Pith. sign in

REVIEW 4 major objections 5 minor 40 references

DidSee: Diffusion-Based Depth Completion for Material-Agnostic Robotic Perception and Manipulation

T0 review · 4 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read DidSee turns noisy RGB-D into accurate depth for transparent and specular objects by making the diffusion final step noise-only, predicting depth in one pass, and adding a semantic segmenter, beating prior methods on three benchmarks.

desk verdict A practical, well-validated diffusion-based depth completion system for non-Lambertian objects; the SoTA claims are likely true, but missing error bars and a clean ablation make verification necessary. read the letter →

arxiv 2506.21034 v2 pith:HLHGFKIN submitted 2025-06-26 cs.CV

Wenzhou Lyu , Jialing Lin , Wenqi Ren , Ruihao Xia , Feng Qian , Yang Tang This is my paper · ORCID
classification cs.CV
keywords depthcompletiontransparentobjectsnon-LambertiansurfacesdiffusionmodelszeroterminalSNRsingle-stepsamplingsemanticsegmentationroboticgrasping
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

DidSee sets out to show that a pre-trained text-to-image diffusion model can be turned into an accurate depth-completion engine for transparent and specular objects, the surfaces where RGB-D cameras fail most. The paper argues that the vanilla diffusion setup silently injects two biases — a faint leftover of the clean depth signal at the final noise step, and errors that grow during multi-step sampling — and that both badly hurt depth accuracy even though they matter little for image generation. To remove them, DidSee rescales the noise schedule so the final step is pure noise, then trains the denoiser for a single step at that final timestep, so the network must produce the full depth latent in one pass. A semantic enhancer adds a parallel segmentation task so the model separates object from background where visual features are missing. If the paper's results hold, depth maps restored this way are good enough to improve category-level pose estimation and robotic grasping on materials ranging from diffuse to transparent.

What carries the argument

The load-bearing mechanism is the zero-terminal-SNR, noise-agnostic single-step formulation. With the rescaled scheduler, the final forward sample is $\mathbf{z}^y_T = \boldsymbol{\epsilon}$ (since $\bar{\alpha}_T = 0$), and the $v$-prediction target in Eq. (7) collapses from $v_T = \sqrt{\bar{\alpha}_T}\,\boldsymbol{\epsilon} + \sqrt{1-\bar{\alpha}_T}\,\mathbf{z}^y_0$ to $v_T = \mathbf{z}^y_0$, so the denoiser's output is directly the clean depth latent and is independent of the sampled noise. This removes exposure bias by training and evaluating at the same single timestep, and it lets the loss operate in pixel space. The semantic enhancer is a second mechanism: discrete segmentation labels are turned into RGB palette images so the VAE can encode them, and two task switchers plus cross-task self-attention let depth and semantic predictions share features, which the paper says gives sharper depth where transparent surfaces blend into backgrounds.

What would settle it

Run the trained DidSee at timestep $t=T$ with several different Gaussian noise draws and compare the decoded depth maps; the paper's formulation predicts $\hat{v}_T = \hat{\mathbf{z}}^y_0$ independent of $\boldsymbol{\epsilon}$, so if the outputs vary noticeably across noise seeds, the noise-agnostic single-step premise is false. A second check is to train with a nonzero terminal $\bar{\alpha}_T$ using otherwise identical settings and show that errors do not rise, which would undermine the signal-leakage explanation.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that the two training-inference mismatches inherited from image-generation diffusion — signal leakage bias and exposure bias — are first-order obstacles to depth completion, not minor artifacts, and that both can be eliminated jointly. DidSee enforces a zero terminal signal-to-noise ratio by rescaling the noise schedule, which makes the last forward step pure Gaussian noise; at that timestep the v-prediction target reduces to the clean depth latent itself, so the model is trained to map raw RGB-D latents plus noise directly to the depth latent in a single forward pass. Because the output is now the predicted clean latent, the model can be supervised with an L1 loss in pixel space instead of latent space. The paper further claims that a semantic enhancer operating through task switchers and cross-task self-attention lets one U-Net predict depth and semantic maps jointly, sharpening boundaries on transparent surfaces. Across TransCG, ClearPose, and DREDS/STD, DidSee reports the lowest errors among the compared methods, and restored depth from DidSee yields higher pose-estimation accuracy and grasping success than depth from earlier completers.

