F2T2-HiT: A U-Shaped FFT Transformer and Hierarchical Transformer for Reflection Removal

Chiu Man Ho; Huiming Sun; Jiaming Ding; Jie Cai; Kangning Yang; Lan Fu; Ling Ouyang; Zibo Meng

REVIEW 3 major objections 5 minor 37 references

F2T2-HiT: A U-Shaped FFT Transformer and Hierarchical Transformer for Reflection Removal

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read F2T2-HiT, a U-shaped network pairing FFT Transformer blocks with hierarchical window Transformer blocks, reports state-of-the-art reflection removal, averaging 25.57 dB PSNR across the Nature, Real, and SIR2 benchmarks.

desk verdict The architecture is a sensible repackaging of known blocks, but the paper's own Table 1 undercuts its state-of-the-art claim, and the Real/Nature training overlap is left unresolved. read the letter →

arxiv 2506.05489 v1 pith:YJRKDQON submitted 2025-06-05 cs.CV

Jie Cai , Kangning Yang , Ling Ouyang , Lan Fu , Jiaming Ding , Huiming Sun , Chiu Man Ho , Zibo Meng This is my paper

classification cs.CV

keywords reflectionremovalsingleimagetransformerfastFouriertransformhierarchicalattentionrestorationU-Netfrequencydomain

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

F2T2-HiT is a U-shaped network for removing reflections from a single photograph taken through glass. The paper's claim is that replacing the plain blocks of a NAFNet-style encoder–decoder with two custom Transformer blocks—an FFT Transformer block that reads global frequency-domain structure and a hierarchical window Transformer block that gathers multi-scale local detail—gives the best average performance yet reported on three real-world reflection benchmarks. The headline number is 25.57 dB average PSNR over 494 test images from the Nature, Real, and SIR2 datasets. If true, this would give photographers and downstream vision systems a practical way to recover the transmitted scene behind glass from one image alone.

What carries the argument

The central objects are the two named blocks. The F2T2 (Fast Fourier Transform Transformer) block follows LayerNorm – FFT layer – LayerNorm – multi-kernel convFFN; its FFT layer runs a spatial-domain branch (dilated convolutions) and a frequency-domain branch (2D FFT, frequency-conditional positional embedding, and frequency dynamic convolution) and fuses them with channel attention, giving the network an image-wide receptive field early on. The HiT (Hierarchical Transformer) block follows LayerNorm – window self-attention – LayerNorm – channel-wise FFN, where the window self-attention uses expanding windows of sizes 4, 8, and 16 and computes spatial and channel self-correlations with linear complexity in window size. These blocks replace the convolutional blocks of a NAFNet U-shaped encoder–decoder, which is what lets the paper test the blocks' individual contributions in ablation.

What would settle it

A decisive check is to verify the split: list the file names of the Real and Nature test images and check them, plus near-duplicates, against the 89 and 200 real training pairs. If any overlap exists and removing it drops the average PSNR below the 25.40 dB of the strongest prior method, the state-of-the-art claim rests on memorization rather than the architecture.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that a single-stage U-shaped architecture does not need bespoke reflection priors or multi-stage refinement to be state of the art: it needs a global receptive field from the first layers and attention that operates at several window scales. The F2T2 block supplies the first property by mixing a spatial-domain branch with a frequency-domain branch built on 2D FFT and frequency-conditional positional encoding, so every feature can see the whole image; the HiT block supplies the second by partitioning features into windows of sizes 4, 8, and 16 and aggregating them with spatial and channel self-correlation. Combined inside a NAFNet U-Net with skip connections, the model reports an average PSNR of 25.57 dB across the three benchmarks, with per-dataset values of 26.08 dB on Nature, 21.64 dB on Real, and 25.72 dB on SIR2. The authors interpret the ablation results—NAFNet alone at 23.97 dB, plus HiT at 25.00 dB, plus F2T2 at 25.57 dB—as evidence that each block carries a distinct part of the improvement.

