Pith. sign in

REVIEW 2 major objections 5 minor 53 references

UprightNet: Geometry-Aware Camera Orientation Estimation from Single Images

T0 review · 2 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 A network that predicts two surface geometries at every pixel estimates camera roll and pitch more accurately than direct regression, by solving for the up vector that best aligns them.

desk verdict Solid geometry-aware calibration paper; the dual surface-frame representation and differentiable weighted solver are the real contributions, and the stress-test concern about global axes dissolves once you see Eq. (2). read the letter →

arxiv 1908.07070 v1 pith:4SQKEOWF submitted 2019-08-19 cs.CV

Wenqi Xian , Zhengqi Li , Matthew Fisher , Jonathan Eisenmann , Eli Shechtman , Noah Snavely This is my paper · ORCID
classification cs.CV
keywords cameraorientationestimationsurfaceframesconstrainedleastsquaressingle-imagecalibrationrollandpitchindoorscenegeometrygeometricdeeplearningupvector
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 proposes that camera roll and pitch can be estimated more accurately from a single indoor image when the network is made to predict per-pixel surface geometry in two coordinate systems and solve for the camera's up vector as the rotation that aligns them, rather than regressing angles directly. At each pixel it predicts an orthonormal surface frame (surface normal plus two tangent directions) in local camera coordinates, and a matching scene-layout vector in a global upright coordinate system. The alignment is a weighted constrained least-squares problem that is solved analytically and differentiably, so the whole model can be trained end-to-end from ground-truth orientations while also being supervised on the intermediate geometry. The result is an average up-vector angular error of 1.17 degrees on the InteriorNet test set, against 1.76 degrees for the strongest prior CNN-based calibration method, with the advantage persisting on real ScanNet frames and on a cross-dataset SUN360 test. Learned per-pixel weights, which concentrate on line junctions and plane boundaries, are shown to be necessary for the improvement.

What carries the argument

The load-bearing object is the per-pixel surface frame $\mathbf{F}(i)=[\mathbf{n}(i)\ \mathbf{t}(i)\ \mathbf{b}(i)]$, an orthonormal basis composed of the surface normal and two tangent vectors. The network predicts these frames in camera coordinates and, in global upright coordinates, only the third row $\mathbf{f}^g_z$ of the global frame, a scene-layout vector whose normal component sets floor, wall, and ceiling pixels to fixed values. The identity $\mathbf{f}^g_z(i)=\mathbf{u}^T\mathbf{F}_c(i)$ turns the unknown up vector into the solution of a constrained least-squares problem, and the solver reduces it to a quadratic eigenvalue problem $I\lambda^2-2H\lambda+H^2-\mathbf{g}\mathbf{g}^T=0$ with $H=A^TA$ and $\mathbf{g}=A^T\mathbf{b}$. Because the needed eigenvalue derivative is closed-form, the orientation loss propagates through the solver to the network, while separately predicted per-pixel weights make the fit robust to unreliable regions by concentrating on line junctions and structure edges.

What would settle it

Build a test set of indoor images with little or no visible floor, ceiling, or wall plane (close-up furniture views, cluttered corners), run UprightNet and the direct-regression baseline on it, and check whether the up-vector error gap shrinks or reverses; the alignment mechanism predicts the method should lose its advantage exactly where planar supporting structure is absent.

Watch

Extended reading notes

Core claim

The central claim is that the information needed for single-image roll and pitch is already present in local surface geometry, but only becomes usable when the same geometry is also predicted in a global upright coordinate system. Writing the per-pixel surface frame as $\mathbf{F}_c(i)$ and the third row of the global upright frame as $\mathbf{f}^g_z(i)$, the scene up vector $\mathbf{u}$ must satisfy $\mathbf{f}^g_z(i)=\mathbf{u}^T\mathbf{F}_c(i)$ at every pixel, so orientation estimation reduces to a weighted constrained least-squares problem in $\mathbf{u}$ alone. The paper shows this problem has an analytic solution through a quadratic eigenvalue problem, and that the eigenvalue derivative is available in closed form, making the entire alignment differentiable and trainable by back-propagating the orientation error. With this formulation, the average up-vector error drops from 1.76 degrees to 1.17 degrees on InteriorNet relative to the best prior CNN-based method, with consistent gains on ScanNet and a large margin on the SUN360 cross-dataset test.

