- munications of the ACM, pages 103–110, 2015. 2, 6, 11

    work page 2015

  8. [7]

    Gatys, Alexander S

    Leon A. Gatys, Alexander S. Ecker, and Matthias Bethge. Image style transfer using convolutional neural networks. In IEEE Conference on Computer Vision and Pattern Recogni- tion (CVPR), pages 2414–2423, 2016. 5

    work page 2016

Show all 42 references
  1. [8]

    Generative adversarial nets

    Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. Generative adversarial nets. In Advances in Neural Information Processing Systems (NeurIPS), pages 2672–2680, 2014. 1, 5, 6

    2014

  2. [9]

    Improved training of wasserstein gans

    Ishaan Gulrajani, Faruk Ahmed, Martin Arjovsky, Vincent Dumoulin, and Aaron Courville. Improved training of wasserstein gans. In Advances in Neural Information Pro- cessing Systems (NeurIPS), pages 5767–5777, 2017. 6

    2017

  3. [10]

    Globally and locally consistent image completion

    Satoshi Iizuka, Edgar Simo-Serra, and Hiroshi Ishikawa. Globally and locally consistent image completion. ACM Transactions on Graphics (TOG) , pages 107:1–107:14,

  4. [11]

    Image-to-image translation with conditional adversar- ial networks

    Phillip Isola, Jun-Yan Zhu, Tinghui Zhou, and Alexei A Efros. Image-to-image translation with conditional adversar- ial networks. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 5967–5976, 2017. 4, 5

    2017

  5. [12]

    Percep- tual losses for real-time style transfer and super-resolution

    Justin Johnson, Alexandre Alahi, and Li Fei-Fei. Percep- tual losses for real-time style transfer and super-resolution. In The European Conference on Computer Vision (ECCV) , volume 9906, pages 694–711, 2016. 5

    2016

  6. [13]

    Kingma and Jimmy Ba

    Diederik P. Kingma and Jimmy Ba. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980 ,

    arXiv

  7. [14]

    Mask-specific inpainting with deep neural networks

    Rolf K ¨ohler, Christian Schuler, Bernhard Sch ¨olkopf, and Stefan Harmeling. Mask-specific inpainting with deep neural networks. In Pattern Recognition (GCPR), pages 523–534,

  8. [15]

    Image completion using efficient belief propagation via priority scheduling and dynamic pruning

    Nikos Komodakis and Georgios Tziritas. Image completion using efficient belief propagation via priority scheduling and dynamic pruning. IEEE Transactions on Image Processing (TIP), pages 2649–2661, 2007. 2

    2007

  9. [16]

    Aitken, Alykhan Tejani, Johannes Totz, Zehan Wang, and Wenzhe Shi

    Christian Ledig, Lucas Theis, Ferenc Huszar, Jose Caballero, Andrew P. Aitken, Alykhan Tejani, Johannes Totz, Zehan Wang, and Wenzhe Shi. Photo-realistic single image super- resolution using a generative adversarial network. In IEEE Conference on Computer Vision and Pattern Re...

    2017

  10. [18]

    Examplar-based inpainting based on local geometry

    Olivier Le Meur, Josselin Gautier, and Christine Guillemot. Examplar-based inpainting based on local geometry. InIEEE International Conference on Image Processing (ICIP), pages 3401–3404, 2011. 1, 2

    2011

  11. [19]

    Edgeconnect: Generative image in- painting with adversarial edge learning

    Kamyar Nazeri, Eric Ng, Tony Joseph, Faisal Qureshi, and Mehran Ebrahimi. Edgeconnect: Generative image in- painting with adversarial edge learning. arXiv preprint arXiv:1901.00212, 2019. 3

    1901 arXiv

  12. [20]

    Context encoders: Feature learn- ing by inpainting

    Deepak Pathak, Philipp Kr ¨ahenb¨uhl, Jeff Donahue, Trevor Darrell, and Alexei Efros. Context encoders: Feature learn- ing by inpainting. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , pages 2536–2544, 2016. 1, 2, 6

    2016

  13. [21]

    Shep- ard convolutional neural networks

    Jimmy SJ Ren, Li Xu, Qiong Yan, and Wenxiu Sun. Shep- ard convolutional neural networks. In Advances in Neural Information Processing Systems (NeurIPS), pages 901–909,

  14. [22]

    U-net: Convolutional networks for biomedical image segmentation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U-net: Convolutional networks for biomedical image segmentation. In Medical Image Computing and Computer-Assisted Inter- vention (MICCAI), volume 9351, pages 234–241, 2015. 2

    2015

  15. [23]

    Berg, and Li Fei-Fei

    Olga Russakovsky, Jia Deng, Hao Su, Jonathan Krause, San- jeev Satheesh, Sean Ma, Zhiheng Huang, Andrej Karpathy, Aditya Khosla, Michael Bernstein, Alexander C. Berg, and Li Fei-Fei. ImageNet Large Scale Visual Recognition Chal- lenge. International Journal of Computer Vision ...