Load-bearing premise

The argument for state-of-the-art generalization assumes that no image from the Real or Nature test benchmarks was included in the 89+200 real training pairs, but the paper only says these datasets contain both training and test data and never states that the test images were left out.

Editorial extensions

If this is right

A single trained F2T2-HiT model can be applied directly to three different real-world benchmarks without per-dataset tuning, and the paper reports it beats all listed prior methods on the average PSNR over 494 images.
Large-area reflections, which local receptive fields miss, become addressable because the FFT layers give even the early encoder layers a view of the whole image.
The hierarchical windows let the model keep local edge detail while still forming long-range dependencies, so reflections of mixed sizes can be handled in one pass.
The ablations show the gains are additive: replacing NAFNet blocks with HiT adds about 1 dB on average, and adding F2T2 adds about 0.6 dB more.
Training takes roughly 24 hours on eight A100 GPUs, indicating the architecture is practical to reproduce at academic scale.

Reading between the lines

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

Because the SIR2 subset contributes 454 of the 494 averaged images, the average PSNR is dominated by that benchmark; a method can top the average while being weaker on the small Real set, which is what the table shows, so per-benchmark numbers rather than the average should drive method comparisons.
The same dual-domain recipe—global frequency context plus hierarchical windows—could transfer to other single-image restoration tasks where artifacts span large areas, such as deraining, dehazing, or glare removal, though the paper does not test those.
A strict test of the generalization claim would retrain with every Real and Nature test image explicitly excluded and report the resulting PSNR; the current paper does not state that this exclusion was done.

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.

Desk Editor's Note

read the letter

I'll give you the short version: this is a workmanlike architecture paper whose headline claim doesn't survive contact with its own numbers. The authors take NAFNet's U-Net, drop in HiT and F2T2 blocks from super-resolution and restoration work, and test on the standard reflection-removal benchmarks. That combination is new for SIRR, and the ablations in Table 2 show each addition helps on most rows. Good: the writing is clear, the block descriptions are adequate, and the work is properly situated in the literature. No code is released, but that's not fatal.

The soft spots are real and central. First, the 'state-of-the-art' claim in the abstract and conclusion doesn't match Table 1. On the Real benchmark, your method is more than 2 dB behind Zhu et al. (21.64 vs 23.82), and the average SSIM is lower (0.894 vs 0.904). The only metric where you lead is average PSNR, by 0.17 dB, and that's driven by the 454-image SIR2 subset. With no error bars and no significance test, that's not a convincing lead.

Second, and more worrying: Section 4.2 says the training set includes 89 real pairs from Zhang et al. and 200 from IBCLN, and then says, without apparent irony, that 'Real and Nature include both training and test data.' The paper never states that the specific test images were held out. If they weren't, the reported numbers on Real and Nature are memorization, not generalization. The authors flag the overlap but don't resolve it. That's exactly the kind of thing a critical reader is entitled to demand before accepting the performance claims.

The ablation study is well-formed and does show a consistent gain from the HiT and F2T2 blocks, so the architectural contribution is probably real, just incremental. The citation pattern is fine; the blocks are properly attributed to [19] and [20].

Net: a competent but incremental engineering paper whose evaluation is not trustworthy as presented. A serious referee would need to see a clear statement of how Real and Nature test images were excluded from training, plus a re-run of Table 1 with error bars and a fair comparison on the terms the authors themselves define. As-is, I wouldn't send this to peer review; the central empirical claim is contradicted by the paper's own data. If you want an example of how not to report train/test overlap, this is a good case study for the reading group.

Referee Report

3 major / 5 minor

Summary. The paper proposes F2T2-HiT, a U-shaped network for single-image reflection removal that combines FFT-based Transformer blocks and hierarchical window Transformer blocks inside a NAFNet-style encoder-decoder. The training set includes 89 real pairs from [30], 200 real pairs from [28], and synthesized data. Evaluation is reported on the Real, Nature, and SIR2 benchmarks, and Table 1 reports the highest average PSNR (25.57 dB) among the compared methods. The paper's central claim is that this constitutes state-of-the-art performance, supported by the quantitative comparison in Table 1 and the ablations in Table 2.