Load-bearing premise

The method assumes a single indoor image contains enough visible flat, structured surfaces—floors, ceilings, walls, and their junctions—to give a clear alignment signal between the two predicted geometries; the paper's own limitations section says images without such supporting structure lead to inaccurate predictions.

Editorial extensions

If this is right

  • Single-image roll and pitch can be framed as an alignment of two predicted geometry maps, and this framing is trainable end-to-end through a differentiable constrained least-squares solver.
  • Supervising the intermediate surface frames in both coordinate systems adds accuracy on top of the orientation loss, so the method can exploit partial geometric ground truth when full camera poses are available.
  • The learned weight maps form an interpretable by-product: they highlight line junctions and plane boundaries, the same regions that carry vanishing-point information in classical calibration.
  • Cross-dataset results on SUN360 indicate the geometric intermediate representation transfers to new indoor scene distributions better than direct regression or horizon classification.

Reading between the lines

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

  • The differentiable two-frame alignment is not tied to indoor scenes: replacing the global upright frame with a sky-versus-ground semantic prior could give outdoor roll and pitch estimation the same structure.
  • Because the orientation loss alone already improves accuracy, a self-supervised variant could be built that aligns predicted frames across generated rotations, reducing the need for dense ground-truth normals.
  • A testable consequence of the paper's mechanism is that accuracy should depend on how much of the image is covered by planar supporting structures; partitioning a test set by that coverage would draw the boundary where the method loses its advantage over direct regression.
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

2 major / 5 minor

Summary. The paper introduces UprightNet, an end-to-end deep network for estimating 2DoF camera orientation (roll and pitch) from a single indoor RGB image. The central idea is to predict per-pixel surface frames in both local camera coordinates (F_c) and a global upright coordinate system (f_z^g), then solve for the camera up vector as the unit vector that best aligns these two predictions via a differentiable weighted constrained least-squares module. Training is end-to-end with a combination of an orientation loss, surface-frame losses, and gradient-consistency losses. Experiments on InteriorNet, ScanNet, and a SUN360 cross-dataset test show consistent improvements over regression, classification, and classical baselines, and ablations support the design choices of the predicted weights and the full surface-frame representation.

Significance. If the formulation is sound, this is a valuable contribution to single-image camera calibration. The paper convincingly demonstrates that injecting explicit geometric reasoning through dual surface-frame prediction and a differentiable constrained least-squares solver can improve both accuracy and cross-dataset generalization relative to black-box regression or classification. The differentiable solver itself is a reusable module, and the extensive experimental comparison and ablation study strengthen the empirical case. The main concern is a gap in the definition of the global upright coordinate system, which affects the mathematical foundation and reproducibility of the method.