Load-bearing premise

The load-bearing premise is that one forward pass of the U-Net, starting from pure Gaussian noise plus the RGB and raw-depth latents, can directly produce a depth latent that the frozen VAE decodes into accurate depth; if the network cannot learn that underdetermined mapping, or if the random input noise makes predictions unstable, the central performance claims collapse.

Editorial extensions

If this is right

  • On TransCG, DidSee reports RMSE 0.011 versus 0.013 for TODE-Trans, with relative and mean absolute errors roughly a quarter lower, so even the strongest prior transparent-object completer is beaten by a clear margin.
  • On ClearPose's hardest scenes, including translucent cover and heavy occlusion, DidSee keeps the lowest errors and its accuracy thresholds stay well above the compared methods, implying the gains are largest exactly where visual cues fail.
  • Trained only on the synthetic DREDS-CatKnown split, DidSee remains competitive on real STD data and on novel object categories, evidence that the diffusion priors transfer without extra real training data.
  • Depth from DidSee improves category-level pose estimation over raw or completed depth in most reported metrics, and raises robotic grasping success on transparent, specular, and diffuse objects (overall 92.8% versus 89.2% for TODE-Trans).
  • Because inference is a single step, the method avoids the multi-step error accumulation shown in the paper and uses one denoising pass plus one VAE decode at test time.

Reading between the lines

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

  • The noise-agnostic single-step trick is not obviously limited to depth: any dense regression task that can be encoded in VAE latent space, such as surface normals or specular-reflection masks, could in principle use the same zero-terminal-SNR one-step training to avoid bias accumulation, though the paper does not test this.
  • If the single-step formulation truly makes output independent of the sampled noise, the model behaves more like a deterministic conditional generator than a diffusion sampler, and one could expect test-time ensembling over noise seeds to yield little or no gain; this is a measurable prediction the paper does not report.
  • The semantic enhancer's gain on transparent objects suggests that fusing a classification-style task into a regression diffusion model helps exactly where local RGB cues are ambiguous, so task switchers plus cross-task attention could be reused as a general module for other material-agnostic perception tasks.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The manuscript proposes DidSee, a Stable-Diffusion-based depth completion method for non-Lambertian (transparent/specular) objects. It identifies two biases in vanilla diffusion: signal leakage from a nonzero terminal SNR and exposure bias from multi-step sampling. It addresses them with a rescaled zero-terminal-SNR scheduler, a noise-agnostic single-step training formulation with a pixel-space L1 task loss, and a semantic enhancer that jointly predicts depth and palette-coded semantic maps via task switchers and cross-task attention. Experiments on TransCG, ClearPose, and DREDS/STD report state-of-the-art depth metrics, ablations for each component, and improved downstream category-level pose estimation and robotic grasping.

Significance. If the reported results hold, DidSee is a meaningful step: it shows that two diffusion-specific biases are material for depth completion and that a single-step zero-SNR formulation with a task-specific loss plus semantic conditioning can outperform task-specific transparent-object depth completion methods on three benchmarks and improve downstream robotics. The paper's strengths are the clear decomposition of the design (zero-SNR scheduler, single-step training, semantic enhancer), comprehensive comparison against strong baselines, and inclusion of downstream evaluations. However, the load-bearing issues below—the v-prediction reconstruction identity, the unexamined dependence on the random input noise at t=T, and the confounded ablation—must be resolved before the state-of-the-art claim is fully established; the semantic enhancer also lacks any semantic-level evaluation and training-data specification. No code is provided, so the reported single-number results cannot be independently reproduced at this stage.