Significance. If the reported results were obtained under a clean and properly split evaluation protocol, the architecture would be a plausible contribution: it combines two existing Transformer designs (F2T2 and HiT blocks) within a well-known UNet/NAFNet framework, and the ablation shows monotone gains from adding each block. However, the empirical evaluation as written does not establish the headline claim. The unresolved overlap between the training set and two of the three test benchmarks, together with a comparison table that contradicts the 'state-of-the-art' conclusion, mean the paper's main contribution is not supported by its evidence. The paper does not provide code, model size, FLOPs, or significance tests, so the efficiency and superiority claims are also not fully documented.

major comments (3)

[Section 4.2] The training and test sets are not shown to be disjoint. The text states that 'Real and Nature include both training and test data' while the training set is described as including 89 real pairs from [30] and 200 real pairs from [28]. Since Real and Nature are benchmarks drawn from the same sources as these training pairs, the results on those benchmarks cannot be interpreted as generalization unless the test images were explicitly excluded from training and the split was documented. No such exclusion is stated anywhere in the manuscript. Tables 1 and 2 therefore contain numbers on Nature and Real that may reflect memorization of the training distribution rather than reflection-removal ability.
[Table 1] The state-of-the-art claim is not supported by the paper's own comparison. On Real(20), the proposed method is substantially worse than Zhu et al. [15] in PSNR (21.64 vs. 23.82). On SSIM, the proposed method is lower than Zhu et al. on Nature (0.837 vs. 0.843), on SIR2 (0.903 vs. 0.910), and on the 494-image average (0.894 vs. 0.904). The only metric that favors the method is average PSNR (25.57 vs. 25.40), a 0.17 dB gap that is dominated by the 454-image SIR2 subset and is reported without error bars or statistical significance tests. The unqualified 'state-of-the-art performance' in the abstract and conclusion therefore does not follow from the evidence.
[Section 4.3 / Table 2] The ablation study inherits the same evaluation-protocol problem. Because the Real and Nature portions of the test set may overlap with the training data, the incremental gains from the HiT and F2T2 blocks shown in Table 2 cannot be attributed to the architectural designs rather than to overfitting of those training examples. Without a clean disjoint split, the ablation does not establish the effectiveness of either proposed block.

minor comments (5)

[Section 4.2] The sentence 'Real and Nature include both training and test data' needs to be replaced with a precise statement of how many test images there are, which sources they come from, and how they were held out from the 89+200 real training pairs.
[Figure 1] The caption contains the typo 'architecutre'; also, '1x1, conv' and similar shorthand should be expanded or defined in the figure caption for clarity.
[Section 4.1] The paper claims computational efficiency in the introduction and conclusion, but no model parameter count, FLOPs, inference time, or memory usage is reported anywhere in Section 4.
[Conclusion] The closing 'state-of-the-art results' phrasing should be qualified or removed until the evaluation protocol is corrected and the comparison with Zhu et al. [15] is reconciled with the numbers in Table 1.
[References] References [6]-[10] are largely self-citations of preprints and workshop papers; the authors should clarify the relationship of the current work to these papers and provide peer-reviewed versions where available.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the architecture is composed from cited prior blocks and compared on benchmarks; the acknowledged Real/Nature train-test overlap is a data-hygiene concern, not a demonstrated circular reduction.

full rationale

The paper's derivation chain is architectural composition, not a formal derivation from inputs. F2T2 and HiT blocks are explicitly attributed to prior work ([19], [20]) and NAFNet is cited as the UNet backbone; no load-bearing claim is justified by the authors' own citations. The one statement that could suggest circularity is 'Real and Nature include both training and test data' (Sec. 4.2), since the training set also draws real pairs from [30] and [28], the source datasets of those benchmarks. However, the paper does not state that the specific test images were used in training, and the acknowledged overlap is a standard description of the benchmark partitions; without an exhibit showing that the test images are the fitted training pairs, this is a possible data-leakage/correctness risk, not a self-definitional reduction. The 'state-of-the-art' claim is weakened by Table 1 (Ours loses on Real PSNR and on most SSIM metrics), but under-support is an empirical-evidence problem, not circularity. Self-citations [6]-[10] are contextual and not used to justify the architecture. No step of the claimed derivation is equivalent to its inputs by construction.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The central claim rests mainly on the evaluation protocol. The most important axiom is that the training and test sets are disjoint, which the paper's own dataset description calls into question. The method itself is a standard deep network with no new physical or mathematical entities.

