Pith. sign in

REVIEW 4 major objections 6 minor 36 references

Image-Guided Depth Sampling and Reconstruction

T0 review · 4 major / 6 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read The paper claims that sampling one depth value per RGB superpixel, then smoothing with a log-domain bilateral filter, reconstructs depth with 3-4 times fewer samples than state-of-the-art image-guided depth completion.

desk verdict Solid adaptive depth sampling paper, but the headline 3-4x claim conflates sampling with reconstruction and needs an ablation before it should be accepted as stated. read the letter →

arxiv 1908.01379 v1 pith:7JRPXFQ7 submitted 2019-08-04 cs.CV eess.IV

Adam Wolff , Shachar Praisler , Ilya Tcenov , Guy Gilboa This is my paper · ORCID
classification cs.CVeess.IV
keywords adaptivedepthsamplingcompletionsuperpixelspiecewiseplanarmodelbilateralfilterRGB-guidedLiDARsparsereconstruction
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 asks how few depth measurements a navigation system truly needs when a color camera can say where to look. It argues that depth scenes are mostly piecewise planar, that 20-60 planar regions describe a typical frame, and therefore that an optimal sampler would need only about one depth reading per 1200 pixels. The proposed practical scheme does not reach that bound, but it cuts the required sampling rate by a factor of 3-4 relative to existing image-guided depth completion methods by taking one measurement per superpixel center of mass and filtering in the logarithmic domain. A physical single-pixel scanning prototype demonstrates the idea outside simulation.

What carries the argument

The central object is the superpixel map, computed by SLIC, which simultaneously defines where samples are taken (superpixel centers of mass) and the initial zero-order reconstruction (each superpixel takes its sample's depth value). The second mechanism is a bilateral filter applied in the log domain, which smooths staircase artifacts on large planar walls and roads while preserving real depth discontinuities. These two pieces are justified by a piecewise planar model of depth scenes, validated statistically on dense indoor and outdoor data.

What would settle it

Measure the required sample count on a benchmark deliberately filled with camouflaged objects, where depth changes occur with no visible RGB edge; if the 3-4x sample savings disappears or reverses, the central claim is bounded to scenes with visible color boundaries. Alternatively, fit the piecewise planar model to highly curved scenes such as vegetation or cables and check whether the number of planes N grows beyond the claimed 20-60 range.

Watch

Extended reading notes

Core claim

Depth sampling and reconstruction can be driven by a superpixel segmentation of the RGB image. The method's sampling places one depth probe at the center of mass of each superpixel; reconstruction assigns that depth to the entire superpixel, then applies a bilateral filter to the logarithm of the depth-plus-one map and exponentiates back. The paper reports that this simple pipeline achieves state-of-the-art RMSE on indoor and outdoor benchmarks (NYU-Depth-v2 and Synthia), needing 3-4 times fewer samples than competing methods for a given accuracy, especially on small obstacles. Under the paper's piecewise planar model, three samples per plane give a theoretical lower bound of roughly 56-200 samples per frame, about 0.08 percent of pixels, which marks how much room remains below the achieved sampling rates.

Load-bearing premise

The load-bearing premise is that RGB superpixel boundaries align closely with true depth discontinuities and that one depth reading at a superpixel's center of mass represents the whole segment; the paper itself shows a camouflaged object where this fails and the method degenerates to ordinary grid sampling.

Editorial extensions

If this is right

  • Solid-state LiDAR with programmable scanning can use an RGB camera to cut scan time and energy by a factor of 3-4 at equal reconstruction accuracy.
  • Because the sampling pattern alone improves other completion methods, the superpixel center-of-mass pattern is a generic front-end that can be paired with any reconstruction algorithm.
  • The 1/1200 optimal sampling ratio gives a concrete target for future adaptive samplers, indicating a large gap between current practical methods and the theoretical lower bound.
  • Reliable reconstruction of thin obstacles like poles and signs at low sampling budgets improves detection of small hazards in autonomous navigation.

