Pith. sign in

REVIEW 3 major objections 5 minor 40 references

CompenNet++: End-to-end Full Projector Compensation

T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read The paper proposes CompenNet++, the first end-to-end network that performs full projector compensation—geometric correction plus photometric compensation—jointly, learning the geometric warp directly from photometric sampling images…

desk verdict First end-to-end full projector compensation, but the headline comparison runs on a proxy metric; worth reviewing with a demand for a direct project-and-capture test. read the letter →

arxiv 1908.06246 v1 pith:TIV55S3A submitted 2019-08-17 cs.CV cs.GR

Bingyao Huang , Haibin Ling This is my paper · ORCID
classification cs.CVcs.GR
keywords projectorcompensationgeometriccorrectionphotometricend-to-endlearningconvolutionalneuralnetworkthinplatesplinestructuredlightprojectionmapping
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

This paper claims that full projector compensation—correcting where pixels land on a nonplanar textured surface and how the surface changes their colors—should be solved as one learning problem, not as two separate calibration steps. It proposes CompenNet++, a single end-to-end network that concatenates a WarpingNet for geometric correction with a CompenNet for photometric compensation. If the claim holds, projector systems can skip the structured-light scan that two-step methods require, avoid error propagation between the two stages, and run faster at test time after network simplification. The reported experiments on 20 nonplanar textured setups, including a surface behind glass with specular reflections, support the advantage over two-step baselines.

What carries the argument

The carrying object is WarpingNet, the geometric correction subnet: a cascaded coarse-to-fine sampling grid predictor. It stacks (1) a learnable $2\times3$ affine matrix that maps the projector field of view toward a front view, (2) a thin-plate-spline grid with 76 learnable parameters for a smooth nonlinear warp, and (3) a grid-refinement network, built on a U-Net-like encoder-decoder, that adds residual displacements to the grid. Instead of resampling the image three times, the finer grids sample the coarser grids in coordinate space, so the input image is warped once with a single bilinear interpolation; this keeps the result sharp and reduces computation. After training, the whole geometric stage collapses to one sampling grid, and the photometric surface-feature branch collapses to additive biases.

What would settle it

Measure the learned geometric warp against an independent structured-light ground truth on a surface with a deep concavity or strong interreflection; if the compensation in the occluded region stays plain gray or the reprojection error exceeds that of a two-step SL baseline, the joint-training claim fails. Alternatively, ablate the grid refinement network while keeping photometric compensation fixed on a specular surface; if the end-to-end advantage disappears, the mechanism is the refinement network rather than the joint formulation.

Watch

Extended reading notes

Core claim

The central claim is that a deep network can learn the composite geometric warp and photometric transfer of a projector-camera-surface system directly from photometric sampling images, without building a pixel-correspondence map by structured light. The paper models a camera capture as $\tilde{x} = \mathcal{T}(\mathcal{F}(x;g,s))$ and trains the network to invert both factors at once, $x^* = \mathcal{F}^{\dagger}(\mathcal{T}^{-1}(x);\mathcal{T}^{-1}(\tilde{s}))$. The geometric part is a cascaded coarse-to-fine warp: a learnable affine transform, a thin-plate-spline grid, and a residual grid-refinement network, all composed in 2D coordinate space so the image is resampled only once. The photometric part is an improved CompenNet initialized as a linear channel-independent model to avoid degenerate solutions such as plain-gray outputs. At test time both subnets simplify to a single sampling grid plus a bias-augmented photometric backbone, with no measured performance drop.

Load-bearing premise

The method assumes that jointly training the warp and the photometric network converges to a useful solution from the proposed initialization; if the photometric loss gives insufficient gradient signal toward the correct warp, the output degrades to a plain gray image, and the paper also assumes every surface patch is directly illuminated by the projector, so self-occluded regions fail.

Editorial extensions

If this is right

  • Projector systems no longer need a structured-light scan: the geometric warp is learned from the same photometric sampling images used for color compensation, saving the 42 extra projected patterns used in the comparisons.
  • Errors stop propagating from geometric mapping into photometric compensation; joint training can tolerate specular highlights and reflections that corrupt structured-light decoding.
  • At test time the trained system reduces to one image resampling and one pass through the photometric backbone, which is faster and uses less memory than the training-time network.
  • Even with only 48 training images and shortened training, the method can keep a structural-similarity advantage over two-step baselines trained on 500 images.
  • The joint formulation provides a template for other projector-camera tasks where geometry and appearance are entangled and traditionally handled in sequence.

