Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

Completion as Enhancement: A Degradation-Aware Selective Image Guided Network for Depth Completion

T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read SigNet claims that depth completion is better solved as depth enhancement: densify first, then let a learned degradation kernel select RGB edges to correct the coarse depth, achieving state-of-the-art RMSE on four benchmarks with a…

desk verdict SigNet is a well-engineered, unusually compact depth completion model with consistent SOTA numbers, but its degradation-bridge story is a fitted assumption rather than a verified mechanism. read the letter →

arxiv 2412.19225 v2 pith:QY3UV7XS submitted 2024-12-26 cs.CV eess.IV

Zhiqiang Yan , Zhengxue Wang , Kun Wang , Jun Li , Jian Yang This is my paper · ORCID
classification cs.CVeess.IV
keywords depthcompletionenhancementdegradation-awarenetworkRGB-DfusionconditionalMambadiscretecosinetransformsparsedensificationstatespacemodel
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to establish that depth completion can be recast as depth enhancement: first fill the sparse depth map with classical densification to get a coarse dense map, then treat the remaining error as a learned degradation that RGB edge information can compensate. SigNet makes this concrete by estimating a per-image degradation kernel that is decomposed in the DCT domain and used both to select high-frequency RGB features and to condition a Mamba sequence model that fuses RGB and depth. The payoff claimed is state-of-the-art RMSE on NYUv2, DIML, SUN RGBD, and TOFDC with only 3.3M parameters, e.g., 83 mm on NYUv2 and 55 mm on DIML. The redefinition matters because it removes the irregular-sparse-data mismatch that direct convolutions face, and it borrows a mature degradation-learning toolkit from image restoration.

What carries the argument

The load-bearing object is the degradation bridge in Eq. (3), $Z = K(D;\vartheta) \otimes Y + n$, which asserts that coarse depth $Z$ is a convolution of target depth $Y$ with a kernel generated from an implicit degradation map $D$, plus noise. DADF then decomposes $D$ with a 2D DCT into three components, builds an adaptive spectrum mask that multiplies RGB features to select high-frequency content, and uses the decomposed degradation as a condition inside a Mamba state-space block, computing the SSM matrices $B$ and $C$ from gated recurrent convolutions over the RGB-D features and degradation. This machinery converts the completion task into an enhancement task and supplies the mechanism for RGB-guided correction.

What would settle it

Re-run the DIML or NYUv2 experiments with the degradation loss $L_d$ removed while keeping the DADF module; if RMSE barely changes and the learned kernel $K$ converges to a near-identity kernel, the degradation bridge is not what carries the performance. Alternatively, construct a synthetic test where coarse depth is corrupted by non-edge-centered noise, such as large smooth-region offsets; if SigNet's edge-selection mechanism cannot correct it, the edge-concentration premise in the degradation model is falsified.

Watch

Extended reading notes

Core claim

SigNet's central claim is that the long-standing sparse-depth difficulty in depth completion is avoidable: dense coarse depth produced by non-CNN methods, together with a self-supervised degradation assumption $Z = K(D;\vartheta) \otimes Y + n$, turns completion into a depth super-resolution/enhancement problem. The degradation kernel $K$ is learned from an implicit degradation representation $D$ and is constrained by the loss $L_d = \|K \otimes Y - Z\|_1$ so that the network explicitly models how coarse depth arises from target depth. Degradation typically sits at edges, so DADF decomposes $D$ via a 2D DCT into components, uses them to mask and select high-frequency RGB features, and feeds RGB-D features plus degradation into a conditional Mamba whose state parameters are generated from both the input and the degradation. The authors report that this design produces state-of-the-art numbers on four benchmarks and that each component, densification plus degradation, denoising, attention selection, DCT decomposition, and conditional Mamba, contributes in ablations.

Load-bearing premise

The approach stands on Eq. (3): that the coarse depth map is the target depth blurred by a learned, edge-concentrated kernel plus noise, so that RGB high-frequency information is the right signal to subtract the error; if real coarse-depth error is not mostly edge-localized blur, the adaptive selection and conditional Mamba lose their principled basis.

Editorial extensions

If this is right

  • If SigNet's reported numbers hold, depth completion can be made both more accurate and far lighter: 3.3M parameters versus 89.9M for BPNet on NYUv2, with lower RMSE.
  • The densify-then-enhance recipe gives a general route around sparse-input convolution, so the same degradation bridge could apply to other dense prediction problems where input and output share resolution.
  • The self-supervised degradation loss $L_d$ links coarse and target depth without paired degradation examples, reducing the need for externally supplied degradation supervision.
  • Conditional Mamba improves over vanilla Vision Mamba by roughly 8 mm RMSE in ablations, showing that degradation-conditioned state parameters help global RGB-D fusion.
  • The method is stable across input densities: on TOFDC it beats TPVD at every tested density from 1% to 95%, with an average improvement of 9.7%.