free parameters (2)

Hierarchical window sizes = 4, 8, 16
Hand-chosen for the HiT block; no ablation over window sizes is reported.
Scheduler restart weights = 1, 0.5, 0.25
Training hyperparameters from the cosine annealing restart scheme, not derived from the task.

assumptions (3)

domain assumption Training and test sets are disjoint
Section 4.2 lists training data from [30] and [28] while evaluating on Real and Nature from the same sources. The paper notes these datasets include both training and test data but does not state that test images were excluded.
domain assumption PSNR/SSIM are sufficient evaluation metrics
The paper relies solely on these metrics without perceptual studies or error bars.
domain assumption Synthetic data from PASCAL VOC and RRW transfers to real reflections
Training uses synthesized pairs generated as in [14]; no analysis of the domain gap is provided.

how reviews work

0 comments

Cite this review

Pith. "Pith review of F2T2-HiT: A U-Shaped FFT Transformer and Hierarchical Transformer for Reflection Removal." pith.science (2026). https://pith.science/paper/YJRKDQON

@misc{pith2026250605489,
  author       = {Pith},
  title        = {Pith review of: F2T2-HiT: A U-Shaped FFT Transformer and Hierarchical Transformer for Reflection Removal},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YJRKDQON}},
  note         = {Machine review of arXiv:2506.05489}
}

read the original abstract

Single Image Reflection Removal (SIRR) technique plays a crucial role in image processing by eliminating unwanted reflections from the background. These reflections, often caused by photographs taken through glass surfaces, can significantly degrade image quality. SIRR remains a challenging problem due to the complex and varied reflections encountered in real-world scenarios. These reflections vary significantly in intensity, shapes, light sources, sizes, and coverage areas across the image, posing challenges for most existing methods to effectively handle all cases. To address these challenges, this paper introduces a U-shaped Fast Fourier Transform Transformer and Hierarchical Transformer (F2T2-HiT) architecture, an innovative Transformer-based design for SIRR. Our approach uniquely combines Fast Fourier Transform (FFT) Transformer blocks and Hierarchical Transformer blocks within a UNet framework. The FFT Transformer blocks leverage the global frequency domain information to effectively capture and separate reflection patterns, while the Hierarchical Transformer blocks utilize multi-scale feature extraction to handle reflections of varying sizes and complexities. Extensive experiments conducted on three publicly available testing datasets demonstrate state-of-the-art performance, validating the effectiveness of our approach.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

37 extracted references · 33 canonical work pages

[19]

Single image reflection separation via component synergy,

Q. Hu and X. Guo, “Single image reflection separation via component synergy,” in ICCV, 2023, pp. 13138–13147

work page 2023
[20]

Revisiting single image reflection removal in the wild,

Y . Zhu, X. Fu, P. T. Jiang, H. Zhang, Q. Sun, J. Chen, Z. Zha, and B. Li, “Revisiting single image reflection removal in the wild,” in CVPR, 2024, pp. 25468–25478

work page 2024
[30]

Single image reflection removal beyond linearity,

Q. Wen, Y . Tan, J. Qin, W. Liu, G. Han, and S. He, “Single image reflection removal beyond linearity,” in CVPR, 2019, pp. 3771–3779

work page 2019
[28]

Swin transformer: Hierarchical vision transformer using shifted windows,

Z. Liu, Y . Lin, Y . Cao, H. Hu, Y . Wei, Z. Zhang, S. Lin, and B. Guo, “Swin transformer: Hierarchical vision transformer using shifted windows,” in ICCV, 2021, pp. 10012–10022

work page 2021
[15]