Reading between the lines

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

  • The cascaded coordinate-space grid composition could be reused for projector-camera calibration or multi-projector blending, where geometric warps are currently estimated from markers or structured light.
  • The initialization recipe—starting the photometric subnet as a linear channel-independent subtractor—is a general defense against degenerate vacuous solutions in end-to-end image-to-image training, and could be tested on other compensation networks.
  • On surfaces with strong interreflections or global light transport, conditioning only on a single surface image may be insufficient; a testable extension would feed multiple surface images captured under different projector inputs.
  • A direct geometric evaluation against structured-light ground truth (for example reprojection error) would isolate how much of the end-to-end gain comes from better warping versus better photometric modeling, since the paper reports only indirect compensation quality.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper proposes CompenNet++, an end-to-end neural network for full projector compensation on nonplanar textured surfaces. It extends the authors' earlier CompenNet by adding a WarpingNet subnet that learns a cascaded coarse-to-fine geometric warp directly from photometric sampling images, concatenating it with a photometric compensation subnet, and training both jointly. A benchmark with 20 setups is introduced, and experiments compare against TPS-, Pix2pix-, and CompenNet-based two-step baselines. The paper also describes task-specific weight initialization, network simplification for test-time efficiency, and reports qualitative and quantitative results claimed to outperform prior methods. The source code and benchmark are released.

Significance. If the empirical claims are confirmed, this would be a useful step: the first end-to-end trainable full compensation pipeline, demonstrated on a public benchmark with 20 setups, with code release. The joint-training idea is principled, and the coarse-to-fine grid refinement and test-time simplification are sensible engineering contributions. The paper explicitly ships reproducible code and a new benchmark, which are strengths. However, the evaluation protocol in Section 3.8 does not directly test the production pipeline described in Section 3.6, and the reported improvements over CompenNet with structured light are small and reported without error bars. The central comparative claim therefore needs additional evidence before the paper can be accepted.

major comments (3)
  1. [Section 3.8, Eq. (9), Table 1, Fig. 5] Section 3.8 defines the reported metric as \hat{y}_i = \pi^\dagger_\theta(\tilde{y}_i; \tilde{s}) compared with y_i, where \tilde{y}_i is the camera capture obtained by projecting y_i. This evaluates the inverse map on the range of the forward imaging process, i.e., reconstruction of the projector input from its capture. The production pipeline in Section 3.6 and Fig. 5 instead feeds the desired viewer image z (after affine fit z' = A z) into the simplified network and projects the output. A desired image z is not generally a camera capture \tilde{y}_i, so it lies outside the input distribution used in Eq. (9). Table 1 therefore does not directly measure full compensation quality for arbitrary desired images; it measures inverse consistency on training-distribution captures. A direct project-and-capture experiment is required: for held-out desired images z, compute z* = \pi'^\dagger_\theta(z'), project z*, capture the result, and compare it with z, for all baselines. Without that, the abstract's claim of outperforming two-step methods is not supported by the reported numbers.
  2. [Section 3.5] Section 3.5 states that the test-time simplification is made 'without any sacrifice in performance quality,' and Section 3.6 uses the simplified \pi'^\dagger_\theta at deployment, but no experiment compares \pi'^\dagger_\theta with the un-simplified \pi^\dagger_\theta under identical training. The rows CompenNet++ fast/faster in Table 1 use different training-set sizes and iteration counts, so they are not an ablation of the simplification. Please add an ablation that trains once and evaluates both the full and simplified networks on the same data.
  3. [Table 1] Table 1 reports averages over 20 setups but no per-setup variances or significance tests. At #Train=500, CompenNet++ attains PSNR 21.11 versus 20.96 for CompenNet w/ SL and SSIM 0.727 versus 0.712; these differences are small relative to likely cross-setup variation. Without error bars, paired tests, or per-setup results, the claim of 'clear advantages' is not quantitatively established. Reporting the standard deviation over the 20 setups, or a per-setup table or scatter plot, would address this.
minor comments (5)
  1. [Section 2.2] There is a likely typo: 'context-aware ones typically assume pro-cam pixels one-to-one mapping and context-aware ones also consider neighborhood/global information' should read 'context-independent ones also consider neighborhood/global information.'
  2. [Sections 3.1 and 3.2] The symbol x is used both for a projector input image and for a desired perceived image (e.g., Eq. (3) versus Eq. (7)). Please use separate symbols or state explicitly that the same symbol is intentionally reused.
  3. [Fig. 5] Figure 5 is underspecified: it should state which tensor is the input to \pi'^\dagger_\theta, in which coordinate frame z' lives, and how the affine transform A is applied.
  4. [Section 3.4, Eq. (13)] The initialization in Eq. (13) uses a synthetic surface image \dot{s} that 'mimics' the warped surface image, but the paper does not describe how \dot{s} is generated; a sentence or reference is needed for reproducibility.
  5. [Section 4.2] The statement 'we use 42 less images than two-step SL-based method' should be qualified: at #Train=500 the total is 500 versus 542, while at #Train=48 the difference is 42, so the claim depends on the training-set size.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation found: the core claim is empirically tested against external baselines, and self-citations to CompenNet are not load-bearing.