major comments (2)
  1. [Section 3, Eqs. (1)-(2) and Eq. (13)] The global upright coordinate system is not fully specified: the paper states only that its up vector aligns with the scene up vector, leaving the global x/y axes undefined. Because the tangent vector t is defined as n × (camera y-axis), the analogous definition in the global frame depends on the arbitrary global y-axis. For a vertical wall, t_g^z is the sine of the angle between the wall normal and the global y-axis, so its sign and magnitude change under a scene-level yaw rotation. Consequently, the ground-truth f_z^g used in the surface-frame loss L_F (Eq. 13) is not well-defined as a function of the image, and Eq. (1) F_g = R F_c holds only if R maps the camera y-axis to the global y-axis, which is generally false for a camera with nonzero yaw. The authors should specify a canonical global yaw convention (e.g., projecting the camera's right vector onto the horizontal plane) or redefine the tangent vectors in a yaw-invariant way, then re-derive the alignment equation and re-run the experiments under that convention.
  2. [Section 4.3, Tables 1-5] The paper repeatedly describes improvements as 'significant,' but no statistical significance tests, confidence intervals, or variance across multiple training runs are reported. On ScanNet, the gain over Hold-Geoffroy et al. is about 0.8 degrees in mean angular error (3.74 vs. 4.55), which may not be robust given that the baselines were re-implemented by the authors and that only a single split is used. Please report error bars, significance tests, or at least a discussion of variance to support the claim of significant improvement.
minor comments (5)
  1. [Section 3.1, Eqs. (7)-(8)] The paper states that the derivative of the eigenvalue is computed in closed form using [45], but the actual derivative formula and its backpropagation details are not presented. A short derivation or a pointer to the supplementary material would improve reproducibility.
  2. [Table 3] In the row 'Ours (w/o Lo)', the median pitch error (2.83°) is reported as larger than the mean pitch error (1.21°), which is atypical for an angular error distribution; please check this entry for a possible typo.
  3. [Section 4.1] The description of InteriorNet says 'Each scene includes 3 images randomly sampled from a rendered videos.' This should read 'a rendered video' or 'rendered videos.'
  4. [Eq. (9)] The nested-norm notation in Eq. (9) is confusing; it should be a single L2 norm of the vector W(i)(u^T F_c(i) - f_z^g(i)).
  5. [Figure 4 caption] The caption says the weight maps are overlaid on the input images, but the overlay alpha or blending mode is not specified; adding this detail would improve clarity.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the up vector is recovered from independently supervised geometry predictions via an external least-squares solver.

full rationale

UprightNet's derivation chain is self-contained. The estimated camera up vector u is the solution of a constrained weighted least-squares problem (Eq. 3, Eq. 9) that aligns two network outputs: local camera surface frames Fc and global upright scene-layout vectors f_g_z. The solver is an external mathematical result (Gander, Golub, and von Matt, Eq. 7-8), not a learned or fitted quantity. The network outputs are supervised directly by ground-truth surface geometry and ground-truth camera poses through the losses in Eq. 11-14, and the orientation loss backpropagates through the differentiable solver. No predicted quantity is also used as a fitted input, and no parameter is renamed as a prediction. The only same-author citations are [19], used as a comparison baseline, and [44], cited only as inspiration for end-to-end intermediate representation learning; neither supplies a load-bearing premise or a uniqueness theorem. The noted ambiguity in the global upright frame's x/y axes is a label-consistency and reproducibility concern, not circularity, because the supervision labels are still computed from ground-truth orientation rather than from the model's own outputs. Therefore the central claim does not reduce to its inputs by construction.

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

The central derivation relies on a published theorem for constrained eigenvalue problems, on the domain assumption of structured indoor planar surfaces, and on several hand-chosen hyperparameters. No physically invented entities are introduced; the 'surface frame' and 'scene layout vector' are representations, not new physical objects.

free parameters (4)
  • Loss weight alpha_L for surface frame loss LF in Eq. 11 = not specified (deferred to supplementary)
    Hyperparameter chosen by hand to balance the orientation and surface geometry losses; not derived from first principles.
  • Loss weight alpha_gradient for gradient consistency loss in Eq. 11 = not specified (deferred to supplementary)
    Hyperparameter chosen by hand to balance the gradient consistency loss.
  • Orientation loss switch threshold epsilon = 1e-6
    Chosen to avoid exploding gradients of arccos near 1; hand-set.
  • Number of scales S for gradient consistency loss in Eq. 14 = 4
    Chosen by hand for the multi-scale gradient loss.
assumptions (4)
  • standard math Constrained least squares solution via Gander-Golub-von Matt theorem (Theorem 1 from reference [14])
    Equations 7 and 8 rely on a published theorem for the constrained eigenvalue problem; proof is deferred to the supplementary material.
  • domain assumption Indoor scenes are predominantly composed of structured planar surfaces (floors, ceilings, walls)
    The method assumes such surfaces provide reliable normals and tangents for alignment; the paper's limitation section notes that images lacking these structures lead to failures.
  • domain assumption Non-degeneracy of the tangent frame definition
    The tangent vector t is defined via cross product with the camera y-axis; the paper notes this degenerates when the normal is parallel to the y-axis but states it is rare in practice.
  • domain assumption Global upright coordinate system is consistent across images
    The supervised learning of the global upright geometry f^g_z assumes that floors, ceilings, and walls have consistent layout vectors across different images.