Degradation-aware image enhancement via vision- language classification,

J. Cai, K. Yang, J. Ding, L. Fu, L. Ouyang, J. Li, J. Shen, and Z. Meng, “Degradation-aware image enhancement via vision- language classification,” in MIPR, 2025

work page 2025
[1]

F2T2-HiT: A U-Shaped FFT Transformer and Hierarchical Transformer for Reflection Removal

INTRODUCTION The SIRR technique aims to decompose a reflection- contaminated input image I into two distinct components: the reflection layer R and the reflection-free transmission layer T . The primary objective is to ensure that the generated transmission layer T closely approximates its corresponding ground truth B, which represents the ideal reflectio...

work page Pith review arXiv 2025
[2]

ghosting

RELA TED WORK Over recent decades, numerous innovative methods have been proposed to address the problem of reflection removal. Some methods rely on additional inputs, such as multi- frames [3], polarization [4], and flash-only priors [5]. Others utilize traditional image priors, including relative smooth- ness, “ghosting” cues, and dark channels, to sepa...

work page
[3]

METHODOLOGY We first introduce the fundamental methodology of the F2T2-HiT architecture in Sec.3.1, followed by the block-level designs detailed in Sec.3.2, Sec.3.3, and Sec.3.4, respectively. 3.1. A U-Shaped F2T2-HiT Architecture To simplify inter-block complexity, we utilize a classic single-stage U-shaped architecture with skip connections, as shown in...

work page

Show all 37 references

[4]

Implementation details Our framework is implemented with the PyTorch plat- form

EXPERIMENTS 4.1. Implementation details Our framework is implemented with the PyTorch plat- form. During the training phase, the network is trained using the Adam optimizer with an initial learning rate of 0.0001, which is adjusted based on a Cosine Annealing Restart scheme. T...

2007
[5]

By integrating two key innova- tions, we demonstrate significant improvements in reflection removal performance

CONCLUSION In this paper, we propose F2T2-HIT, an innovative and efficient U-shaped Transformer-based architecture especially designed for SIRR task. By integrating two key innova- tions, we demonstrate significant improvements in reflection removal performance. First, the inc...
[6]

Simple baselines for image restoration,

L. Chen, X. Chu, X. Zhang, and J. Sun, “Simple baselines for image restoration,” in ECCV. Springer, 2022, pp. 17–33

2022
[7]

Fast fourier convolution,

L. Chi, B. Jiang, and Y . Mu, “Fast fourier convolution,” NeurIPS, vol. 33, pp. 4479–4488, 2020

2020
[8]

Learned dual-view reflection removal,

S. Niklaus, X. C. Zhang, J. T. Barron, N. Wadhwa, R. Garg, F. Liu, and T. Xue, “Learned dual-view reflection removal,” in WACV, 2021, pp. 3713–3722

2021
[9]

Polarized reflection removal with perfect alignment in the wild,

C. Lei, X. Huang, M. Zhang, Q. Yan, W. Sun, and Q. Chen, “Polarized reflection removal with perfect alignment in the wild,” in CVPR, 2020, pp. 1750–1758

2020
[10]

Robust reflection removal with flash-only cues in the wild,

C. Lei, X. Jiang, and Q. Chen, “Robust reflection removal with flash-only cues in the wild,” IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023

2023
[11]

Openrr-1k: A scalable dataset for real-world reflection removal,

K. Yang, L. Ouyang, H. Sun, J. Cai, L. Fu, J. Ding, C. M. Ho, and Z. Meng, “Openrr-1k: A scalable dataset for real-world reflection removal,” in ICIP, 2025

2025
[12]

Ntire 2025 challenge on single image reflection removal in the wild: Datasets, methods and results,

K. Yang, J. Cai, L. Ouyang, F. Vasluianu, R. Timofte, et al., “Ntire 2025 challenge on single image reflection removal in the wild: Datasets, methods and results,” in CVPR Workshops, 2025