major comments (4)
  1. [Sec. 3, Eqs. (2)-(4)] The reconstruction formula in Eq. (4) is inconsistent with the v-prediction target in Eq. (2). With v_t = sqrt(alpha_t) eps + sqrt(1-alpha_t) z0 and z_t = sqrt(alpha_t) z0 + sqrt(1-alpha_t) eps, substituting into hat_z0 = sqrt(alpha_t) z_t + sqrt(1-alpha_t) hat_v_t gives hat_z0 = z0 + 2 sqrt(alpha_t(1-alpha_t)) eps, not z0. The identity is exact only at the endpoints alpha_t=0 or 1, so the single-step t=T case is unaffected, but the multi-step baseline, Fig. 4, and the Table 6 baseline rely on this formula for intermediate timesteps. The authors should correct the sign convention (standard v-prediction uses v_t = sqrt(alpha_t) eps - sqrt(1-alpha_t) z0 and hat_z0 = sqrt(alpha_t) z_t - sqrt(1-alpha_t) hat_v_t) and recompute the affected experiments.
  2. [Sec. 3.2, Eq. (7)] The 'noise-agnostic' claim is not established for a finite-capacity network. Setting alpha_T=0 makes the training target v_T = z0 independent of eps, but the model input is z_T = eps, a fresh standard Gaussian sample. The optimal predictor is the conditional mean, but nothing in the U-Net architecture forces invariance to this uninformative input, and SGD may leave substantial sensitivity. The paper does not report the variance of the restored depth over noise realizations at inference, does not give the noise seed used for Tables 2-5, and reports all metrics without error bars. Please add a noise-invariance experiment (e.g., mean +/- std of RMSE/REL over several eps at fixed inputs) and report seeds; without this, the single-number SoTA results could be seed-dependent.
  3. [Table 6, Sec. 3.4] The '+Noise-Agnostic Single-step Training' ablation changes two variables simultaneously: the sampling schedule is moved to one step at t=T and the training objective is switched from the standard diffusion/latent objective to the pixel-space task-specific loss in Eq. (8). The REL drop from 0.020 to 0.017 therefore cannot be attributed solely to the single-step formulation, and the subsequent claim that this row demonstrates the adverse effect of exposure bias is not uniquely supported. Please run a multi-step baseline with the same task-specific loss, or otherwise decouple the loss change from the step-count change.
  4. [Sec. 3.3, Eq. (8)] The semantic enhancer is a load-bearing component (Table 6, last row), but the paper never states where the semantic ground-truth images y_s come from for TransCG, ClearPose, and DREDS/STD, and it reports no semantic segmentation metric (e.g., mIoU) or any predicted semantic map. Given that the training loss in Eq. (8) is an L1 regression on palette-colored images rather than a classification loss, the claim that the model 'jointly performs semantic segmentation' is not verifiable as presented. Please specify the semantic-label source/annotation protocol and report semantic accuracy or at least qualitative semantic outputs on the benchmark splits.
minor comments (5)
  1. [Tables 2-5, Table 8] All quantitative claims are reported as single numbers without error bars; for the robotic grasping results, the number of trials and standard deviations should also be reported.
  2. [Fig. 4] The experimental setup for Fig. 4 is underspecified: it should state which model and training loss are used, and whether each curve is an average over multiple noise samples.
  3. [Supplementary Eq. (9)] The notation f_theta([z_x, z_d], m_d) omits the noisy latent z_t and timestep t that appear in main-text Eq. (3); please clarify the relationship between these two notations.
  4. [Appendix D] The limitations section only mentions efficiency; it should also discuss the sensitivity to the random input noise at t=T and the dependence on semantic ground-truth labels.
  5. [References] Reference [9] has an empty author field ('Hyojun Go, , Yunsung Lee'), and Table 5 uses inconsistent capitalization ('Didsee' vs. 'DidSee').