how reviews work

0 comments
Cite this review

Pith. "Pith review of UprightNet: Geometry-Aware Camera Orientation Estimation from Single Images." pith.science (2026). https://pith.science/paper/4SQKEOWF

@misc{pith2026190807070,
  author       = {Pith},
  title        = {Pith review of: UprightNet: Geometry-Aware Camera Orientation Estimation from Single Images},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4SQKEOWF}},
  note         = {Machine review of arXiv:1908.07070}
}
read the original abstract

We introduce UprightNet, a learning-based approach for estimating 2DoF camera orientation from a single RGB image of an indoor scene. Unlike recent methods that leverage deep learning to perform black-box regression from image to orientation parameters, we propose an end-to-end framework that incorporates explicit geometric reasoning. In particular, we design a network that predicts two representations of scene geometry, in both the local camera and global reference coordinate systems, and solves for the camera orientation as the rotation that best aligns these two predictions via a differentiable least squares module. This network can be trained end-to-end, and can be supervised with both ground truth camera poses and intermediate representations of surface geometry. We evaluate UprightNet on the single-image camera orientation task on synthetic and real datasets, and show significant improvements over prior state-of-the-art approaches.

Figures

Figures reproduced from arXiv: 1908.07070 by the authors.

Figure 1
Figure 1. UprightNet overview. UprightNet takes a single RGB image and predicts surface geometry in both local camera and global upright coordinate systems. The camera orientation is then computed as the alignment between these two predictions, solved for by a differentiable least squares module, and weighted using predicted weight maps. generalization to a broader class of images, because geom￾etry affords generally applicab… view at source ↗
Figure 2
Figure 2. Visualization of surface geometry. From left to right: (a) image, (b-d) local camera surface frames F c , (e) the third row of global upright surface frames F g . • F c : the surface geometry in local camera coordinates. • F g : the surface geometry in global upright coordinates. Surface frames. To represent surface geometry, we define a surface frame F(i) at every pixel location i as a 3 × 3 matrix formed by three … view at source ↗
Figure 3
Figure 3. Qualitative comparison of horizon line predic￾tions. From top row to bottom row: InteriorNet, ScanNet and SUN360. Our trained model outperforms other baselines in terms of accuracy on all three datasets. every 5 epochs. More details on hyperparameter settings are included in the supplemental material. 4.3. Comparisons to baseline methods We compare UprightNet with four baseline methods: • A regression baseline: a CN… view at source ↗
Figures from the paper (4 more)
Figure 5
Figure 5. Figure 5: visualizes these alignment scores, and compares them with the predicted weights. The network indeed tends to predict large weights where the alignment error is small. This suggests that, while our surface geometry predictions are not always accurate, the weights can ca…
Figure 4
Figure 4. Figure 4: Visualizations of predictions in InteriorNet (top 3 rows) and ScanNet (bottom 3 rows). In (d), we overlay weight maps (combined weights for n, t and b) over input images. Blue=small weights, red=large weights. and similarly for the two tangent vectors [PITH_FULL_IMAGE…
Figure 7
Figure 7. Figure 7: Application. Virtual insertion of 3D objects in images from SUN360 dataset using the camera orientation estimated by UprightNet. supporting structure for the network to reason about geom￾etry, resulting in inaccurate surface geometry and camera orientation predictions.…
Figure 6
Figure 6. Figure 6: Visualizations of predictions on the SUN360 testset. The last row shows a failure case. predictions of our model in [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

53 extracted references · 51 canonical work pages

  1. [1]

    http://pytorch.org

    Pytorch, 2016. http://pytorch.org

    work page 2016

  2. [2]

    A machine learning approach to horizon line detection using local features

    Touqeer Ahmad, George Bebis, Emma E Regentova, and Ara Nefian. A machine learning approach to horizon line detection using local features. In International Symposium on Visual Computing, pages 181–193. Springer, 2013

    work page 2013

  3. [3]