Reading between the lines

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

  • Editorial inference: because the authors note depth completion and depth super-resolution share the same end goal, the degradation-aware DADF module could be applied directly to RGB-guided depth super-resolution with minimal changes.
  • Editorial inference: the reported failure on KITTI suggests the edge-localized degradation assumption is range- and sensor-dependent; a testable fix hinted by the paper is to supervise the degradation branch with dense edge maps from monocular depth priors.
  • Editorial inference: since the densification step is fixed and non-CNN, the method's accuracy ceiling may be tied to the quality of that initial fill; replacing it with a learned densifier could improve cases with large holes or very sparse input.
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. SigNet reframes depth completion as depth enhancement: sparse depth is first densified by non-CNN tools (IP-Basic/colorization) to obtain a coarse dense map, which is then enhanced by a CNN under an assumed degradation model Z = K(D;θ) ⊗ Y + n. A Degradation-Aware Decomposition and Fusion (DADF) module estimates an implicit degradation representation, decomposes it via DCT into components, uses those components to select high-frequency RGB features, and fuses the resulting RGB-D features with a conditional Mamba. The paper reports state-of-the-art RMSE on NYUv2, DIML, SUN RGBD, and TOFDC with a 3.3M-parameter model, along with ablations, density experiments, complexity comparisons, and a KITTI failure-case analysis.

Significance. If the reported numbers are reproducible, the paper demonstrates a strong empirical result: replacing direct sparse-depth convolution with densification-plus-enhancement yields competitive or better depth completion with substantially fewer parameters. The DCT-based decomposition of an implicit degradation and the conditional Mamba fusion are interesting mechanisms, and the evaluation across four datasets is broad. The paper is also honest about a KITTI failure case, which is a positive sign. However, the central conceptual claim—that the method is degradation-aware in a meaningful sense—rests on an untested assumption about how coarse depth relates to target depth, and the reported gains could plausibly be attributed to the extra attention/global-modeling components rather than to the degradation mechanism.

major comments (3)
  1. [Sec. 3.1, Eq. (3)] The degradation model Z = K(D;θ) ⊗ Y + n is the conceptual foundation of the paper, but it is asserted rather than validated. The coarse depth Z is obtained by IP-Basic morphological hole-filling or colorization-based interpolation, which are nonlinear operations that need not be representable as a convolution of the target depth with any kernel. The auxiliary loss Ld in Eq. (13) only enforces that the learned K applied to the ground-truth Y reproduces Z, so a degenerate K can satisfy this objective without the representation D capturing the actual degradation mechanism. As a result, the claim that DADF is 'degradation-aware' is not currently supported by any direct diagnostic. I request at least one of: (i) quantitative evaluation of how well K(D;θ)⊗Y predicts the actual coarse-depth errors, (ii) an ablation in which D is replaced by a fixed or random condition, or (iii) visual/statistical evidence that the DCT-decomposed degradation components correlate with depth boundaries and errors. Without such evidence, the reported improvements could come from the attention or Mamba components rather than from modeling degradation.
  2. [Abstract and Sec. 3.3, Eq. (13)] The degradation loss Ld uses the ground-truth dense depth Y to supervise the kernel K, so the degradation bridge is not 'self-supervised' as stated in the abstract and introduction; it is supervised by the same ground truth that trains the reconstruction loss Lr. If the intended claim is that no annotations beyond the existing GT depth are required, that should be stated precisely. The distinction matters because the paper presents the degradation bridge as a conceptual novelty, and the current terminology is misleading.
  3. [Sec. 4, Tables 1-4] The state-of-the-art claim rests on single-run metrics with no error bars or repeated runs, and most baseline numbers are borrowed from AGG-Net and TPVD rather than re-run under the authors' protocol. Several improvements are small relative to typical run-to-run variation (e.g., NYUv2 RMSE 83 vs 88 mm and REL 0.012 vs 0.011; TOFDC RMSE 87 vs 92 mm). Please report mean ± std over at least three runs for the main comparisons and, if possible, re-run the strongest baselines under the same training settings. In addition, Table 3 mislabels the final row as 'v' when the text describes 'SigNet-vi'; as printed, the table does not unambiguously support the stated 13 mm gain from conditional Mamba.
minor comments (5)
  1. [Table 2 and Sec. 4.1.3] Table 2 caption says the density comparison is on TOFDC, but Sec. 4.1.3 states the comparison is on DIML; please correct the dataset label.
  2. [Sec. 6.3, Eq. (14)] The implementation details call the loss weight γ, whereas Eq. (14) defines λ; unify the notation.
  3. [Fig. 8 and Sec. 4.2] The Figure 8 caption labels (a) as 'Iterations of DADF' and (b) as 'Mamba Fusion', while the text refers to Fig. 8(a) as Mamba fusion and Fig. 8(b) as iterations; the caption and text are inconsistent.
  4. [Sec. 3.1, Eqs. (1) and (3)] Equation (1) uses h for the degradation while Eq. (3) uses a kernel K(D;θ); the relationship between h and K, and the origin of the implicit degradation D (which network, with what input), should be defined more precisely.
  5. [Reproducibility] The paper does not state whether code or trained models will be released; please provide code or a clear reproducibility statement.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the degradation bridge is an auxiliary learned model assumption, and the depth predictions are independently supervised and externally benchmarked.