Reading between the lines

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

  • If one quantified per frame how well superpixel boundaries match depth discontinuities, the 3-4x savings could be predicted in advance, and the sampler could gracefully fall back to grid sampling on poorly matched frames—a testable extension the paper does not run.
  • The log-domain bilateral smoothing approximates relative depth error; filtering in inverse-depth or disparity space might yield similar or better accuracy, especially for close objects.
  • The 20-60 plane model suggests a compressed depth representation for transmission: rather than dense depth, send plane parameters plus a superpixel segmentation, which could approach the theoretical sampling bound.
  • The paper's assumptions of clean range readings are optimistic; under realistic LiDAR noise the advantage may shrink, so a study of sample-noise sensitivity would bound the method's practical gain.
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

4 major / 6 minor

Summary. The paper proposes an adaptive depth sampling and reconstruction framework guided by RGB superpixels. The authors first introduce a piecewise planar depth model, estimate on Synthia and NYU-Depth-v2 that 20-60 planar regions approximate typical indoor and outdoor depth maps, and derive an optimal sampling ratio near 1/1200 from the 3N lower bound. The proposed algorithm samples once at the center of mass of each SLIC superpixel and reconstructs by zero-order assignment followed by bilateral filtering in the log domain. Experiments compare against several depth completion baselines on Synthia and NYU, reporting lower RMSE and, on a Synthia obstacle subset, a 3-4x reduction in required sampling density. A single-pixel mechanical depth sampler demonstrates the concept in a laboratory setting.

Significance. If the central claims hold, the work offers a simple, non-learning, image-guided sampling strategy that is relevant to emerging solid-state LiDAR systems, and it provides a falsifiable prediction about the number of planar structures needed for depth approximation. The paper's explicit strengths are the breadth of the evaluation (multiple reconstruction baselines, indoor/outdoor data, a resolution analysis via MTF, and a physical prototype) and the consistent qualitative gains shown across settings. The main weaknesses are that the headline sample-reduction factor is not cleanly attributed to the reconstruction rather than the sampling pattern, and several implementation details needed for reproducibility are missing. The contribution is useful and appears technically sound in its core comparison, but the claims need to be re-scoped or supported by additional ablations before publication.

major comments (4)
  1. [Sec. 5.1 (Table 1) and Abstract] The statement that 'our proposed reconstruction achieves state-of-the-art results... reducing the required sampling rate by a factor of 3-4' attributes the gain to the reconstruction, but the comparison in Table 1 varies both the sampling pattern and the reconstruction jointly. For all baselines the paper simulates uniform random samples, while the proposed method samples once per RGB superpixel at the center of mass; Sec. 5.1 and Fig. 14 show that this sampling pattern alone improves other completion methods. Consequently, the 3-4x factor in Table 1 is not evidence that the zero-order-plus-bilateral reconstruction is responsible for the gain. Please provide an ablation that fixes the sampling pattern (e.g., our sampling with the baselines' reconstructions, or our reconstruction with random/grid samples) and report sample-requirement curves for that ablation.
  2. [Sec. 3.1] The procedure used to compute the piecewise planar approximations is not described. The paper reports average N, δ, and ε for Synthia and NYU, but without specifying how the partition E and plane parameters (a_i, b_i, c_i) are estimated, how the validity set V is determined, and what optimization or segmentation algorithm is used, these numbers and the derived 1/1200 optimal sampling ratio cannot be reproduced. This is load-bearing because Eq. (3) and the model-validity claim rest on these estimates.
  3. [Sec. 4.2 (R.3) and Sec. 5] The bilateral filter parameters are never given; the text only says they are 'fixed for a given number of samples n and type of scene'. Since the reconstruction is central to the claimed improvements, the spatial and range standard deviations (and any iteration count) must be reported for each experiment. The SLIC compactness parameter in step S.1 is also unspecified. Without these values, the method cannot be reimplemented or compared fairly.
  4. [Sec. 5.1 (Figs. 12a/12b, Table 1)] The 3-4x sample reduction is demonstrated only on the Synthia obstacles subset. On the full Synthia scene the improvement at fixed density is about 30% RMSE (Fig. 12a), and on NYU-Depth-v2 (Table 2) the gain at a fixed budget of 200 samples is modest. The abstract and introduction state the 3-4x factor without this scope restriction. Please qualify the claim to the obstacles evaluation or provide equivalent density-vs-RMSE curves for the full scenes, including the sample counts required for a given RMSE.
minor comments (6)
  1. [Sec. 1] In the sentence 'An example of output of a our algorithm', 'a our' should be 'our'.
  2. [Sec. 3.1, Eq. (3)] The notation 'Number−of−Samples' is awkward; consider using a mathematical symbol such as n_min.
  3. [Fig. 3 caption] The caption says 'percentage of approximated planes N per image' but the correct term is 'number of approximated planes'.
  4. [Sec. 4.2, step S.1] The text says the desired number of SPs is set to n, but it should clarify how the actual sample count relates to n when a center of mass falls outside a non-convex superpixel.
  5. [Sec. 5.1] Uniform random sampling is stochastic, so the baseline curves in Figs. 12 and Table 1 should include error bars or be averaged over multiple trials to establish that the reported differences are not due to sampling noise.
  6. [Sec. 5.2, Table 2] Please specify the training data and preprocessing for the deep-learning baselines, since a fair comparison requires that they are evaluated under conditions consistent with the NYU test set.