2025
[13]

Survey on single-image reflection removal using deep learn- ing techniques,

K. Yang, H. Sun, J. Cai, L. Fu, J. Ding, J. Li, and Z. Meng, “Survey on single-image reflection removal using deep learn- ing techniques,” in MIPR, 2025

2025
[14]

Openrr-5k: A large-scale benchmark for reflection removal in the wild,

J. Cai, K. Yang, L. Ouyang, L. Fu, J. Ding, J. Shen, and Z. Meng, “Openrr-5k: A large-scale benchmark for reflection removal in the wild,” in MIPR, 2025

2025
[16]

Location-aware single image reflection removal,

Z. Dong, K. Xu, Y . Yang, H. Bao, W. Xu, and R. W. Lau, “Location-aware single image reflection removal,” in ICCV, 2021, pp. 5017–5026

2021
[17]

Robust single image reflection removal against adver- sarial attacks,

Z. Song, Z. Zhang, K. Zhang, W. Luo, Z. Fan, W. Ren, and J. Lu, “Robust single image reflection removal against adver- sarial attacks,” in CVPR, 2023, pp. 24688–24698

2023
[18]

Two-stage single image reflection removal with reflection-aware guid- ance,

Y . Li, M. Liu, Y . Yi, Q. Li, D. Ren, and W. Zuo, “Two-stage single image reflection removal with reflection-aware guid- ance,” Applied Intelligence, pp. 1–16, 2023

2023
[21]

Language- guided image reflection separation,

H. Zhong, Y . Hong, S. Weng, J. Liang, and B. Shi, “Language- guided image reflection separation,” in CVPR, 2024, pp. 24913–24922

2024
[22]

V-desirr: Very fast deep embedded single image reflection re- moval,

B. H. Prasad, L. R Boregowda, K. Mitra, S. Chowdhury, et al., “V-desirr: Very fast deep embedded single image reflection re- moval,” in ICCV, 2021, pp. 2390–2399

2021
[23]

Trash or treasure? an interactive dual-stream strategy for single image reflection separation,

Q. Hu and X. Guo, “Trash or treasure? an interactive dual-stream strategy for single image reflection separation,” NeurIPS, vol. 34, 2021

2021
[24]

Hit-sr: Hierarchical transformer for efficient image super-resolution,

X. Zhang, Y . Zhang, and F. Yu, “Hit-sr: Hierarchical transformer for efficient image super-resolution,” in ECCV. Springer, 2024, pp. 483–500

2024
[25]

When fast fourier transform meets transformer for image restoration,

X. Jiang, X. Zhang, N. Gao, and Y . Deng, “When fast fourier transform meets transformer for image restoration,” in ECCV. Springer, 2024, pp. 381–402

2024
[26]

Attention is all you need,

A. Vaswani, “Attention is all you need,” NeurIPS, 2017

2017
[27]

Swinir: Image restoration using swin transformer,

J. Liang, J. Cao, G. Sun, K. Zhang, L. Van Gool, and R. Tim- ofte, “Swinir: Image restoration using swin transformer,” in ICCV, 2021, pp. 1833–1844

2021
[29]

Seeing deeply and bidirectionally: A deep learning approach for single image re- flection removal,

J. Yang, D. Gong, L. Liu, and Q. Shi, “Seeing deeply and bidirectionally: A deep learning approach for single image re- flection removal,” in ECCV, 2018, pp. 654–669

2018
[31]

Single im- age reflection removal exploiting misaligned training data and network enhancements,

K. Wei, J. Yang, Y . Fu, D. Wipf, and H. Huang, “Single im- age reflection removal exploiting misaligned training data and network enhancements,” in CVPR, 2019, pp. 8178–8187

2019
[32]

Single image reflection re- moval with physically-based training images,

S. Kim, Y . Huo, and S. E. Yoon, “Single image reflection re- moval with physically-based training images,” inCVPR, 2020, pp. 5164–5173

2020
[33]