full rationale

The paper's derivation chain is generative rather than circular: sparse depth S is densified by non-CNN tools to form coarse depth Z (Eq. 2), a degradation relation Z = K(D;θ)⊗Y + n is assumed (Eq. 3), and a degradation network produces D that conditions feature selection and fusion in DADF. The final depth prediction is trained by the L1 reconstruction loss (Eq. 12) directly against ground truth and is evaluated on held-out test splits in Table 1, so the reported metric improvements are not forced by construction. The auxiliary degradation loss (Eq. 13) does fit the kernel K to reproduce the coarse input Z from ground truth Y, but this fitted kernel is used only as an intermediate regularizer/conditioning signal; it is not the source of the predicted depth, and the paper does not rename a fit as a benchmark prediction. The load-bearing Eq. (3) is explicitly introduced as an assumption ('we assume that there exists unknown degradation h'), not derived from the output, and the paper's own Sec. 4.3 concedes that the degradation model fails to transfer to KITTI, which is a generalization/validity weakness rather than a circular step. Self-citations (RigNet, TPVD, SGNet, etc.) are used for context, for baseline comparisons, or for future-work suggestions; the degradation assumption itself is grounded in external works [43, 59, 62], and no uniqueness theorem from the authors' prior work is invoked to rule out alternatives. Hence the central claim is self-contained with respect to circularity.

Assumptions & free parameters 3 free parameters · 3 assumptions · 1 invented entities

The central claim rests on a learned degradation model that is assumed rather than derived, plus several hand-set hyperparameters. The implicit degradation D is an invented latent entity with no independent falsifiable evidence. The performance gain is empirical, so the method should be judged on reproducibility and benchmark fairness rather than on the conceptual framing.

free parameters (3)
  • lambda (degredation loss weight) = 0.1
    Balance coefficient for the degradation loss in Eq. (14); set empirically, with no sensitivity analysis shown.
  • number of DCT decomposition paths = 3
    Ablation in Table 4 shows RMSE keeps decreasing with more paths (11 paths gives 54.0 vs 55.0 for 3), but 3 is chosen for computational cost.
  • number of DADF iterations = 5
    Ablation in Fig. 8(b) shows performance improves with more iterations; 5 is chosen as an efficiency/effectiveness trade-off.
assumptions (3)
  • ad hoc to paper There is an unknown degradation h relating coarse depth Z and target depth Y, Z = hY + n.
    Introduced in Eq. (1) and Eq. (3); no independent justification, and the kernel K is learned rather than derived.
  • domain assumption Degradation typically occurs near edges, so high-frequency RGB components can compensate for coarse depth.
    Stated in Sec. 3.2 as a 'fundamental observation'; it is an empirical heuristic, not proven.
  • domain assumption Non-CNN densification produces a coarse dense depth that eliminates mismatch and ambiguity of sparse convolution.
    Stated in Sec. 1 and Sec. 3.1; plausible but not formally established, and the densification tools introduce their own errors.
invented entities (1)
  • Implicit degradation representation D and derived kernel K(D;theta)
    purpose: Models the blur-like relationship between coarse depth and target depth; used to select high-frequency RGB features and condition the Mamba fusion.
    D is generated by a learned degradation network and is only constrained by the auxiliary loss Ld in Eq. (13), which uses ground truth depth during training. There is no external or cross-modal evidence that D captures a physically meaningful degradation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Completion as Enhancement: A Degradation-Aware Selective Image Guided Network for Depth Completion." pith.science (2026). https://pith.science/paper/QY3UV7XS

@misc{pith2026241219225,
  author       = {Pith},
  title        = {Pith review of: Completion as Enhancement: A Degradation-Aware Selective Image Guided Network for Depth Completion},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QY3UV7XS}},
  note         = {Machine review of arXiv:2412.19225}
}
read the original abstract

In this paper, we introduce the Selective Image Guided Network (SigNet), a novel degradation-aware framework that transforms depth completion into depth enhancement for the first time. Moving beyond direct completion using convolutional neural networks (CNNs), SigNet initially densifies sparse depth data through non-CNN densification tools to obtain coarse yet dense depth. This approach eliminates the mismatch and ambiguity caused by direct convolution over irregularly sampled sparse data. Subsequently, SigNet redefines completion as enhancement, establishing a self-supervised degradation bridge between the coarse depth and the targeted dense depth for effective RGB-D fusion. To achieve this, SigNet leverages the implicit degradation to adaptively select high-frequency components (e.g., edges) of RGB data to compensate for the coarse depth. This degradation is further integrated into a multi-modal conditional Mamba, dynamically generating the state parameters to enable efficient global high-frequency information interaction. We conduct extensive experiments on the NYUv2, DIML, SUN RGBD, and TOFDC datasets, demonstrating the state-of-the-art (SOTA) performance of SigNet.