full rationale

The paper's derivation chain is a supervised inverse-modeling construction, not a self-referential one. Equations (1)-(6) define the forward projection/capture process and formally write the compensation image as F^†(T^{-1}(x); T^{-1}(~s)); Eqs. (7)-(9) then train a network π^†_θ on sampled pairs (~x_i, x_i) by minimizing a loss between the network output and the known projector input x_i. The evaluation in Section 3.8 uses held-out pairs (~y_i, y_i) under the same setup, so the reported PSNR/RMSE/SSIM measure generalization of the learned inverse, not a quantity that equals the training target by construction. No parameter is fitted to the validation metric and then reported as a prediction. CompenNet [15] is cited extensively as the photometric backbone and as the source of the evaluation protocol, and it is by the same authors. However, the central claim of this paper—that jointly learning geometry and photometry outperforms two-step structured-light methods—is supported by comparisons with TPS [10], Pix2pix [17], and CompenNet w/ SL, all trained and evaluated under the same protocol. The reused CompenNet is a published component, not an unverified theorem imported to force the conclusion, so the self-citations are routine prior work rather than load-bearing circularity. The only notable concern is an evaluation/deployment mismatch: Section 3.8 evaluates ŷ_i = π^†_θ(~y_i; ~s) against y_i, whereas the test-time pipeline in Fig. 5 feeds a desired image z through the simplified network. This is a validity or correctness issue about whether the headline 'outperforms' is directly supported, not a circularity: the reported metric is not equal to the training loss by construction, and all compared methods are subject to the same protocol. The limitations section also explicitly identifies self-occlusion as a failure mode, which is an external, falsifiable scope condition rather than a circular defensive move.

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

The method relies on several domain assumptions about the separability of geometry and photometry and the sufficiency of a single surface image. The only fitted quantities are the network parameters and standard hyperparameters; no new physical entities are introduced.

free parameters (4)
  • Network weights θ = {θF, θT} = Learned on each setup
    These encode the geometric warp and photometric transfer; they are fitted to the training data, not derived.
  • TPS control point grid (6x6, 76 parameters) = 6x6 grid
    A hand-chosen design decision for the spline granularity in WarpingNet.
  • Loss weighting (L1 + SSIM, equal weights) = 1 and 1
    The combination in Eq. (10) is chosen without derivation; other weightings are not explored.
  • Training hyperparameters (learning rate, batch size, iterations) = 1e-3, 48, 1500
    Set in Section 3.7; these affect convergence and final performance.
assumptions (4)
  • domain assumption The image formation process (Eq. 1) can be disentangled into a geometric warp T and a photometric transfer F (Eq. 2).
    The paper states this reformulation in Section 3.1; if the interaction of geometry and photometry is not separable, the model is misspecified.
  • domain assumption The camera-captured surface image \tilde{s} under a plain gray input captures the spectral interaction needed for compensation (Eq. 5).
    This substitutes g and s with T^{-1}(\tilde{s}) in Eq. (6), a crucial simplification.
  • domain assumption The inverse mappings F† and T^{-1} are representable by the chosen CNN architectures and learnable from the sampled pairs.
    Section 3.2 assumes the deep network can approximate these inverse functions; no universal approximation argument is given.
  • domain assumption Each surface patch is directly illuminated by the projector (no self-occlusion).
    Explicitly stated as a limitation in Section 5; the failure case in Fig. 7 shows the method cannot handle occluded regions.

how reviews work

0 comments
Cite this review

Pith. "Pith review of CompenNet++: End-to-end Full Projector Compensation." pith.science (2026). https://pith.science/paper/TIV55S3A

