Pith. sign in

REVIEW 3 major objections 3 minor 1 cited by

Unsupervised Homography Estimation on Multimodal Image Pair via Alternating Optimization

T0 review · 3 major / 3 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read AltO trains homography estimators on multimodal image pairs with no ground-truth homography, using alternating optimization to separate geometry and modality gaps.

desk verdict A genuinely new unsupervised multimodal homography method with strong empirical results, but the global-statistic geometry loss and missing error bars keep it from being fully settled. read the letter →

arxiv 2411.13036 v1 pith:O264Y2VI submitted 2024-11-20 cs.CV cs.AI

Sanghyeob Song , Jaihyun Lew , Hyemi Jang , Sungroh Yoon This is my paper · ORCID
classification cs.CVcs.AI
keywords homographyestimationmultimodalimagealignmentunsupervisedlearningalternatingoptimizationBarlowTwinsGeometryregistrationmodality-agnosticrepresentation
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

AltO is an unsupervised learning framework that estimates the homography between two images—the planar projective transform aligning two views—when the images come from different modalities, such as map versus satellite or RGB versus near-infrared, and no ground-truth homography is available. The paper argues that existing unsupervised methods fail on such pairs because they try to close the geometry gap and the modality gap simultaneously, which collapses into a trivial solution where the encoder produces a constant output and the network predicts the identity warp. AltO avoids this by alternating two optimization phases, similar to Expectation-Maximization: a Geometry Learning phase trains the registration network with a spatial variant of the Barlow Twins loss, and a Modality-Agnostic Representation Learning phase trains the encoder and projector with the standard Barlow Twins loss. On Google Map, Google Earth, and Deep NIR, AltO beats other unsupervised methods and approaches supervised performance; for example, IHN-1 trained with AltO reaches a mean corner error of 3.06 on Google Map compared with 0.92 for supervised IHN-1. This matters because ground-truth homographies are hard to collect in real stitching, fusion, and remote-sensing settings.

What carries the argument

The machinery is the alternating two-phase optimization schedule together with the two Barlow Twins-derived losses. Geometry Barlow Twins extends the Barlow Twins objective to 2D feature maps by treating the spatial locations (height and width) as samples, so the cross-correlation between channels of the warped and fixed feature maps is pulled toward the identity matrix, aligning local geometric features without requiring exact pixel equality. The Modality-Agnostic Representation Learning phase applies the standard Barlow Twins loss to global-average-pooled embeddings, making the representation space shared across modalities. Alternation is the anti-collapse device: each phase trains only one part of the model while freezing the other, so the encoder and the registration network cannot jointly 'solve' the loss by outputting a constant and an identity warp. Global average pooling is also load-bearing; without it, the MARL phase forces fine local features to match before the registration network is trained, and the ablation in Section 6.2 shows this re-introduces the trivial solution.

What would settle it

Train IHN-1 with AltO on a new multimodal pair dataset with a large modality gap using the same hyperparameters; if the mean corner error stays near the roughly 24-pixel identity-warp baseline, or the encoder's output becomes near-constant while the registration network predicts near-identity homographies, the alternating schedule failed to prevent collapse and the paper's central claim is undercut.

Watch

Extended reading notes

Core claim

The paper's central claim is that the trivial-solution collapse that ruins naive unsupervised homography training on multimodal pairs can be avoided by an alternating schedule, so a registration network can be trained without any ground-truth homography even when the input pair spans different domains. In the Geometry Learning phase, the registration network is trained by warping the moving image and maximizing the similarity between its features and those of the fixed image using the proposed Geometry Barlow Twins loss, which reinterprets the spatial dimensions of feature maps as the batch dimension and drives the channel cross-correlation matrix toward the identity. In the Modality-Agnostic Representation Learning phase, with the registration network frozen, the shared encoder and projector are trained with the original Barlow Twins loss so the two modalities map into a common feature space. The two phases alternate within each mini-batch, and the paper reports that this, together with global average pooling in the projector, prevents the encoder from collapsing and the network from predicting the identity homography. The framework is architecture-agnostic: the same alternating procedure trains DHN, RAFT, IHN-1, and RHWF-1, and in each case it outperforms existing unsupervised methods by a wide margin while approaching supervised accuracy.

Load-bearing premise

The load-bearing premise is that the alternating GL/MARL schedule prevents the trivial-solution collapse in which the encoder outputs a constant and the registration network predicts the identity homography; the paper supports this only with the Google Map ablation in Section 6.1 and offers no formal guarantee.