Figures

Figures reproduced from arXiv: 2412.19225 by the authors.

Figure 1
Figure 1. Illustration of our main concept. It redefines depth com [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Pipeline of SigNet. The sparse depth data is initially filled to create a coarse depth map. We then utilize the degradation [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Overview of our DADF. ⊙ and CM refer to element￾wise multiplication and the conditional Mamba, respectively. match and ambiguity caused by direct convolution over ir￾regularly sampled sparse data. Subsequently, we employ a degradation network to estimate the implicit degradation D. The degradation kernel K is produced from D using convolutions. Consequently, given the coarse depth Z and target depth Y, Eq. (1) can b… view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Visual comparison with SOTA methods on TOFDC dataset, including NLSPN [ [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Visual comparison with SOTA approaches on DIML dataset, including CSPN [ [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Visual examples of filled coarse depth maps and degradation representations on the DIML dataset. [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]
Figure 7
Figure 7. Figure 7: Complexity comparison on the TOFDC dataset, where [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: Ablation study of our DADF on the DIML dataset. [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 9
Figure 9. Figure 9: Visual results of failure cases on KITTI benchmark [ [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]
Figure 10
Figure 10. Figure 10: Visual comparisons with SOTAs on DIML. of 1 × 10−4 , and a weight decay of 1 × 10−6 . The balance coefficient γ in the loss function is set to 0.1. To further improve model performance and generalization, we apply data augmentation techniques, including random croppin…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. Event-Driven Dynamic Scene Depth Completion

    cs.CV 2025-05 conditional novelty 6.0 of 10

    EventDC uses event-stream-conditioned dynamic convolutions (EMA and LDF modules) to align and refine RGB and sparse-depth features, achieving state-of-the-art results on a new event-based depth completion benchmark.

Reference graph

Works this paper leans on

70 extracted references · 67 canonical work pages · cited by 1 Pith paper

  1. [1]

    A non-local algorithm for image denoising

    Antoni Buades, Bartomeu Coll, and J-M Morel. A non-local algorithm for image denoising. InCVPR, pages 60–65, 2005. 5, 6

    work page 2005

  2. [2]

    Agg-net: Attention guided gated- convolutional network for depth image completion

    Dongyue Chen, Tingxuan Huang, Zhimin Song, Shizhuo Deng, and Tong Jia. Agg-net: Attention guided gated- convolutional network for depth image completion. InICCV, pages 8853–8862, 2023. 5, 6, 7, 2

    work page 2023

  3. [3]

    Learning joint 2d-3d representations for depth completion

    Yun Chen, Bin Yang, Ming Liang, and Raquel Urtasun. Learning joint 2d-3d representations for depth completion. In ICCV, pages 10023–10032, 2019. 2

    work page 2019

  4. [4]

    Learning depth with convolutional spatial propagation network

    Xinjing Cheng, Peng Wang, and Ruigang Yang. Learning depth with convolutional spatial propagation network. In ECCV, pages 103–119, 2018. 5, 6, 7

    work page 2018

  5. [5]

    Depth esti- mation via affinity learned with convolutional spatial propa- gation network

    Xinjing Cheng, Peng Wang, and Ruigang Yang. Depth esti- mation via affinity learned with convolutional spatial propa- gation network. In ECCV, pages 103–119, 2018. 2

    work page 2018

  6. [6]

    Learning depth with convolutional spatial propagation network

    Xinjing Cheng, Peng Wang, and Ruigang Yang. Learning depth with convolutional spatial propagation network. IEEE Transactions on Pattern Analysis and Machine Intelligence, 42(10):2361–2379, 2019. 2

    work page 2019

  7. [7]

    Cspn++: Learning context and resource aware convo- lutional spatial propagation networks for depth completion

    Xinjing Cheng, Peng Wang, Chenye Guan, and Ruigang Yang. Cspn++: Learning context and resource aware convo- lutional spatial propagation networks for depth completion. In AAAI, pages 10615–10622, 2020. 2

    work page 2020

  8. [8]

    A Large RGB-D Dataset for Semi-supervised Monocular Depth Estimation

    Jaehoon Cho, Dongbo Min, Youngjung Kim, and Kwanghoon Sohn. A large rgb-d dataset for semi- supervised monocular depth estimation. arXiv preprint arXiv:1904.10230, 2019. 2

    work page Pith review arXiv 1904

Show all 70 references
  1. [9]

    Confidence propagation through cnns for guided sparse depth regression

    Abdelrahman Eldesokey, Michael Felsberg, and Fahad Shah- baz Khan. Confidence propagation through cnns for guided sparse depth regression. IEEE Transactions on Pattern Anal- ysis and Machine Intelligence, 42(10):2423–2436, 2020. 2

    2020

  2. [10]

    Mamba: Linear-time sequence modeling with selective state spaces

    Albert Gu and Tri Dao. Mamba: Linear-time sequence modeling with selective state spaces. arXiv preprint arXiv:2312.00752, 2023. 4