Circularity Check

1 steps flagged · score 2.0 of 10

Minor presentational circularity: the abstract's 'prediction' of 20-60 planar structures and 1/1200 sampling ratio is a restatement of fitted model parameters plus the definitional lower bound; the central sampling/reconstruction result is not circular.

  1. fitted input called prediction [Abstract; Section 3.1, Eq. (3)]
    "Our model and experiments predict that, in the optimal case, about 20-60 piece-wise linear structures can approximate well a depth map. This translates to a depth-to-image sampling ratio of about 1/1200."

    The 20-60 range is not an independent prediction: it is the fitted number of planar regions recovered by the piecewise-planar approximation in Section 3.1 (average N=66.6 on Synthia and N=18.5 on NYU). The '1/1200' ratio is obtained by applying the definitional lower bound nmin>=3N from Eq. (3) to those fitted N values and dividing by the image area. Thus the abstract's 'prediction' is an algebraic restatement of the fitted model parameters, not a separate forecast that could confirm the model. The paper is transparent that these are recovered parameters ('average model parameters recovered'), so the circularity is presentational and confined to a secondary claim; the sampling and reconstruction comparisons do not depend on it.

full rationale

The central contribution is not circular. The sampling pattern is generated by SLIC superpixels, the reconstruction is zero-order assignment followed by log-domain bilateral filtering, and no parameter is fitted to the evaluation benchmark in a way that forces the reported RMSE comparisons. The piecewise-planar model is an empirical hypothesis test, not used to tune the final algorithm. The only reduction-by-construction I found is the abstract's wording that the model 'predicts' 20-60 structures and a 1/1200 ratio, which is just the fitted N plus Eq. (3). The headline 3-4x factor compares the full proposed pipeline against baselines given uniform random samples, while Fig. 14 shows the proposed sampling pattern alone improves those baselines; that is an attribution/experimental-design concern rather than a circular derivation, since the factor is measured, not derived from the model. There is a minor self-citation to prior work by one author ([14]) but it is not load-bearing. Overall, no significant circularity in the main method; the only issue is a minor presentational fitted-input-as-prediction in a secondary claim, hence score 2.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

No new physical entities, particles, forces or conserved quantities are introduced. The only 'invention' is an algorithmic sampling and reconstruction scheme. The main free parameters are model tolerances and algorithm settings, several of which are underreported.

free parameters (5)
  • Model validity tolerance delta = 0.1 (Synthia), 0.07 (NYU-v2)
    Threshold for fraction of pixels allowed to violate the piecewise planar model; chosen by the authors and estimated from the same datasets used for evaluation.
  • Model error tolerance epsilon = 1.35 m (Synthia), 0.18 m (NYU-v2)
    Allowed RMSE of the piecewise planar approximation; selected post hoc to make the model 'valid' on the benchmark data.
  • Number of piecewise planar regions N = 66.6 (Synthia), 18.5 (NYU-v2)
    Estimated per image from an unspecified piecewise planar fitting procedure; it directly determines the claimed optimal sample count nmin=3N.
  • Bilateral filter parameters = not reported
    Fixed per sample count and scene type (Section 4.2 R.3), but values and selection method are not provided; results depend on this tuning.
  • SLIC compactness = 'high value'
    Set to produce regularly shaped superpixels (Section 4.2 S.1), exact value not reported.
assumptions (4)
  • domain assumption Depth maps are piecewise planar almost everywhere
    Section 3.1 states the hypothesis; it is validated empirically on two datasets but not derived.
  • domain assumption RGB edges predict depth discontinuities well enough for superpixel boundaries to approximate the depth partition
    Section 3.2 reports conditional probabilities Prgb-d and Pd-rgb; the low Pd-rgb (22.9-33.7%) means many RGB edges are false depth boundaries, an acknowledged limitation.
  • domain assumption A single depth sample at the superpixel center of mass is representative of the whole segment
    Section 4.3.1 justifies this for planar regions and near discontinuities, but it breaks for camouflaged objects (Fig. 9).
  • domain assumption Depth measurements are noiseless relative to reconstruction error
    Assumption 1 in Section 4; it simplifies the model but is not validated against real sensor noise.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Image-Guided Depth Sampling and Reconstruction." pith.science (2026). https://pith.science/paper/7JRPXFQ7