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: DidSee's components are externally benchmarked and its derivations follow from stated equations, not from the target results.

full rationale

The paper's central claims are empirical comparisons against external baselines on TransCG, ClearPose, DREDS, and STD, with the method evaluated on held-out benchmark splits. The diffusion modifications are either mathematical consequences of the paper's own equations or adoptions of prior work: Eq. (7) sets v_T = z_0 when alpha_T = 0, which is a direct simplification, not a definition of the benchmark metrics in terms of the model output. The zero-terminal-SNR scheduler is taken from the cited external work [19], and the single-step formulation is a training choice supervised by ground-truth depth; no fitted parameter is renamed as a prediction. The only self-citation is [32] (color palette, by co-author Ruihao Xia), used to map discrete semantic labels to RGB images. This is a simple deterministic encoding step that is externally usable and not load-bearing for the paper's depth-completion performance claims; it does not invoke a uniqueness theorem or forbid alternatives. The ablation in Table 6 introduces the single-step formulation together with the pixel-space task-specific loss, which is an experimental attribution confound rather than a circularity, since the final numbers still come from external benchmark evaluation. Overall, the derivation chain is self-contained against external data and baselines, so no circular step is exhibited.

Assumptions & free parameters 1 free parameters · 3 assumptions · 0 invented entities

The central claim rests on the single-step mapping from pure noise to depth and on VAE representations of depth and semantic images. No new physical entities are introduced.

free parameters (1)
  • gamma = 0.1
    Balancing coefficient between depth L1 loss and semantic L1 loss in Eq. (8); set empirically across all datasets.
assumptions (3)
  • domain assumption A pre-trained Stable Diffusion U-Net, fine-tuned on a single timestep t=T, can learn to map pure Gaussian noise plus RGB and raw depth latents directly to a clean depth latent in one step.
    This is the core single-step training assumption; the paper's ablations support it empirically, but it is not derived.
  • domain assumption The Stable Diffusion VAE can encode depth maps and color-palette semantic maps into a latent space where decoding yields accurate predictions.
    The method relies on the VAE's ability to represent depth and semantic images, as in prior work [8, 11, 14], without fine-tuning the VAE.
  • domain assumption Preserving semantic information through a color palette and the VAE is sufficient for joint training to improve depth completion.
    The semantic enhancer uses a color palette [32] to convert discrete labels to RGB; the paper assumes the VAE retains the needed semantic signal.

how reviews work

0 comments
Cite this review

Pith. "Pith review of DidSee: Diffusion-Based Depth Completion for Material-Agnostic Robotic Perception and Manipulation." pith.science (2026). https://pith.science/paper/HLHGFKIN

@misc{pith2026250621034,
  author       = {Pith},
  title        = {Pith review of: DidSee: Diffusion-Based Depth Completion for Material-Agnostic Robotic Perception and Manipulation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HLHGFKIN}},
  note         = {Machine review of arXiv:2506.21034}
}
read the original abstract

Commercial RGB-D cameras often produce noisy, incomplete depth maps for non-Lambertian objects. Traditional depth completion methods struggle to generalize due to the limited diversity and scale of training data. Recent advances exploit visual priors from pre-trained text-to-image diffusion models to enhance generalization in dense prediction tasks. However, we find that biases arising from training-inference mismatches in the vanilla diffusion framework significantly impair depth completion performance. Additionally, the lack of distinct visual features in non-Lambertian regions further hinders precise prediction. To address these issues, we propose \textbf{DidSee}, a diffusion-based framework for depth completion on non-Lambertian objects. First, we integrate a rescaled noise scheduler enforcing a zero terminal signal-to-noise ratio to eliminate signal leakage bias. Second, we devise a noise-agnostic single-step training formulation to alleviate error accumulation caused by exposure bias and optimize the model with a task-specific loss. Finally, we incorporate a semantic enhancer that enables joint depth completion and semantic segmentation, distinguishing objects from backgrounds and yielding precise, fine-grained depth maps. DidSee achieves state-of-the-art performance on multiple benchmarks, demonstrates robust real-world generalization, and effectively improves downstream tasks such as category-level pose estimation and robotic grasping.

Figures

Figures reproduced from arXiv: 2506.21034 by the authors.