    2023 arXiv

  3. [11]

    Penet: Towards precise and efficient image guided depth completion

    Mu Hu, Shuling Wang, Bin Li, Shiyu Ning, Li Fan, and Xiaojin Gong. Penet: Towards precise and efficient image guided depth completion. In ICRA, 2021. 2, 5

    2021

  4. [12]

    Sparse and dense data with cnns: Depth completion and semantic segmentation

    Maximilian Jaritz, Raoul De Charette, Emilie Wirbel, Xavier Perrotton, and Fawzi Nashashibi. Sparse and dense data with cnns: Depth completion and semantic segmentation. In3DV, pages 52–60, 2018. 2

    2018

  5. [13]

    Adam: A method for stochastic optimization

    Diederik P Kingma and Jimmy Ba. Adam: A method for stochastic optimization. In Computer Ence, 2014. 2

    2014

  6. [14]

    In defense of classical image processing: Fast depth completion on the cpu

    Jason Ku, Ali Harakeh, and Steven L Waslander. In defense of classical image processing: Fast depth completion on the cpu. In CRV, pages 16–22, 2018. 1, 2, 3

    2018

  7. [15]

    Colorization using optimization

    Anat Levin, Dani Lischinski, and Yair Weiss. Colorization using optimization. In SIGGRAPH, pages 689–694, 2004. 1, 2, 3

    2004

  8. [16]

    A multi-scale guided cascade hourglass net- work for depth completion

    Ang Li, Zejian Yuan, Yonggen Ling, Wanchao Chi, Chong Zhang, et al. A multi-scale guided cascade hourglass net- work for depth completion. In WACV, pages 32–40, 2020. 2

    2020

  9. [17]

    All-in-one image restoration for unknown cor- ruption

    Boyun Li, Xiao Liu, Peng Hu, Zhongqin Wu, Jiancheng Lv, and Xi Peng. All-in-one image restoration for unknown cor- ruption. In CVPR, pages 17452–17462, 2022. 3

    2022

  10. [18]

    Efficient and degradation-adaptive network for real-world image super- resolution

    Jie Liang, Hui Zeng, and Lei Zhang. Efficient and degradation-adaptive network for real-world image super- resolution. In ECCV, pages 574–591, 2022. 3

    2022

  11. [19]

    Dynamic spatial propagation network for depth com- pletion

    Yuankai Lin, Tao Cheng, Qi Zhong, Wending Zhou, and Hua Yang. Dynamic spatial propagation network for depth com- pletion. In AAAI, pages 1638–1646, 2022. 2

    2022

  12. [20]

    Dyspn: Learning dynamic affinity for image- guided depth completion

    Yuankai Lin, Hua Yang, Tao Cheng, Wending Zhou, and Zhouping Yin. Dyspn: Learning dynamic affinity for image- guided depth completion. IEEE Transactions on Circuits and Systems for Video Technology, pages 1–1, 2023. 2

    2023

  13. [21]

    Learning steering kernels for guided depth comple- tion

    Lina Liu, Yiyi Liao, Yue Wang, Andreas Geiger, and Yong Liu. Learning steering kernels for guided depth comple- tion. IEEE Transactions on Image Processing , 30:2850– 2861, 2021. 1

    2021

  14. [22]

    Fcfr-net: Fea- ture fusion based coarse-to-fine residual learning for depth completion

    Lina Liu, Xibin Song, Xiaoyang Lyu, Junwei Diao, Meng- meng Wang, Yong Liu, and Liangjun Zhang. Fcfr-net: Fea- ture fusion based coarse-to-fine residual learning for depth completion. In AAAI, pages 2136–2144, 2021. 1, 2

    2021

  15. [23]

    Learning affinity via spa- tial propagation networks

    Sifei Liu, Shalini De Mello, Jinwei Gu, Guangyu Zhong, Ming-Hsuan Yang, and Jan Kautz. Learning affinity via spa- tial propagation networks. In NeurIPS, 2017. 2

    2017

  16. [24]

    Graphcspn: Geometry-aware depth completion via dynamic gcns

    Xin Liu, Xiaofei Shao, Bo Wang, Yali Li, and Shengjin Wang. Graphcspn: Geometry-aware depth completion via dynamic gcns. In ECCV, pages 90–107. Springer, 2022. 2, 5

    2022

  17. [25]

    Vmamba: Visual state space model

    Yue Liu, Yunjie Tian, Yuzhong Zhao, Hongtian Yu, Lingxi Xie, Yaowei Wang, Qixiang Ye, and Yunfan Liu. Vmamba: Visual state space model. arXiv preprint arXiv:2401.10166,

    arXiv

  18. [26]

    From depth what can you see? depth completion via aux- iliary image reconstruction

    Kaiyue Lu, Nick Barnes, Saeed Anwar, and Liang Zheng. From depth what can you see? depth completion via aux- iliary image reconstruction. In CVPR, pages 11306–11315,

  19. [27]