Single image reflection removal through cascaded refinement,

C. Li, Y . Yang, K. He, S. Lin, and J. E. Hopcroft, “Single image reflection removal through cascaded refinement,” in CVPR, 2020, pp. 3565–3574

2020
[34]

Personalized single image reflection removal network through adaptive cascade re- finement,

M. Wang, X. Zhang, Y . Gong, and Y . Yin, “Personalized single image reflection removal network through adaptive cascade re- finement,” in Proceedings of the 31st ACM International Con- ference on Multimedia, 2023, pp. 8204–8213

2023
[35]

Single image reflection sepa- ration with perceptual losses,

X. Zhang, R. Ng, and Q. Chen, “Single image reflection sepa- ration with perceptual losses,” in CVPR, 2018, pp. 4786–4794

2018
[36]

The pascal visual object classes (voc) chal- lenge,

M. Everingham, L. Van Gool, C. KI Williams, J. Winn, and A. Zisserman, “The pascal visual object classes (voc) chal- lenge,” International Journal of Computer Vision , vol. 88, pp. 303–338, 2010

2010
[37]

Restormer: Efficient transformer for high- resolution image restoration,

S. W. Zamir, A. Arora, S. Khan, M. Hayat, F. S. Khan, and M. H. Yang, “Restormer: Efficient transformer for high- resolution image restoration,” in CVPR, 2022, pp. 5728–5739

2022

Pith tools

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

[1] [19]

Single image reflection separation via component synergy,

Q. Hu and X. Guo, “Single image reflection separation via component synergy,” in ICCV, 2023, pp. 13138–13147

work page 2023

[2] [20]

Revisiting single image reflection removal in the wild,

Y . Zhu, X. Fu, P. T. Jiang, H. Zhang, Q. Sun, J. Chen, Z. Zha, and B. Li, “Revisiting single image reflection removal in the wild,” in CVPR, 2024, pp. 25468–25478

work page 2024

[3] [30]

Single image reflection removal beyond linearity,

Q. Wen, Y . Tan, J. Qin, W. Liu, G. Han, and S. He, “Single image reflection removal beyond linearity,” in CVPR, 2019, pp. 3771–3779

work page 2019

[4] [28]

Swin transformer: Hierarchical vision transformer using shifted windows,

Z. Liu, Y . Lin, Y . Cao, H. Hu, Y . Wei, Z. Zhang, S. Lin, and B. Guo, “Swin transformer: Hierarchical vision transformer using shifted windows,” in ICCV, 2021, pp. 10012–10022

work page 2021

[5] [15]

Degradation-aware image enhancement via vision- language classification,

J. Cai, K. Yang, J. Ding, L. Fu, L. Ouyang, J. Li, J. Shen, and Z. Meng, “Degradation-aware image enhancement via vision- language classification,” in MIPR, 2025

work page 2025

[6] [1]

F2T2-HiT: A U-Shaped FFT Transformer and Hierarchical Transformer for Reflection Removal

INTRODUCTION The SIRR technique aims to decompose a reflection- contaminated input image I into two distinct components: the reflection layer R and the reflection-free transmission layer T . The primary objective is to ensure that the generated transmission layer T closely approximates its corresponding ground truth B, which represents the ideal reflectio...

work page Pith review arXiv 2025

[7] [2]

ghosting

RELA TED WORK Over recent decades, numerous innovative methods have been proposed to address the problem of reflection removal. Some methods rely on additional inputs, such as multi- frames [3], polarization [4], and flash-only priors [5]. Others utilize traditional image priors, including relative smooth- ness, “ghosting” cues, and dark channels, to sepa...

work page

[8] [3]

METHODOLOGY We first introduce the fundamental methodology of the F2T2-HiT architecture in Sec.3.1, followed by the block-level designs detailed in Sec.3.2, Sec.3.3, and Sec.3.4, respectively. 3.1. A U-Shaped F2T2-HiT Architecture To simplify inter-block complexity, we utilize a classic single-stage U-shaped architecture with skip connections, as shown in...

work page