    2015

  16. [24]

    Improved techniques for training gans

    Tim Salimans, Ian Goodfellow, Wojciech Zaremba, Vicki Cheung, Alec Radford, Xi Chen, and Xi Chen. Improved techniques for training gans. InAdvances in Neural Informa- tion Processing Systems (NeurIPS), pages 2234–2242, 2017. 6

    2017

  17. [25]

    Very deep con- volutional networks for large-scale image recognition

    Karen Simonyan and Andrew Zisserman. Very deep con- volutional networks for large-scale image recognition. In In- ternational Conference on Learning Representations (ICLR),

  18. [26]

    Contextual Based Image Inpainting: Infer, Match and Translate

    Yuhang Song, Chao Yang, Zhe Lin, Xiaofeng Liu, Hao Li, and Qin Huang. Contextual Based Image Inpainting: Infer, Match and Translate. In The European Conference on Com- puter Vision (ECCV), volume 11206, pages 3–18, 2018. 1, 3

    2018

  19. [27]

    Condi- tional image generation with pixelcnn decoders

    Aaron van den Oord, Nal Kalchbrenner, Lasse Espeholt, ko- ray kavukcuoglu, Oriol Vinyals, and Alex Graves. Condi- tional image generation with pixelcnn decoders. InAdvances in Neural Information Processing Systems (NeurIPS), pages 4790–4798, 2016. 6

    2016

  20. [28]

    Image inpainting via generative multi-column convolu- tional neural networks

    Yi Wang, Xin Tao, Xiaojuan Qi, Xiaoyong Shen, and Jiaya Jia. Image inpainting via generative multi-column convolu- tional neural networks. In Advances in Neural Information Processing Systems (NeurIPS), pages 329–338, 2018. 3

    2018

  21. [29]

    Brostow, Ben Tordoff, and Roberto Cipolla

    Marta Wilczkowiak, Gabriel J. Brostow, Ben Tordoff, and Roberto Cipolla. Hole filling through photomontage. In British Machine Vision Conference (BMVC), pages 492–501,

  22. [30]

    Image denois- ing and inpainting with deep neural networks

    Junyuan Xie, Linli Xu, and Enhong Chen. Image denois- ing and inpainting with deep neural networks. In Advances in Neural Information Processing Systems (NeurIPS), pages 341–349, 2012. 2

    2012

  23. [31]

    Foreground-aware image in- painting

    Wei Xiong, Jiahui Yu, Zhe Lin, Jimei Yang, Xin Lu, Con- nelly Barnes, and Jiebo Luo. Foreground-aware image in- painting. In IEEE Conference on Computer Vision and Pat- tern Recognition (CVPR), 2019. 3

    2019

  24. [32]

    Image inpainting by patch prop- agation using patch sparsity

    Zongben Xu and Jian Sun. Image inpainting by patch prop- agation using patch sparsity. IEEE Transactions on Image Processing (TIP), pages 1153–1165, 2010. 1, 2

    2010

  25. [33]

    Shift-net: Image inpainting via deep feature rearrangement

    Zhaoyi Yan, Xiaoming Li, Mu Li, Wangmeng Zuo, and Shiguang Shan. Shift-net: Image inpainting via deep feature rearrangement. In The European Conference on Computer Vision (ECCV), volume 11218, pages 3–19, 2018. 1, 3, 4, 6

    2018

  26. [34]

    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 IEEE Conference on Com- puter Vision and Pattern Recognition (CVPR), pages 4076– 4084, 2017. 1, 3, 6

    2017

  27. [35]

    Free-form image inpainting with gated convolution

    Jiahui Yu, Zhe Lin, Jimei Yang, Xiaohui Shen, Xin Lu, and Thomas S Huang. Free-form image inpainting with gated convolution. arXiv preprint arXiv:1806.03589, 2018. 3

    2018 arXiv

  28. [36]

    Jiahui Yu, Zhe Lin, Jimei Yang, Xiaohui Shen, Xin Lu, and Thomas S. Huang. Generative image inpainting with contex- tual attention. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 5505–5514, 2018. 1, 2, 3, 5, 6, 7, 11, 15, 16, 17

    2018

  29. [37]

    Semantic image inpainting with pro- gressive generative networks

    Haoran Zhang, Zhenzhen Hu, Changzhi Luo, Wangmeng Zuo, and Meng Wang. Semantic image inpainting with pro- gressive generative networks. In ACM International Confer- ence on Multimedia (ACM MM), pages 1939–1947, 2018. 3

    1939

  30. [38]

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

    Han Zhang, Tao Xu, Hongsheng Li, Shaoting Zhang, Xiao- gang Wang, Xiaolei Huang, and Dimitris Metaxas. Stackgan: Text to photo-realistic image synthesis with stacked genera- tive adversarial networks. In IEEE International Conference on Computer Vision (ICCV), pages 5908–5916, 2017. 6

    2017

  31. [39]

    Pluralistic image completion

    Chuanxia Zheng, Tat-Jen Cham, and Jianfei Cai. Pluralistic image completion. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 1438–1447, 2019. 3

    2019

  32. [40]

    Places: A 10 million image database for scene recognition

    Bolei Zhou, Agata Lapedriza, Aditya Khosla, Aude Oliva, and Antonio Torralba. Places: A 10 million image database for scene recognition. IEEE Transactions on Pattern Anal- ysis and Machine Intelligence (TPAMI) , pages 1452–1464,

  33. [42]

    9) for activation the attention map and the modified activation function (Eqn

    2, 6, 7, 11 Supplementary Material Visual comparison of several LBAM variants on Paris StreetView dataset We implement our bidirectional attention maps by em- ploying an asymmetric Gaussian shaped form (Eqn. 9) for activation the attention map and the modified activation functi...

  34. [2017]

    1, 2, 3, 5, 6, 7, 8, 11, 15, 16, 17

Pith tools

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