    Sparse-to-dense: Depth prediction from sparse depth samples and a single image

    Fangchang Ma and Sertac Karaman. Sparse-to-dense: Depth prediction from sparse depth samples and a single image. In ICRA, pages 4796–4803. IEEE, 2018. 1

    2018

  20. [28]

    Self-supervised sparse-to-dense: Self-supervised depth completion from lidar and monocular camera

    Fangchang Ma, Guilherme Venturelli Cavalheiro, and Sertac Karaman. Self-supervised sparse-to-dense: Self-supervised depth completion from lidar and monocular camera. In ICRA, 2019. 2, 5, 6

    2019

  21. [29]

    Non-local spatial propagation network for depth completion

    Jinsun Park, Kyungdon Joo, Zhe Hu, Chi-Kuei Liu, and In So Kweon. Non-local spatial propagation network for depth completion. In ECCV, 2020. 2, 5, 6

    2020

  22. [30]

    Depth prompting for sensor-agnostic depth estimation

    Jin-Hwi Park, Chanhwi Jeong, Junoh Lee, and Hae-Gon Jeon. Depth prompting for sensor-agnostic depth estimation. In CVPR, pages 9859–9869, 2024. 2

    2024

  23. [31]

    Deepli- dar: Deep surface normal guided depth prediction for out- door scene from sparse lidar data and single color image

    Jiaxiong Qiu, Zhaopeng Cui, Yinda Zhang, Xingdi Zhang, Shuaicheng Liu, Bing Zeng, and Marc Pollefeys. Deepli- dar: Deep surface normal guided depth prediction for out- door scene from sparse lidar data and single color image. In CVPR, pages 3313–3322, 2019. 5, 2

    2019

  24. [32]

    Guide- former: Transformers for image guided depth completion

    Kyeongha Rho, Jinsung Ha, and Youngjung Kim. Guide- former: Transformers for image guided depth completion. In CVPR, pages 6250–6259, 2022. 2

    2022

  25. [33]

    Decoder modulation for in- door depth completion

    Dmitry Senushkin, Mikhail Romanov, Ilia Belikov, Nikolay Patakin, and Anton Konushin. Decoder modulation for in- door depth completion. In IROS, pages 2181–2188. IEEE,

  26. [34]

    Dfusenet: Deep fusion of rgb and sparse depth information for image guided dense depth completion

    Shreyas S Shivakumar, Ty Nguyen, Ian D Miller, Steven W Chen, Vijay Kumar, and Camillo J Taylor. Dfusenet: Deep fusion of rgb and sparse depth information for image guided dense depth completion. In ITSC, pages 13–20. IEEE, 2019. 5, 6

    2019

  27. [35]

    Indoor segmentation and support inference from rgbd images

    Nathan Silberman, Derek Hoiem, Pushmeet Kohli, and Rob Fergus. Indoor segmentation and support inference from rgbd images. In ECCV, pages 746–760. Springer, 2012. 2

    2012

  28. [36]

    Sun rgb-d: A rgb-d scene understanding benchmark suite

    Shuran Song, Samuel P Lichtenberg, and Jianxiong Xiao. Sun rgb-d: A rgb-d scene understanding benchmark suite. In CVPR, pages 567–576, 2015. 1, 2

    2015

  29. [37]

    Learning guided convolutional network for depth comple- tion

    Jie Tang, Fei-Peng Tian, Wei Feng, Jian Li, and Ping Tan. Learning guided convolutional network for depth comple- tion. IEEE Transactions on Image Processing , 30:1116– 1129, 2020. 2, 5

    2020

  30. [38]

    Bi- lateral propagation network for depth completion

    Jie Tang, Fei-Peng Tian, Boshi An, Jian Li, and Ping Tan. Bi- lateral propagation network for depth completion. In CVPR, pages 9763–9772, 2024. 1, 2, 5, 6, 7

    2024

  31. [39]

    A flexible scene rep- resentation for 3d reconstruction using an rgb-d camera

    Diego Thomas and Akihiro Sugimoto. A flexible scene rep- resentation for 3d reconstruction using an rgb-d camera. In ICCV, pages 2800–2807, 2013. 1

    2013

  32. [40]

    Sparsity invariant cnns

    Jonas Uhrig, Nick Schneider, Lukas Schneider, Uwe Franke, Thomas Brox, and Andreas Geiger. Sparsity invariant cnns. In 3DV, pages 11–20, 2017. 1, 2, 8

    2017

  33. [41]

    Sparse and noisy lidar completion with rgb guidance and uncertainty

    Wouter Van Gansbeke, Davy Neven, Bert De Brabandere, and Luc Van Gool. Sparse and noisy lidar completion with rgb guidance and uncertainty. In MVA, pages 1–6, 2019. 2

    2019

  34. [42]

    Rgb-depth fusion gan for indoor depth completion

    Haowen Wang, Mingyuan Wang, Zhengping Che, Zhiyuan Xu, Xiuquan Qiao, Mengshi Qi, Feifei Feng, and Jian Tang. Rgb-depth fusion gan for indoor depth completion. InCVPR, pages 6209–6218, 2022. 5