    A global approach for the detection of vanishing points and mutually orthogonal vanishing directions

    Michel Antunes and Joao P Barreto. A global approach for the detection of vanishing points and mutually orthogonal vanishing directions. In Proc. Computer Vision and Pattern Recognition (CVPR), pages 1336–1343, 2013

    work page 2013

  4. [4]

    Marr revisited: 2d-3d alignment via surface normal prediction

    Aayush Bansal, Bryan Russell, and Abhinav Gupta. Marr revisited: 2d-3d alignment via surface normal prediction. In Proc. Computer Vision and Pattern Recognition (CVPR), pages 5965–5974, 2016

    work page 2016

  5. [5]

    3-line ransac for orthogonal vanishing point detection

    Jean-Charles Bazin and Marc Pollefeys. 3-line ransac for orthogonal vanishing point detection. In IEEE/RSJ Interna- tional Conference on Intelligent Robots and Systems, pages 4282–4287. IEEE, 2012

    work page 2012

  6. [6]

    Globally optimal line clustering and vanishing point estimation in manhattan world

    Jean-Charles Bazin, Yongduek Seo, C´edric Demonceaux, Pas- cal Vasseur, Katsushi Ikeuchi, Inso Kweon, and Marc Polle- feys. Globally optimal line clustering and vanishing point estimation in manhattan world. In Proc. Computer Vision and Pattern Recognition (CVPR), pages 638–645. IEEE, 2012

    work page 2012

  7. [7]

    Dsac-differentiable ransac for camera localization

    Eric Brachmann, Alexander Krull, Sebastian Nowozin, Jamie Shotton, Frank Michel, Stefan Gumhold, and Carsten Rother. Dsac-differentiable ransac for camera localization. In Proc. Computer Vision and Pattern Recognition (CVPR) , pages 6684–6692, 2017

    work page 2017

  8. [8]

    Learning less is more- 6d camera localization via 3d surface regression

    Eric Brachmann and Carsten Rother. Learning less is more- 6d camera localization via 3d surface regression. In Proc. Computer Vision and Pattern Recognition (CVPR) , pages 4654–4662, 2018

    work page 2018

Show all 53 references
  1. [9]

    Camera cali- bration with two arbitrary coplanar circles

    Qian Chen, Haiyuan Wu, and Toshikazu Wada. Camera cali- bration with two arbitrary coplanar circles. In Proc. European Conf. on Computer Vision (ECCV), pages 521–532. Springer, 2004

    2004

  2. [10]

    Scannet: Richly- annotated 3d reconstructions of indoor scenes

    Angela Dai, Angel X Chang, Manolis Savva, Maciej Halber, Thomas Funkhouser, and Matthias Nießner. Scannet: Richly- annotated 3d reconstructions of indoor scenes. In Proc. Com- puter Vision and Pattern Recognition (CVPR), pages 5828– 5839, 2017

    2017

  3. [11]

    Automatic camera calibration from a single manhattan image

    Jonathan Deutscher, Michael Isard, and John MacCormick. Automatic camera calibration from a single manhattan image. In Proc. European Conf. on Computer Vision (ECCV), pages 175–188. Springer, 2002

    2002

  4. [12]

    Predicting depth, surface nor- mals and semantic labels with a common multi-scale convo- lutional architecture

    David Eigen and Rob Fergus. Predicting depth, surface nor- mals and semantic labels with a common multi-scale convo- lutional architecture. In Proc. Int. Conf. on Computer Vision (ICCV), pages 2650–2658, 2015

    2015

  5. [13]

    Horizon detection using machine learning techniques

    Sergiy Fefilatyev, V olha Smarodzinava, Lawrence O Hall, and Dmitry B Goldgof. Horizon detection using machine learning techniques. In International Conference on Machine Learning and Applications, pages 17–21. IEEE, 2006

    2006

  6. [14]

    A con- strained eigenvalue problem

    Walter Gander, Gene H Golub, and Urs von Matt. A con- strained eigenvalue problem. Linear Algebra and its applica- tions, 114:815–839, 1989