Figure 1
Figure 1. We propose DidSee, a diffusion-based depth completion framework for non-Lambertian objects. DidSee reduces depth restoration errors by mitigating signal leakage and exposure biases while incorporating a novel semantic enhancer (Right). Without scaling up data size, it generalizes robustly to in-the-wild scenes (Left) and facilitates material-agnostic robotic perception and manipulation in real-world scenarios. Abstr… view at source ↗
Figure 2
Figure 2. The overall framework of DidSee. During training, the pre-trained VAE encoder [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Comparison of original and rescaled noise schedules. (a) [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Prediction errors for different numbers of inference steps [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Depth maps without and with semantic enhancer. Mod [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 6
Figure 6. Figure 6: Qualitative comparison on STD-CatKnown dataset. Red bounding boxes delineate challenging non-Lambertian objects. [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Qualitative comparison on in-the-wild scenes. [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: Qualitative comparison on in-the-wild scenes (1/2). 2 [PITH_FULL_IMAGE:figures/full_fig_p012_8.png]
Figure 9
Figure 9. Figure 9: Qualitative comparison on in-the-wild scenes (2/2). 3 [PITH_FULL_IMAGE:figures/full_fig_p013_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

40 extracted references · 38 canonical work pages

  1. [19]

    Common diffusion noise schedules and sample steps are flawed

    Shanchuan Lin, Bingchen Liu, Jiashi Li, and Xiao Yang. Common diffusion noise schedules and sample steps are flawed. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) , pages 5404– 5411, 2024. 2, 3, 4

    work page 2024

  2. [8]

    Geowiz- ard: Unleashing the diffusion priors for 3d geometry estima- tion from a single image

    Xiao Fu, Wei Yin, Mu Hu, Kaixuan Wang, Yuexin Ma, Ping Tan, Shaojie Shen, Dahua Lin, and Xiaoxiao Long. Geowiz- ard: Unleashing the diffusion priors for 3d geometry estima- tion from a single image. In ECCV, pages 241–258, 2024. 2, 3, 5, 1

    work page 2024

  3. [20]

    Wonder3d: Single image to 3d using cross-domain diffusion

    Xiaoxiao Long, Yuan-Chen Guo, Cheng Lin, Yuan Liu, Zhiyang Dou, Lingjie Liu, Yuexin Ma, Song-Hai Zhang, Marc Habermann, Christian Theobalt, and Wenping Wang. Wonder3d: Single image to 3d using cross-domain diffusion. In CVPR, pages 9970–9980, 2024. 5, 1

    work page 2024

  4. [21]

    Fine-tuning image-conditional diffusion models is easier than you think

    Gonzalo Martin Garcia, Karim Abou Zeid, Christian Schmidt, Daan de Geus, Alexander Hermans, and Bastian Leibe. Fine-tuning image-conditional diffusion models is easier than you think. In Proceedings of the IEEE/CVF Win- ter Conference on Applications of Computer Vision (WACV), pages 753–762, 2025. 2, 3

    work page 2025

  5. [1]

    Tode-trans: Transparent object depth es- timation with transformer

    Kang Chen, Shaochen Wang, Beihao Xia, Dongxu Li, Zhen Kan, and Bin Li. Tode-trans: Transparent object depth es- timation with transformer. In IEEE International Confer- ence on Robotics and Automation (ICRA), pages 4880–4886,

    work page

  6. [2]

    Clearpose: Large-scale trans- parent object dataset and benchmark

    Xiaotong Chen, Huijie Zhang, Zeren Yu, Anthony Opipari, and Odest Chadwicke Jenkins. Clearpose: Large-scale trans- parent object dataset and benchmark. In ECCV, pages 381– 396, 2022. 2, 5, 8, 1

    work page 2022

  7. [3]

    Domain randomization- enhanced depth simulation and restoration for perceiving and grasping specular and transparent objects

    Qiyu Dai, Jiyao Zhang, Qiwei Li, Tianhao Wu, Hao Dong, Ziyuan Liu, Ping Tan, and He Wang. Domain randomization- enhanced depth simulation and restoration for perceiving and grasping specular and transparent objects. In ECCV, pages 374–391, 2022. 2, 5, 7, 8, 1

    work page 2022

  8. [4]