    2022

  35. [43]

    Unsuper- vised degradation representation learning for blind super- resolution

    Longguang Wang, Yingqian Wang, Xiaoyu Dong, Qingyu Xu, Jungang Yang, Wei An, and Yulan Guo. Unsuper- vised degradation representation learning for blind super- resolution. In CVPR, pages 10581–10590, 2021. 1, 2, 3

    2021

  36. [44]

    Lrru: Long-short range recurrent updating networks for depth completion

    Yufei Wang, Bo Li, Ge Zhang, Qi Liu, Tao Gao, and Yuchao Dai. Lrru: Long-short range recurrent updating networks for depth completion. In ICCV, pages 9422–9432, 2023. 2

    2023

  37. [45]

    Improving depth completion via depth feature upsampling

    Yufei Wang, Ge Zhang, Shaoqian Wang, Bo Li, Qi Liu, Le Hui, and Yuchao Dai. Improving depth completion via depth feature upsampling. In CVPR, pages 21104–21113, 2024. 1, 2

    2024

  38. [46]

    Sgnet: Struc- ture guided network via gradient-frequency awareness for depth map super-resolution

    Zhengxue Wang, Zhiqiang Yan, and Jian Yang. Sgnet: Struc- ture guided network via gradient-frequency awareness for depth map super-resolution. In AAAI, pages 5823–5831,

  39. [47]

    Multi-modal masked pre-training for monoc- ular panoramic depth completion

    Zhiqiang Yan, Xiang Li, Kun Wang, Zhenyu Zhang, Jun Li, and Jian Yang. Multi-modal masked pre-training for monoc- ular panoramic depth completion. In ECCV, pages 378–395,

  40. [48]

    Rignet: Repetitive image guided network for depth completion

    Zhiqiang Yan, Kun Wang, Xiang Li, Zhenyu Zhang, Jun Li, and Jian Yang. Rignet: Repetitive image guided network for depth completion. In ECCV, pages 214–230, 2022. 1, 2, 5, 6

    2022

  41. [49]

    Desnet: Decomposed scale-consistent net- work for unsupervised depth completion

    Zhiqiang Yan, Kun Wang, Xiang Li, Zhenyu Zhang, Jun Li, and Jian Yang. Desnet: Decomposed scale-consistent net- work for unsupervised depth completion. In AAAI, pages 3109–3117, 2023. 1

    2023

  42. [50]

    Learnable differ- encing center for nighttime depth perception

    Zhiqiang Yan, Yupeng Zheng, Kun Wang, Xiang Li, Zhenyu Zhang, Shuo Chen, Jun Li, and Jian Yang. Learnable differ- encing center for nighttime depth perception. arXiv preprint arXiv:2306.14538, 2023. 1

    2023 arXiv

  43. [51]

    Tri- perspective view decomposition for geometry-aware depth completion

    Zhiqiang Yan, Yuankai Lin, Kun Wang, Yupeng Zheng, Yufei Wang, Zhenyu Zhang, Jun Li, and Jian Yang. Tri- perspective view decomposition for geometry-aware depth completion. In CVPR, pages 4874–4884, 2024. 1, 2, 5, 7

    2024

  44. [52]

    Depth any- thing v2

    Lihe Yang, Bingyi Kang, Zilong Huang, Zhen Zhao, Xiao- gang Xu, Jiashi Feng, and Hengshuang Zhao. Depth any- thing v2. arXiv preprint arXiv:2406.09414, 2024. 8

    2024 arXiv

  45. [53]

    Dense depth posterior (ddp) from single image and sparse range

    Yanchao Yang, Alex Wong, and Stefano Soatto. Dense depth posterior (ddp) from single image and sparse range. In CVPR, pages 3353–3362, 2020. 2

    2020

  46. [54]

    Neural degradation representation learning for all-in-one image restoration

    Mingde Yao, Ruikang Xu, Yuanshen Guan, Jie Huang, and Zhiwei Xiong. Neural degradation representation learning for all-in-one image restoration. IEEE Transactions on Im- age Processing, 2024. 3

    2024

  47. [55]

    Con- ditional hyper-network for blind super-resolution with mul- tiple degradations

    Guanghao Yin, Wei Wang, Zehuan Yuan, Wei Ji, Dongdong Yu, Shouqian Sun, Tat-Seng Chua, and Changhu Wang. Con- ditional hyper-network for blind super-resolution with mul- tiple degradations. IEEE Transactions on Image Processing, 31:3949–3960, 2022. 2, 3

    2022

  48. [56]

    Aggregating feature point cloud for depth completion

    Zhu Yu, Zehua Sheng, Zili Zhou, Lun Luo, Si-Yuan Cao, Hong Gu, Huaqi Zhang, and Hui-Liang Shen. Aggregating feature point cloud for depth completion. In ICCV, pages 8732–8743, 2023. 2, 5

    2023

  49. [57]

    All-in-one multi-degradation image restoration network via hierarchical degradation representation