@misc{pith2026190806246,
  author       = {Pith},
  title        = {Pith review of: CompenNet++: End-to-end Full Projector Compensation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TIV55S3A}},
  note         = {Machine review of arXiv:1908.06246}
}
read the original abstract

Full projector compensation aims to modify a projector input image such that it can compensate for both geometric and photometric disturbance of the projection surface. Traditional methods usually solve the two parts separately, although they are known to correlate with each other. In this paper, we propose the first end-to-end solution, named CompenNet++, to solve the two problems jointly. Our work non-trivially extends CompenNet, which was recently proposed for photometric compensation with promising performance. First, we propose a novel geometric correction subnet, which is designed with a cascaded coarse-to-fine structure to learn the sampling grid directly from photometric sampling images. Second, by concatenating the geometric correction subset with CompenNet, CompenNet++ accomplishes full projector compensation and is end-to-end trainable. Third, after training, we significantly simplify both geometric and photometric compensation parts, and hence largely improves the running time efficiency. Moreover, we construct the first setup-independent full compensation benchmark to facilitate the study on this topic. In our thorough experiments, our method shows clear advantages over previous arts with promising compensation quality and meanwhile being practically convenient.

Figures

Figures reproduced from arXiv: 1908.06246 by the authors.

Figure 1
Figure 1. Full projector geometric correction and photo [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Training of CompenNet++ in two major steps. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. WarpingNet (T −1 ) architecture (activations layers [21, 23] omitted). It warps the input camera-captured image x˜ to the projector’s view using a cascaded coarse-to-fine structure. The red and green blocks are learnable parameters and grid generation functions, respectively. Operator ⊗ denotes bilinear interpolation, i.e., φ(·; ·). The grid refinement network Wθr consists of a UNet-like [32] structure, it generates… view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Projector FOV mask, bounding rectangle (green) [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: The testing phase of the proposed CompenNet++. Due to our novel WarpingNet structure and sampling strategy, [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Qualitative comparison of TPS [10] w/ SL, TPS textured w/ SL, Pix2pix [17] w/ SL, CompenNet [15] w/ SL, proposed CompenNet++ w/o refine and proposed CompenNet++ on two different surfaces. The 1st to 3rd columns are the camera-captured projection surface, desired viewer…
Figure 7
Figure 7. Figure 7: Failed example. CompenNet++ is unable to com [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

40 extracted references · 37 canonical work pages

  1. [1]

    Fast high-resolution appear- ance editing using superimposed projections

    Daniel G Aliaga, Yu Hong Yeung, Alvin Law, Behzad Sa- jadi, and Aditi Majumder. Fast high-resolution appear- ance editing using superimposed projections. ACM Tran. on Graphics, 2012. 1, 2

    work page 2012

  2. [2]

    Fabri- cating diminishable visual markers for geometric registration in projection mapping

    Hirotaka Asayama, Daisuke Iwai, and Kosuke Sato. Fabri- cating diminishable visual markers for geometric registration in projection mapping. IEEE TVCG, 2018. 1, 2

    work page 2018

  3. [3]

    Robust content-dependent photometric projector com- pensation

    Mark Ashdown, Takahiro Okabe, Imari Sato, and Yoichi Sato. Robust content-dependent photometric projector com- pensation. In CVPRW PROCAMS, 2006. 1, 2

    work page 2006

  4. [4]

    Embedded entertainment with smart projectors

    Oliver Bimber, Andreas Emmerling, and Thomas Klemmer. Embedded entertainment with smart projectors. Computer,

    work page

  5. [5]

    Saliency-guided pro- jection geometric correction using a projector-camera sys- tem

    Ameneh Boroomand, Hicham Sekkati, Mark Lamm, David A Clausi, and Alexander Wong. Saliency-guided pro- jection geometric correction using a projector-camera sys- tem. In ICIP, 2016. 1, 2

    work page 2016

  6. [6]

    Approximate thin plate spline mappings

    Gianluca Donato and Serge Belongie. Approximate thin plate spline mappings. In ECCV, 2002. 4, 6

    work page 2002

  7. [7]

    A projector-camera system with real-time photometric adapta- tion for dynamic environments

    Kensaku Fujii, Michael D Grossberg, and Shree K Nayar. A projector-camera system with real-time photometric adapta- tion for dynamic environments. In CVPR, 2005. 2

    work page 2005

  8. [8]

    Structured-light 3D surface imaging: a tutorial

    Jason Geng. Structured-light 3D surface imaging: a tutorial. Advances in Optics and Photonics, 2011. 1

    work page 2011

Show all 40 references
  1. [9]