    1989

  7. [15]

    Robust manhattan frame estimation from a single rgb-d image

    Bernard Ghanem, Ali Thabet, Juan Carlos Niebles, and Fabian Caba Heilbron. Robust manhattan frame estimation from a single rgb-d image. In Proc. Computer Vision and Pattern Recognition (CVPR), pages 3772–3780, 2015

    2015

  8. [16]

    Unsupervised monocular depth estimation with left-right con- sistency

    Cl´ement Godard, Oisin Mac Aodha, and Gabriel J Brostow. Unsupervised monocular depth estimation with left-right con- sistency. In Proc. Computer Vision and Pattern Recognition (CVPR), pages 270–279, 2017

    2017

  9. [17]

    Multiple view geom- etry in computer vision

    Richard Hartley and Andrew Zisserman. Multiple view geom- etry in computer vision. Cambridge university press, 2003

    2003

  10. [18]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In Proc. Com- puter Vision and Pattern Recognition (CVPR), pages 770–778, 2016

    2016

  11. [19]

    A perceptual measure for deep single image camera calibration

    Yannick Hold-Geoffroy, Kalyan Sunkavalli, Jonathan Eisen- mann, Matthew Fisher, Emiliano Gambaretto, Sunil Hadap, and Jean-Franc ¸ois Lalonde. A perceptual measure for deep single image camera calibration. In Proc. Computer Vision and Pattern Recognition (CVPR), pages 2354–2363, 2018

    2018

  12. [20]

    Texturenet: Consis- tent local parametrizations for learning from high-resolution signals on meshes

    Jingwei Huang, Haotian Zhang, Li Yi, Thomas Funkhouser, Matthias Nießner, and Leonidas J Guibas. Texturenet: Consis- tent local parametrizations for learning from high-resolution signals on meshes. In Proc. Computer Vision and Pattern Recognition (CVPR), pages 4440–4449, 2019

    2019

  13. [21]

    Framenet: Learning local canonical frames of 3d surfaces from a single rgb image

    Jingwei Huang, Yichao Zhou, Thomas Funkhouser, and Leonidas Guibas. Framenet: Learning local canonical frames of 3d surfaces from a single rgb image. arXiv preprint arXiv:1903.12305, 2019

    1903 arXiv

  14. [22]

    Globally optimal manhattan frame estimation in real-time

    Kyungdon Joo, Tae-Hyun Oh, Junsik Kim, and In So Kweon. Globally optimal manhattan frame estimation in real-time. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 1763–1771, 2016

    2016

  15. [23]

    Posenet: A convolutional network for real-time 6-dof camera relocal- ization

    Alex Kendall, Matthew Grimes, and Roberto Cipolla. Posenet: A convolutional network for real-time 6-dof camera relocal- ization. In Proc. Int. Conf. on Computer Vision (ICCV), pages 2938–2946, 2015

    2015

  16. [24]

    Indoor rgb-d compass from a single line and plane

    Pyojin Kim, Brian Coltin, and H Jin Kim. Indoor rgb-d compass from a single line and plane. In CVPR, 2018

    2018

  17. [25]

    Adam: A method for stochastic optimization

    Diederik P Kingma and Jimmy Ba. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980 , 2014

    2014 arXiv

  18. [26]

    Automatic upright adjustment of photographs with robust camera calibration

    Hyunjoon Lee, Eli Shechtman, Jue Wang, and Seungyong Lee. Automatic upright adjustment of photographs with robust camera calibration. Trans. Pattern Analysis and Machine Intelligence, 36(5):833–844, 2014

    2014

  19. [27]

    Finding vanishing points via point alignments in image primal and dual domains

    Jos´e Lezama, Rafael Grompone von Gioi, Gregory Randall, and Jean-Michel Morel. Finding vanishing points via point alignments in image primal and dual domains. In Proc. Com- puter Vision and Pattern Recognition (CVPR), pages 509–515, 2014

    2014

  20. [28]