    Cheng Zhang, Yu Zhu, Qingsen Yan, Jinqiu Sun, and Yan- ning Zhang. All-in-one multi-degradation image restoration network via hierarchical degradation representation. InACM MM, pages 2285–2293, 2023. 3

    2023

  50. [58]

    Ingredient-oriented multi- degradation learning for image restoration

    Jinghao Zhang, Jie Huang, Mingde Yao, Zizheng Yang, Hu Yu, Man Zhou, and Feng Zhao. Ingredient-oriented multi- degradation learning for image restoration. In CVPR, pages 5825–5835, 2023. 3

    2023

  51. [59]

    Designing a practical degradation model for deep blind image super-resolution

    Kai Zhang, Jingyun Liang, Luc Van Gool, and Radu Timo- fte. Designing a practical degradation model for deep blind image super-resolution. In ICCV, pages 4791–4800, 2021. 1, 2, 3

    2021

  52. [60]

    Image super-resolution using very deep residual channel attention networks

    Yulun Zhang, Kunpeng Li, Kai Li, Lichen Wang, Bineng Zhong, and Yun Fu. Image super-resolution using very deep residual channel attention networks. In ECCV, pages 286– 301, 2018. 3

    2018

  53. [61]

    Completionformer: Depth completion with convolutions and vision transform- ers

    Youmin Zhang, Xianda Guo, Matteo Poggi, Zheng Zhu, Guan Huang, and Stefano Mattoccia. Completionformer: Depth completion with convolutions and vision transform- ers. In CVPR, pages 18527–18536, 2023. 2, 5

    2023

  54. [62]

    Guided depth map super-resolution: A survey

    Zhiwei Zhong, Xianming Liu, Junjun Jiang, Debin Zhao, and Xiangyang Ji. Guided depth map super-resolution: A survey. ACM Computing Surveys, 55(14s):1–36, 2023. 1, 2

    2023

  55. [63]

    Unsu- pervised cumulative domain adaptation for foggy scene op- tical flow

    Hanyu Zhou, Yi Chang, Wending Yan, and Luxin Yan. Unsu- pervised cumulative domain adaptation for foggy scene op- tical flow. In CVPR, pages 9569–9578, 2023. 1

    2023

  56. [64]

    Learn- ing correction filter via degradation-adaptive regression for blind single image super-resolution

    Hongyang Zhou, Xiaobin Zhu, Jianqing Zhu, Zheng Han, Shi-Xue Zhang, Jingyan Qin, and Xu-Cheng Yin. Learn- ing correction filter via degradation-adaptive regression for blind single image super-resolution. In ICCV, pages 12365– 12375, 2023. 3

    2023

  57. [65]

    Adverse weather op- tical flow: Cumulative homogeneous-heterogeneous adapta- tion

    Hanyu Zhou, Yi Chang, Zhiwei Shi, Wending Yan, Gang Chen, Yonghong Tian, and Luxin Yan. Adverse weather op- tical flow: Cumulative homogeneous-heterogeneous adapta- tion. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024. 1

    2024

  58. [66]

    Bev@ dc: Bird’s-eye view assisted training for depth completion

    Wending Zhou, Xu Yan, Yinghong Liao, Yuankai Lin, Jin Huang, Gangming Zhao, Shuguang Cui, and Zhen Li. Bev@ dc: Bird’s-eye view assisted training for depth completion. In CVPR, pages 9233–9242, 2023. 2

    2023

  59. [67]

    Curricular object manipulation in lidar- based object detection

    Ziyue Zhu, Qiang Meng, Xiao Wang, Ke Wang, Liujiang Yan, and Jian Yang. Curricular object manipulation in lidar- based object detection. In CVPR, pages 1125–1135, 2023. 1

    2023

  60. [68]

    V oxelsplat: Dynamic gaussian splatting as an effective loss for occupancy and flow predic- tion

    Ziyue Zhu, Shenlong Wang, Jin Xie, Jiang-Jiang Liu, Jing- dong Wang, and Jian Yang. V oxelsplat: Dynamic gaussian splatting as an effective loss for occupancy and flow predic- tion. In CVPR, 2025. 1 Completion as Enhancement: A Degradation-Aware Selective Image Guided Network ...

    2025

  61. [69]

    Dataset NYUv2 [35] is the most commonly used dataset for depth completion, consisting of 1,449 sets from 464 different in- door scenes using Microsoft Kinect

    Setup 6.1. Dataset NYUv2 [35] is the most commonly used dataset for depth completion, consisting of 1,449 sets from 464 different in- door scenes using Microsoft Kinect. AGG-Net divides this dataset into 420 images for training and 1,029 for testing. The initial resolution of ...

  62. [70]

    10 shows additional visual results on the DIML dataset

    More Visualizations Fig. 10 shows additional visual results on the DIML dataset. As evident, our method reconstructs depth structures with better accuracy and more detailed information. For exam- ple, in the first row, the cabinet predicted by our SigNet appears clearer and sh...

Pith tools

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