    Exploiting the signal-leak bias in diffusion models

    Martin Nicolas Everaert, Athanasios Fitsios, Marco Bocchio, Sami Arpa, Sabine S ¨usstrunk, and Radhakrishna Achanta. Exploiting the signal-leak bias in diffusion models. In Pro- ceedings of the IEEE/CVF Winter Conference on Applica- tions of Computer Vision (WACV), pages 4025–4034, 2024. 2, 4

    work page 2024

Show all 40 references
  1. [5]

    Transcg: A large-scale real-world dataset for transparent object depth completion and a grasping baseline

    Hongjie Fang, Hao-Shu Fang, Sheng Xu, and Cewu Lu. Transcg: A large-scale real-world dataset for transparent object depth completion and a grasping baseline. IEEE Robotics and Automation Letters, 7(3):7383–7390, 2022. 2, 5, 6, 7, 8, 1

    2022

  2. [6]

    Graspnet-1billion: A large-scale benchmark for general ob- ject grasping

    Hao-Shu Fang, Chenxi Wang, Minghao Gou, and Cewu Lu. Graspnet-1billion: A large-scale benchmark for general ob- ject grasping. In CVPR, pages 11441–11450, 2020. 8, 1

    2020

  3. [7]

    D-sco: Dual-stream conditional diffusion for monocular hand-held object reconstruction

    Bowen Fu, Gu Wang, Chenyangguang Zhang, Yan Di, Ziqin Huang, Zhiying Leng, Fabian Manhardt, Xiangyang Ji, and Federico Tombari. D-sco: Dual-stream conditional diffusion for monocular hand-held object reconstruction. In ECCV, pages 376–394, 2024. 1

    2024

  4. [9]

    Addressing nega- tive transfer in diffusion models

    Hyojun Go, , Yunsung Lee, Seunghyun Lee, Shinhyeok Oh, Hyeongdon Moon, and Seungtaek Choi. Addressing nega- tive transfer in diffusion models. In NeurIPS, pages 27199– 27222, 2023. 5

    2023

  5. [10]

    Efficient diffu- sion training via min-snr weighting strategy

    Tiankai Hang, Shuyang Gu, Chen Li, Jianmin Bao, Dong Chen, Han Hu, Xin Geng, and Baining Guo. Efficient diffu- sion training via min-snr weighting strategy. In ICCV, pages 7441–7451, 2023. 4, 5

    2023

  6. [11]

    Lotus: Diffusion-based visual foundation model for high-quality dense prediction

    Jing He, Haodong Li, Wei Yin, Yixun Liang, Leheng Li, Kaiqiang Zhou, Hongbo Liu, Bingbing Liu, and Ying- Cong Chen. Lotus: Diffusion-based visual foundation model for high-quality dense prediction. arXiv preprint arXiv:2409.18124, 2024. 2, 3, 5

    2024 arXiv

  7. [12]

    Fourier transporter: Bi-equivariant robotic manipulation in 3d

    Haojie Huang, Owen Lewis Howell, Dian Wang, Xupeng Zhu, Robert Platt, and Robin Walters. Fourier transporter: Bi-equivariant robotic manipulation in 3d. In ICLR, 2024. 1

    2024

  8. [13]

    Robotic perception of transparent objects: A review

    Jiaqi Jiang, Guanqun Cao, Jiankang Deng, Thanh-Toan Do, and Shan Luo. Robotic perception of transparent objects: A review. IEEE Transactions on Artificial Intelligence , 5(6): 2547–2567, 2024. 2

    2024

  9. [14]

    Repurpos- ing diffusion-based image generators for monocular depth estimation

    Bingxin Ke, Anton Obukhov, Shengyu Huang, Nando Met- zger, Rodrigo Caye Daudt, and Konrad Schindler. Repurpos- ing diffusion-based image generators for monocular depth estimation. In CVPR, pages 9492–9502, 2024. 2, 3

    2024

  10. [15]

    Open3dsg: Open- vocabulary 3d scene graphs from point clouds with queryable objects and open-set relationships