    Interiornet: Mega-scale multi-sensor photo-realistic indoor scenes dataset

    Wenbin Li, Sajad Saeedi, John McCormac, Ronald Clark, Dimos Tzoumanikas, Qing Ye, Yuzhong Huang, Rui Tang, and Stefan Leutenegger. Interiornet: Mega-scale multi-sensor photo-realistic indoor scenes dataset. InProc. British Machine Vision Conf. (BMVC), 2018

    2018

  21. [29]

    Huttenlocher

    Yunpeng Li, Noah Snavely, and Daniel P. Huttenlocher. Loca- tion recognition using prioritized feature matching. In Proc. European Conf. on Computer Vision (ECCV), 2010

    2010

  22. [30]

    Learning the depths of moving people by watching frozen people

    Zhengqi Li, Tali Dekel, Forrester Cole, Richard Tucker, Noah Snavely, Ce Liu, and William T Freeman. Learning the depths of moving people by watching frozen people. In Proc. Com- puter Vision and Pattern Recognition (CVPR), pages 4521– 4530, 2019

    2019

  23. [31]

    CGIntrinsics: Better intrinsic image decomposition through physically-based rendering

    Zhengqi Li and Noah Snavely. CGIntrinsics: Better intrinsic image decomposition through physically-based rendering. In Proc. European Conf. on Computer Vision (ECCV) , pages 371–387, 2018

    2018

  24. [32]

    Megadepth: Learning single- view depth prediction from internet photos

    Zhengqi Li and Noah Snavely. Megadepth: Learning single- view depth prediction from internet photos. In Proc. Com- puter Vision and Pattern Recognition (CVPR), pages 2041– 2050, 2018

    2018

  25. [33]

    Planenet: Piece-wise planar reconstruction from a single rgb image

    Chen Liu, Jimei Yang, Duygu Ceylan, Ersin Yumer, and Ya- sutaka Furukawa. Planenet: Piece-wise planar reconstruction from a single rgb image. In Proc. Computer Vision and Pat- tern Recognition (CVPR), pages 2579–2588, 2018

    2018

  26. [34]

    Groundnet: Segmentation-aware monocular ground plane estimation with geometric consistency

    Yunze Man, Xinshuo Weng, and Kris Kitani. Groundnet: Segmentation-aware monocular ground plane estimation with geometric consistency. arXiv preprint arXiv:1811.07222 , 2018

    2018 arXiv

  27. [35]

    Optimal estima- tion of vanishing points in a manhattan world

    Faraz M Mirzaei and Stergios I Roumeliotis. Optimal estima- tion of vanishing points in a manhattan world. In Proc. Int. Conf. on Computer Vision (ICCV), pages 2454–2461. IEEE, 2011

    2011

  28. [36]

    Simple camera calibration from a single image using five points on two orthogonal 1-d objects

    Isao Miyagawa, Hiroyuki Arai, and Hideki Koike. Simple camera calibration from a single image using five points on two orthogonal 1-d objects. IEEE Transactions on Image Processing, 19(6):1528–1538, 2010

    2010

  29. [37]

    U-net: Convolutional networks for biomedical image segmentation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U-net: Convolutional networks for biomedical image segmentation. In International Conference on Medical image computing and computer-assisted intervention, pages 234–241. Springer, 2015

    2015

  30. [38]

    Imagenet large scale visual recognition challenge

    Olga Russakovsky, Jia Deng, Hao Su, Jonathan Krause, San- jeev Satheesh, Sean Ma, Zhiheng Huang, Andrej Karpathy, Aditya Khosla, Michael Bernstein, et al. Imagenet large scale visual recognition challenge. Int. J. of Computer Vision, 115(3):211–252, 2015

    2015

  31. [39]

    Fast image- based localization using direct 2d-to-3d matching

    Torsten Sattler, Bastian Leibe, and Leif Kobbelt. Fast image- based localization using direct 2d-to-3d matching. In Proc. Int. Conf. on Computer Vision (ICCV), 2011

    2011

  32. [40]

    Planar group- ing for automatic detection of vanishing lines and points

    Frederik Schaffalitzky and Andrew Zisserman. Planar group- ing for automatic detection of vanishing lines and points. Image and Vision Computing, 18(9):647–658, 2000