    Making one object look like another: controlling appearance using a projector-camera system

    Michael D Grossberg, Harish Peri, Shree K Nayar, and Pe- ter N Belhumeur. Making one object look like another: controlling appearance using a projector-camera system. In CVPR, 2004. 1, 2

    2004

  2. [10]

    Robust, error-tolerant photometric projector compensation

    Anselm Grundh ¨ofer and Daisuke Iwai. Robust, error-tolerant photometric projector compensation. IEEE TIP, 2015. 1, 2, 6, 7

    2015

  3. [11]

    Recent advances in projection mapping algorithms, hardware and applications

    Anselm Grundh ¨ofer and Daisuke Iwai. Recent advances in projection mapping algorithms, hardware and applications. In Computer Graphics Forum. Wiley Online Library, 2018. 1, 3

    2018

  4. [12]

    Practical methods for geometric and photometric correction of tiled projector

    Michael Harville, Bruce Culbertson, Irwin Sobel, Dan Gelb, Andrew Fitzhugh, and Donald Tanguay. Practical methods for geometric and photometric correction of tiled projector. In CVPRW, 2006. 1, 2

    2006

  5. [13]

    Delving deep into rectifiers: Surpassing human-level perfor- mance on imagenet classification

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Delving deep into rectifiers: Surpassing human-level perfor- mance on imagenet classification. In ICCV, 2015. 5

    2015

  6. [14]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In CVPR,

  7. [15]

    End-to-end projector pho- tometric compensation

    Bingyao Huang and Haibin Ling. End-to-end projector pho- tometric compensation. In CVPR, 2019. 1, 2, 3, 4, 5, 6, 7, 8

    2019

  8. [16]

    A single-shot-per-pose camera- projector calibration system for imperfect planar targets

    Bingyao Huang, Samed Ozdemir, Ying Tang, Chunyuan Liao, and Haibin Ling. A single-shot-per-pose camera- projector calibration system for imperfect planar targets. In 2018 IEEE International Symposium on Mixed and Aug- mented Reality Adjunct (ISMAR-Adjunct) , pages 15–20. IEEE, 2018. 1

    2018

  9. [17]

    Image-to-image translation with conditional adver- sarial networks

    Phillip Isola, Jun-Yan Zhu, Tinghui Zhou, and Alexei A Efros. Image-to-image translation with conditional adver- sarial networks. CVPR, 2017. 6, 7, 8

    2017

  10. [18]

    Spatial transformer networks

    Max Jaderberg, Karen Simonyan, Andrew Zisserman, and Koray Kavukcuoglu. Spatial transformer networks. In NIPS,

  11. [19]

    Kingma and Jimmy Ba

    Diederik P. Kingma and Jimmy Ba. Adam: A method for stochastic optimization. In ICLR, 2015. 6

    2015

  12. [20]

    Practical radiometric compensation for projection display on textured surfaces using a multidi- mensional model

    Yuqi Li, Aditi Majumder, Meenakshisundaram Gopi, Chong Wang, and Jieyu Zhao. Practical radiometric compensation for projection display on textured surfaces using a multidi- mensional model. In Computer Graphics Forum. Wiley On- line Library, 2018. 2

    2018

  13. [21]

    Recti- fier nonlinearities improve neural network acoustic models

    Andrew L Maas, Awni Y Hannun, and Andrew Y Ng. Recti- fier nonlinearities improve neural network acoustic models. In ICML, 2013. 5

    2013

  14. [22]

    Simple, accurate, and robust projector-camera calibration

    Daniel Moreno and Gabriel Taubin. Simple, accurate, and robust projector-camera calibration. In 3DIMPVT, 2012. 1, 6

    2012

  15. [23]

    Rectified linear units improve restricted boltzmann machines

    Vinod Nair and Geoffrey E Hinton. Rectified linear units improve restricted boltzmann machines. In ICML, 2010. 5

    2010

  16. [24]

    Dynamic projection mapping onto deforming non-rigid sur- face using deformable dot cluster marker.IEEE TVCG, 2017

    Gaku Narita, Yoshihiro Watanabe, and Masatoshi Ishikawa. Dynamic projection mapping onto deforming non-rigid sur- face using deformable dot cluster marker.IEEE TVCG, 2017. 1, 2

    2017

  17. [25]