    Sebastian Koch, Narunas Vaskevicius, Mirco Colosi, Pe- dro Hermosilla, and Timo Ropinski. Open3dsg: Open- vocabulary 3d scene graphs from point clouds with queryable objects and open-set relationships. In CVPR, pages 14183–14193, 2024. 1

    2024

  11. [16]

    Alleviating exposure bias in diffusion mod- els through sampling with shifted time steps

    Mingxiao Li, Tingyu Qu, Ruicong Yao, Wei Sun, and Marie- Francine Moens. Alleviating exposure bias in diffusion mod- els through sampling with shifted time steps. In ICLR, 2024. 2, 4

    2024

  12. [17]

    Fdct: Fast depth completion for transparent objects.IEEE Robotics and Automation Letters, 8(9):1893–1912, 2023

    Tianan Li, Zhehan Chen, Huan Liu, and Chen Wang. Fdct: Fast depth completion for transparent objects.IEEE Robotics and Automation Letters, 8(9):1893–1912, 2023. 2, 5, 6, 7, 8, 1

    1912

  13. [18]

    On error propa- gation of diffusion models

    Yangming Li and Mihaela van der Schaar. On error propa- gation of diffusion models. In ICLR, 2024. 2, 4

    2024

  14. [22]

    Elucidating the exposure bias in diffusion models

    Mang Ning, Mingxiao Li, Jianlin Su, Albert Ali Salah, and Itir Onal Ertugrul. Elucidating the exposure bias in diffusion models. In ICLR, 2024. 2, 4

    2024

  15. [23]

    Discovering clone negatives via adaptive contrastive learning for image-text matching

    Renjie Pan, Jihao Dong, and Hua Yang. Discovering clone negatives via adaptive contrastive learning for image-text matching. In ICLR, 2025. 1

    2025

  16. [24]

    Switch diffusion trans- former: Synergizing denoising tasks with sparse mixture-of- experts

    Byeongjun Park, Hyojun Go, Jin-Young Kim, Sangmin Woo, Seokil Ham, and Changick Kim. Switch diffusion trans- former: Synergizing denoising tasks with sparse mixture-of- experts. In ECCV, pages 461–477, 2024. 5

    2024

  17. [25]

    High-resolution image syn- thesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj¨orn Ommer. High-resolution image syn- thesis with latent diffusion models. In CVPR, pages 10684– 10695, 2022. 2, 3, 5, 6, 1

    2022

  18. [26]

    Clear grasp: 3d shape estimation of transparent objects for manip- ulation

    Shreeyak Sajjan, Matthew Moore, Mike Pan, Ganesh Na- garaja, Johnny Lee, Andy Zeng, and Shuran Song. Clear grasp: 3d shape estimation of transparent objects for manip- ulation. In IEEE International Conference on Robotics and Automation (ICRA), pages 3634–3642, 2020. 2, 5, 6

    2020

  19. [27]

    Progressive distillation for fast sampling of diffusion models

    Tim Salimans and Jonathan Ho. Progressive distillation for fast sampling of diffusion models. In ICLR, 2022. 3 9

    2022

  20. [28]

    Mor- pheus: Neural dynamic 360deg surface reconstruction from monocular rgb-d video

    Hengyi Wang, Jingwen Wang, and Lourdes Agapito. Mor- pheus: Neural dynamic 360deg surface reconstruction from monocular rgb-d video. In CVPR, pages 20965–20976,

  21. [29]

    Gˆ3-lq: Marrying hyperbolic alignment with explicit semantic-geometric mod- eling for 3d visual grounding

    Yuan Wang, Yali Li, and Shengjin Wang. Gˆ3-lq: Marrying hyperbolic alignment with explicit semantic-geometric mod- eling for 3d visual grounding. InCVPR, pages 13917–13926,

  22. [30]

    D3roma: Disparity diffusion-based depth sensing for material-agnostic robotic manipulation

    Songlin Wei, Haoran Geng, Jiayi Chen, Congyue Deng, Wenbo Cui, Chengyang Zhao, Xiaomeng Fang, Leonidas Guibas, and He Wang. D3roma: Disparity diffusion-based depth sensing for material-agnostic robotic manipulation. In Proceedings of the 8th Conference on Robot Learning (CoRL)...