    2000

  33. [41]

    Semantic scene comple- tion from a single depth image

    Shuran Song, Fisher Yu, Andy Zeng, Angel X Chang, Mano- lis Savva, and Thomas Funkhouser. Semantic scene comple- tion from a single depth image. Proc. Computer Vision and Pattern Recognition (CVPR), 2017

    2017

  34. [42]

    Real-time manhattan world rotation estimation in 3d

    Julian Straub, Nishchal Bhandari, John J Leonard, and John W Fisher. Real-time manhattan world rotation estimation in 3d. In 2015 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 1913–1920. IEEE, 2015

    2015

  35. [43]

    The manhattan frame modelmanhat- tan world inference in the space of surface normals

    Julian Straub, Oren Freifeld, Guy Rosman, John J Leonard, and John W Fisher. The manhattan frame modelmanhat- tan world inference in the space of surface normals. Trans. Pattern Analysis and Machine Intelligence, 40(1):235–249, 2018

    2018

  36. [44]

    Discovery of latent 3d keypoints via end-to-end geometric reasoning

    Supasorn Suwajanakorn, Noah Snavely, Jonathan J Tompson, and Mohammad Norouzi. Discovery of latent 3d keypoints via end-to-end geometric reasoning. In Neural Information Processing Systems, pages 2063–2074, 2018

    2018

  37. [45]

    Perturbation theory for eigenvalue problems

    Nico van der Aa. Perturbation theory for eigenvalue problems. 2005

    2005

  38. [46]

    Design- ing deep networks for surface normal estimation

    Xiaolong Wang, David Fouhey, and Abhinav Gupta. Design- ing deep networks for surface normal estimation. In Proc. Computer Vision and Pattern Recognition (CVPR) , pages 539–547, 2015

    2015

  39. [47]

    Robust camera self- calibration from monocular images of manhattan worlds

    Horst Wildenauer and Allan Hanbury. Robust camera self- calibration from monocular images of manhattan worlds. In Proc. Computer Vision and Pattern Recognition (CVPR) , pages 2831–2838. IEEE, 2012

    2012

  40. [48]

    Hori- zon lines in the wild

    Scott Workman, Menghua Zhai, and Nathan Jacobs. Hori- zon lines in the wild. In Proc. British Machine Vision Conf. (BMVC), 2016

    2016

  41. [49]

    Recognizing scene viewpoint using panoramic place representation

    Jianxiong Xiao, Krista A Ehinger, Aude Oliva, and Anto- nio Torralba. Recognizing scene viewpoint using panoramic place representation. In Proc. Computer Vision and Pattern Recognition (CVPR), pages 2695–2702, 2012

    2012

  42. [50]

    Pose estimation from line correspondences: A complete analysis and a series of solutions.Trans

    Chi Xu, Lilian Zhang, Li Cheng, and Reinhard Koch. Pose estimation from line correspondences: A complete analysis and a series of solutions.Trans. Pattern Analysis and Machine Intelligence, 39(6):1209–1222, 2017

    2017

  43. [51]

    A minimum error vanishing point detection approach for uncalibrated monocular images of man-made environments

    Yiliang Xu, Sangmin Oh, and Anthony Hoogs. A minimum error vanishing point detection approach for uncalibrated monocular images of man-made environments. In Proc. Com- puter Vision and Pattern Recognition (CVPR), pages 1376– 1383, 2013

    2013

  44. [52]

    Detect- ing vanishing points using global image context in a non- manhattan world

    Menghua Zhai, Scott Workman, and Nathan Jacobs. Detect- ing vanishing points using global image context in a non- manhattan world. In Proc. Computer Vision and Pattern Recognition (CVPR), pages 5657–5665, 2016

    2016

  45. [53]

    Deep depth comple- tion of a single rgb-d image

    Yinda Zhang and Thomas Funkhouser. Deep depth comple- tion of a single rgb-d image. In Proc. Computer Vision and Pattern Recognition (CVPR), pages 175–185, 2018

    2018

Pith tools

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