@misc{pith2026190801379,
  author       = {Pith},
  title        = {Pith review of: Image-Guided Depth Sampling and Reconstruction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7JRPXFQ7}},
  note         = {Machine review of arXiv:1908.01379}
}
read the original abstract

Depth acquisition, based on active illumination, is essential for autonomous and robotic navigation. LiDARs (Light Detection And Ranging) with mechanical, fixed, sampling templates are commonly used in today's autonomous vehicles. An emerging technology, based on solid-state depth sensors, with no mechanical parts, allows fast, adaptive, programmable scans. In this paper, we investigate the topic of adaptive, image-driven, sampling and reconstruction strategies. First, we formulate a piece-wise linear depth model with several tolerance parameters and estimate its validity for indoor and outdoor scenes. Our model and experiments predict that, in the optimal case, about 20-60 piece-wise linear structures can approximate well a depth map. This translates to a depth-to-image sampling ratio of about 1/1200. We propose a simple, generic, sampling and reconstruction algorithm, based on super-pixels. We reach a sampling rate which is still far from the optimal case. However, our sampling improves grid and random sampling, consistently, for a wide variety of reconstruction methods. Moreover, our proposed reconstruction achieves state-of-the-art results, compared to image-guided depth completion algorithms, reducing the required sampling rate by a factor of 3-4. A single-pixel depth camera built in our lab illustrates the concept.

Figures

Figures reproduced from arXiv: 1908.01379 by the authors.

Figure 1
Figure 1. We developed an algorithm for RGB-based depth sampling and reconstruc [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Examples for piece-wise planar approximation. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Statistics of piece-wise planar approximation. Top: percentage of approxi [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (14 more)
Figure 4
Figure 4. Figure 4: Examples of image parts which fit (top) and parts which do not fit (bottom) [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Examples of depth discontinuities (red) and RGB edges (green) correlation. [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Algorithm block diagram. 4.2 Proposed algorithm The proposed algorithm is divided into two parts, sampling and reconstruction. It in￾cludes the following steps: • Sampling: S.1 A super-pixel map is generated from the RGB image using SLIC [1]. The desired number of SPs …
Figure 7
Figure 7. Figure 7: Reconstruction variants study on NYU-Depth-v2 dataset. We compare 1st [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]
Figure 8
Figure 8. Figure 8: Toy example 1: Comparing 1st-order estimation of larger segments (middle [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 9
Figure 9. Figure 9: Toy example 2: Camouflaged object. In some rare cases, the object is cam [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]
Figure 10
Figure 10. Figure 10: presents the results computed from the reconstructed images in [PITH_FULL_IMAGE:figures/full_fig_p011_10.png]
Figure 11
Figure 11. Figure 11: Qualitative results for depth reconstruction on our modified MTF road-like [PITH_FULL_IMAGE:figures/full_fig_p012_11.png]
Figure 12
Figure 12. Figure 12: Quantitative comparison on Synthia (left) and obstacles (right) datasets [PITH_FULL_IMAGE:figures/full_fig_p013_12.png]
Figure 13
Figure 13. Figure 13: Qualitative results of depth completion on Synthia and our obstacles dataset [PITH_FULL_IMAGE:figures/full_fig_p013_13.png]
Figure 14
Figure 14. Figure 14: Sampling methods comparison: 4 sampling patterns are compared for dif￾ferent reconstruction method: uniform random, grid, Liu et al. [22] and ours (sampling only). The reconstructions are bilinear interp., IP-Basic [19] and L1diag [24] 5.2 Indoor data (NYU-Depth-v2) T…
Figure 15
Figure 15. Figure 15: Qualitative results of depth completion on NYU-Depth-v2 dataset at 0.7% [PITH_FULL_IMAGE:figures/full_fig_p015_15.png]
Figure 16
Figure 16. Figure 16: Our mechanical sampler. 15 [PITH_FULL_IMAGE:figures/full_fig_p015_16.png]
Figure 17
Figure 17. Figure 17: Experimental results on simple (top) and challenging (bottom) scenes taken [PITH_FULL_IMAGE:figures/full_fig_p016_17.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references · 31 canonical work pages