    A projection system with radiometric com- pensation for screen imperfections

    Shree K Nayar, Harish Peri, Michael D Grossberg, and Pe- ter N Belhumeur. A projection system with radiometric com- pensation for screen imperfections. In ICCVW PROCAMS, volume 3, 2003. 2, 5

    2003

  18. [26]

    A threshold selection method from gray- level histograms

    Nobuyuki Otsu. A threshold selection method from gray- level histograms. IEEE T-SMC, 1979. 5

    1979

  19. [27]

    Automatic differentiation in pytorch

    Adam Paszke, Sam Gross, Soumith Chintala, Gregory Chanan, Edward Yang, Zachary DeVito, Zeming Lin, Al- ban Desmaison, Luca Antiga, and Adam Lerer. Automatic differentiation in pytorch. In NIPS-W, 2017. 6

    2017

  20. [28]

    A self-correcting pro- jector

    Ramesh Raskar and Paul Beardsley. A self-correcting pro- jector. In CVPR, 2001. 1, 2

    2001

  21. [29]

    ilamps: geo- metrically aware and self-configuring projectors

    Ramesh Raskar, Jeroen Van Baar, Paul Beardsley, Thomas Willwacher, Srinivas Rao, and Clifton Forlines. ilamps: geo- metrically aware and self-configuring projectors. ACM Tran. on Graphics, 2003. 1, 2, 5, 6

    2003

  22. [30]

    Shader lamps: Animating real objects with image-based illumination

    Ramesh Raskar, Greg Welch, Kok-Lim Low, and Deepak Bandyopadhyay. Shader lamps: Animating real objects with image-based illumination. In Rendering Techniques. Springer, 2001. 1, 2

    2001

  23. [31]

    Convo- lutional neural network architecture for geometric matching

    Ignacio Rocco, Relja Arandjelovic, and Josef Sivic. Convo- lutional neural network architecture for geometric matching. In CVPR, 2017. 4

    2017

  24. [32]

    U-net: Convolutional networks for biomedical image segmentation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U-net: Convolutional networks for biomedical image segmentation. In MICCAI. Springer, 2015. 5

    2015

  25. [33]

    Adict: accurate direct and inverse color transformation

    Behzad Sajadi, Maxim Lazarov, and Aditi Majumder. Adict: accurate direct and inverse color transformation. In ECCV. Springer, 2010. 2

    2010

  26. [34]

    Simultaneous geometric and ra- diometric calibration of a projector-camera pair

    Marjan Shahpaski, Luis Ricardo Sapaico, Gaspard Chevas- sus, and Sabine Susstrunk. Simultaneous geometric and ra- diometric calibration of a projector-camera pair. In CVPR,

  27. [35]

    Adaptive stray-light compensation in dynamic multi-projection mapping

    Christian Siegl, Matteo Colaianni, Marc Stamminger, and Frank Bauer. Adaptive stray-light compensation in dynamic multi-projection mapping. Computational Visual Media ,

  28. [36]

    Real-time pixel luminance optimization for dynamic multi-projection mapping

    Christian Siegl, Matteo Colaianni, Lucas Thies, Justus Thies, Michael Zollh ¨ofer, Shahram Izadi, Marc Stamminger, and Frank Bauer. Real-time pixel luminance optimization for dynamic multi-projection mapping. ACM Tran. on Graph- ics, 2015. 1, 2

    2015

  29. [37]

    Inter- reflection compensation of immersive projection display by spatio-temporal screen reflectance modulation

    Shoichi Takeda, Daisuke Iwai, and Kosuke Sato. Inter- reflection compensation of immersive projection display by spatio-temporal screen reflectance modulation. IEEE TVCG,

  30. [38]

    Multi-projectors for arbitrary surfaces without explicit cal- ibration nor reconstruction

    Jean-Philippe Tardif, S ´ebastien Roy, and Martin Trudeau. Multi-projectors for arbitrary surfaces without explicit cal- ibration nor reconstruction. In 3DIM, 2003. 1, 2

    2003

  31. [39]

    A vir- tual color reconstruction system for real heritage with light projection

    Takenobu Yoshida, Chinatsu Horii, and Kosuke Sato. A vir- tual color reconstruction system for real heritage with light projection. In Proceedings of International Conference on Virtual Systems and Multimedia, volume 3, 2003. 1, 2

    2003

  32. [40]

    Loss functions for image restoration with neural networks

    Hang Zhao, Orazio Gallo, Iuri Frosio, and Jan Kautz. Loss functions for image restoration with neural networks. IEEE TCI, 2017. 4

    2017

Pith tools

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