    2024

  23. [31]

    Layeredflow: A real-world benchmark for non-lambertian multi-layer optical flow

    Hongyu Wen, Erich Liang, and Jia Deng. Layeredflow: A real-world benchmark for non-lambertian multi-layer optical flow. In ECCV, pages 477–495, 2025. 2

    2025

  24. [32]

    Unsupervised modality adapta- tion with text-to-image diffusion models for semantic seg- mentation

    Ruihao Xia, Yu Liang, Peng-Tao Jiang, Hao Zhang, Bo Li, Yang Tang, and Pan Zhou. Unsupervised modality adapta- tion with text-to-image diffusion models for semantic seg- mentation. In NeurIPS, pages 93964–93984, 2024. 5

    2024

  25. [33]

    What matters when repurposing diffusion models for general dense perception tasks? arXiv preprint arXiv:2403.06090 ,

    Guangkai Xu, Yongtao Ge, Mingyu Liu, Chengxiang Fan, Kangyang Xie, Zhiyue Zhao, Hao Chen, and Chunhua Shen. What matters when repurposing diffusion models for general dense perception tasks? arXiv preprint arXiv:2403.06090 ,

    arXiv

  26. [34]

    Seeing glass: Joint point-cloud and depth completion for transparent objects

    Haoping Xu, Yi Ru Wang, Sagi Eppel, Alan Aspuru-Guzik, Florian Shkurti, and Animesh Garg. Seeing glass: Joint point-cloud and depth completion for transparent objects. In Proceedings of the 5th Conference on Robot Learning (CoRL), pages 827–838, 2022. 2, 6

    2022

  27. [35]

    3d diffusion policy: Gen- eralizable visuomotor policy learning via simple 3d repre- sentations

    Yanjie Ze, Gu Zhang, Kangning Zhang, Chenyuan Hu, Muhan Wang, and Huazhe Xu. 3d diffusion policy: Gen- eralizable visuomotor policy learning via simple 3d repre- sentations. In Proceedings of Robotics: Science and Systems (RSS), 2024. 1

    2024

  28. [36]

    Tcrnet: Transparent object depth completion with cascade refinements

    Di-Hua Zhai, Sheng Yu, Wei Wang, Yuyin Guan, and Yuan- qing Xia. Tcrnet: Transparent object depth completion with cascade refinements. IEEE Transactions on Automation Sci- ence and Engineering, 22:1–20, 2024. 2, 6

    2024

  29. [37]

    Omni6dpose: A benchmark and model for universal 6d object pose estima- tion and tracking

    Jiyao Zhang, Weiyao Huang, Bo Peng, Mingdong Wu, Fei Hu, Zijian Chen, Bo Zhao, and Hao Dong. Omni6dpose: A benchmark and model for universal 6d object pose estima- tion and tracking. In ECCV, pages 199–216, 2024. 1

    2024

  30. [38]

    Beta-tuned timestep diffusion model

    Tianyi Zheng, Peng-Tao Jiang, Ben Wan, Hao Zhang, Jinwei Chen, Jia Wang, and Bo Li. Beta-tuned timestep diffusion model. In ECCV, pages 114–130, 2024. 5

    2024

  31. [39]

    Point cloud matters: Rethinking the impact of different observation spaces on robot learning

    Haoyi Zhu, Yating Wang, Di Huang, Weicai Ye, Wanli Ouyang, and Tong He. Point cloud matters: Rethinking the impact of different observation spaces on robot learning. In NeurIPS, pages 77799–77830, 2024. 1

    2024

  32. [40]

    Rgb-d local implicit function for depth completion of transparent objects

    Luyang Zhu, Arsalan Mousavian, Yu Xiang, Hammad Mazhar, Jozef van Eenbergen, Shoubhik Debnath, and Di- eter Fox. Rgb-d local implicit function for depth completion of transparent objects. In CVPR, pages 4649–4658, 2021. 2, 6, 7, 8 10 DidSee: Diffusion-Based Depth Completion fo...

    2021

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.