  1. [1]

    Achanta, A

    R. Achanta, A. Shaji, K. Smith, A. Lucchi, P. Fua, S. S ¨usstrunk, et al. Slic superpixels compared to state-of-the-art superpixel methods. IEEE transactions on pattern analysis and machine intelligence, 34(11):2274–2282, 2012

    work page 2012

  2. [2]

    Aldroubi and K

    A. Aldroubi and K. Gr ¨ochenig. Nonuniform sampling and reconstruction in shift-invariant spaces. SIAM review, 43(4):585–620, 2001

    work page 2001

  3. [3]

    I. Amidror. Scattered data interpolation methods for electronic imaging systems: a survey. Journal of electronic imaging, 11(2):157–177, 2002

    work page 2002

  4. [4]

    Babu and P

    P. Babu and P. Stoica. Spectral analysis of nonuniformly sampled data–a review. Digital Signal Processing, 20(2):359–378, 2010

    work page 2010

  5. [5]

    Baker, R

    S. Baker, R. Szeliski, and P. Anandan. A layered approach to stereo reconstruction. In Proceedings. 1998 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (Cat. No. 98CB36231), pages 434–441. IEEE, 1998

    work page 1998

  6. [6]

    J. T. Barron and B. Poole. The fast bilateral solver. In European Conference on Computer Vision, pages 617–632. Springer, 2016

    work page 2016

  7. [7]

    G. D. Boreman. Modulation transfer function in optical and electro-optical systems , vol- ume 21. SPIE press Bellingham, W A, 2001

    work page 2001

  8. [8]

    E. J. Candes, J. K. Romberg, and T. Tao. Stable signal recovery from incomplete and inaccurate measurements. Communications on Pure and Applied Mathematics: A Journal Issued by the Courant Institute of Mathematical Sciences, 59(8):1207–1223, 2006

    work page 2006

Show all 36 references
  1. [9]

    Cheben, R

    P. Cheben, R. Halir, J. H. Schmid, H. A. Atwater, and D. R. Smith. Subwavelength inte- grated photonics. Nature, 560(7720):565, 2018

    2018

  2. [10]

    Z. Chen, V . Badrinarayanan, G. Drozdov, and A. Rabinovich. Estimating depth from rgb and sparse sensing. arXiv preprint arXiv:1804.02771, 2018

    2018 arXiv

  3. [11]

    Chodosh, C

    N. Chodosh, C. Wang, and S. Lucey. Deep convolutional compressed sensing for lidar depth completion. arXiv preprint arXiv:1803.08949, 2018

    2018 arXiv

  4. [12]

    Cordts, M

    M. Cordts, M. Omran, S. Ramos, T. Rehfeld, M. Enzweiler, R. Benenson, U. Franke, S. Roth, and B. Schiele. The cityscapes dataset for semantic urban scene understanding. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 3213–3223, 2016

    2016

  5. [13]

    Doll ´ar and C

    P. Doll ´ar and C. L. Zitnick. Structured forests for fast edge detection. In Proceedings of the IEEE international conference on computer vision, pages 1841–1848, 2013

    2013

  6. [14]

    Drozdov, Y

    G. Drozdov, Y . Shapiro, and G. Gilboa. Robust recovery of heavily degraded depth mea- surements. In 3D Vision (3DV), 2016 Fourth International Conference on , pages 56–65. IEEE, 2016

    2016

  7. [15]

    Eldesokey, M

    A. Eldesokey, M. Felsberg, and F. S. Khan. Propagating confidences through cnns for sparse data regression. arXiv preprint arXiv:1805.11913, 2018

    2018 arXiv

  8. [16]

    S. Hawe, M. Kleinsteuber, and K. Diepold. Dense disparity maps from sparse disparity measurements. In 13th International Conference on Computer Vision, 2011

    2011

  9. [17]

    Huang, J

    Z. Huang, J. Fan, S. Yi, X. Wang, and H. Li. Hms-net: Hierarchical multi-scale sparsity- invariant network for sparse depth completion. arXiv preprint arXiv:1808.08685, 2018

    2018 arXiv

  10. [18]

    Jaritz, R

    M. Jaritz, R. De Charette, E. Wirbel, X. Perrotton, and F. Nashashibi. Sparse and dense data with cnns: Depth completion and semantic segmentation. In 2018 International Con- ference on 3D Vision (3DV), pages 52–60. IEEE, 2018

    2018

  11. [19]

    J. Ku, A. Harakeh, and S. L. Waslander. In defense of classical image processing: Fast depth completion on the cpu. arXiv preprint arXiv:1802.00036, 2018

    2018 arXiv

  12. [20]

    Y . Li, K. Qian, T. Huang, and J. Zhou. Depth estimation from monocular image and coarse depth points based on conditional gan. In MATEC Web of Conferences, volume 175, page 03055. EDP Sciences, 2018. 18

    2018

  13. [21]

    Y . Liao, L. Huang, Y . Wang, S. Kodagoda, Y . Yu, and Y . Liu. Parse geometry from a line: Monocular depth estimation with partial laser observation. In 2017 IEEE International Conference on Robotics and Automation (ICRA), pages 5059–5066. IEEE, 2017

    2017

  14. [22]

    L.-K. Liu, S. H. Chan, and T. Q. Nguyen. Depth reconstruction from sparse sam- ples: Representation, algorithm, and sampling. IEEE Transactions on Image Processing, 24(6):1983–1996, 2015

    1983

  15. [23]

    Loebich, D

    C. Loebich, D. Wueller, B. Klingen, and A. Jaeger. Digital camera resolution measurement using sinusoidal siemens stars. In Digital Photography III , volume 6502, page 65020N. International Society for Optics and Photonics, 2007

    2007

  16. [24]

    F. Ma, L. Carlone, U. Ayaz, and S. Karaman. Sparse depth sensing for resource-constrained robots. arXiv preprint arXiv:1703.01398, 2017

    2017 arXiv

  17. [25]

    F. Ma, G. V . Cavalheiro, and S. Karaman. Self-supervised sparse-to-dense: Self-supervised depth completion from lidar and monocular camera. arXiv preprint arXiv:1807.00275 , 2018

    2018 arXiv

  18. [26]

    Ma and S

    F. Ma and S. Karaman. Sparse-to-dense: Depth prediction from sparse depth samples and a single image. In 2018 IEEE International Conference on Robotics and Automation (ICRA), pages 1–8. IEEE, 2018

    2018

  19. [27]

    Marvasti

    F. Marvasti. Nonuniform sampling: theory and practice . Springer Science & Business Media, 2012

    2012

  20. [28]

    Mishali and Y

    M. Mishali and Y . C. Eldar. From theory to practice: Sub-nyquist sampling of sparse wideband analog signals. IEEE Journal of Selected Topics in Signal Processing, 4(2):375– 391, 2010

    2010

  21. [29]

    C. V . Poulton, A. Yaacobi, D. B. Cole, M. J. Byrd, M. Raval, D. Vermeulen, and M. R. Watts. Coherent solid-state lidar with silicon photonic optical phased arrays.Optics letters, 42(20):4091–4094, 2017

    2017

  22. [30]

    G. Ros, L. Sellart, J. Materzynska, D. Vazquez, and A. M. Lopez. The synthia dataset: A large collection of synthetic images for semantic segmentation of urban scenes. InProceed- ings of the IEEE conference on computer vision and pattern recognition, pages 3234–3243, 2016

    2016

  23. [31]

    B. Schwarz. Lidar: Mapping the world in 3d. Nature Photonics, 4(7):429, 2010

    2010

  24. [32]

    Silberman, D

    N. Silberman, D. Hoiem, P. Kohli, and R. Fergus. Indoor segmentation and support in- ference from rgbd images. In European Conference on Computer Vision, pages 746–760. Springer, 2012

    2012

  25. [33]

    H. Tao, H. S. Sawhney, and R. Kumar. A global matching framework for stereo compu- tation. In Proceedings Eighth IEEE International Conference on Computer Vision. ICCV 2001, volume 1, pages 532–539. IEEE, 2001

    2001

  26. [34]

    Tomasi and R

    C. Tomasi and R. Manduchi. Bilateral filtering for gray and color images. In Computer Vision, 1998. Sixth International Conference on, pages 839–846. IEEE, 1998

    1998

  27. [35]

    Uhrig, N

    J. Uhrig, N. Schneider, L. Schneider, U. Franke, T. Brox, and A. Geiger. Sparsity invariant cnns. arXiv preprint arXiv:1708.06500, 2017

    2017 arXiv

  28. [36]

    J. Yen. On nonuniform sampling of bandwidth-limited signals. IRE Transactions on circuit theory, 3(4):251–257, 1956. 19

    1956

Pith tools

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