Editorial extensions

If this is right

  • AltO can replace ground-truth homography labels during training for any compatible registration network, and the reported mean corner errors approach those of supervised training of the same network (IHN-1+AltO: 3.06 versus 0.92 on Google Map, 1.82 versus 1.60 on Google Earth, 3.11 versus 2.11 on Deep NIR).
  • Without the alternating schedule, training fails: Table 2 shows MACE values around 24 pixels (no better than an identity warp) for every tested backbone, while the alternating version drops to roughly 3-6 pixels.
  • Removing global average pooling from the projector also collapses training (MACE stays near 24 pixels), so the MARL phase must compare global rather than fine local features until the registration network is ready.
  • Because the framework is architecture-agnostic, stronger or future registration networks can be plugged in directly and inherit the same unsupervised training procedure.
  • The label-free regime extends the practical reach of homography estimation to multimodal stitching, multispectral fusion, and remote-sensing alignment, where collecting ground-truth warps is impractical.

Reading between the lines

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

  • An implication the paper leaves implicit is that its alternating schedule is essentially a coordinate-descent heuristic with no formal convergence or anti-collapse guarantee; proving a condition under which alternation avoids the trivial solution would turn the empirical recipe into a general principle.
  • The GAP requirement suggests a resolution-versus-representation trade-off: a testable extension would be replacing GAP with positional pooling or a lightweight attention pooling to see whether multimodal alignment can be pushed to higher-resolution images without collapse.
  • Because AltO is label-free and architecture-agnostic, the same two-phase schedule may transfer to other geometric alignment tasks on multimodal data, such as optical flow or stereo disparity, although the paper only demonstrates homography.
  • The residual gap to supervised performance (for instance 3.06 versus 0.92 on Google Map) suggests the modality-agnostic representation still loses some spatial precision; a concrete next experiment is to couple AltO with a refinement stage that uses the original image resolution.
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 / 3 minor

Summary. The paper proposes AltO, an unsupervised framework for homography estimation on multimodal image pairs (e.g., satellite/map, RGB/NIR). The training alternates two phases: a Geometry Learning (GL) phase that trains the registration network with a proposed Geometry Barlow Twins (GBT) loss on spatial feature maps, and a Modality-Agnostic Representation Learning (MARL) phase that trains an encoder/projector with a standard Barlow Twins loss. The framework is evaluated on Google Map, Google Earth, and Deep NIR datasets using four different registration backbones (DHN, RAFT, IHN-1, RHWF-1). The authors report that AltO outperforms existing unsupervised methods and approaches supervised performance, and they provide ablations on alternating optimization, global average pooling, loss-function combinations, and encoder/projector architecture.

Significance. If the central claim holds, AltO would be a notable step for unsupervised multimodal homography estimation: it decouples the geometry and modality gaps in a principled alternating scheme, works across several registration architectures, and comes with released source code. The MACE numbers in Table 1 are substantially better than the existing unsupervised baselines on all three benchmarks, and the consistency across four backbone architectures is a genuine strength. However, the empirical claims are currently supported by single-run results, and one load-bearing failure mode of the GBT loss is not tested, so the significance is conditional on those points being resolved.

major comments (3)
  1. [§4.2, Eq. (5)] The GBT loss Lg computes normalized cross-correlations over spatial positions but does not constrain the absolute feature magnitude or the information content of the region onto which the moving image is warped. If the fixed image contains low-texture areas, the loss can in principle be minimized by a homography that maps the moving image onto such a region, since the diagonal correlation term (1-C(n,ii))^2 can be driven to zero for low-variance features. The paper's trivial-solution discussion in §3.2 covers only the constant-encoder/identity-homography collapse, and the ablations in §6.1–6.2 do not vary texture content, occlusion, or initial displacement magnitude to rule out this complementary failure mode. Please add experiments (e.g., synthetic pairs with low-texture fixed images, or monitoring of the region where the warped image lands) that demonstrate the GL phase cannot collapse to low-information regions, or otherwise modify the loss to explicitly penalize such degenerate warps.
  2. [Table 1 and §5.3] All quantitative claims rest on single-run MACE values with no error bars, confidence intervals, or number of seeds. Given that the reported performance differences between AltO and the strongest unsupervised baselines are sometimes large but the variance is unknown, the claim that AltO 'outperforms' existing methods is not yet statistically supported. Please report mean and standard deviation over at least three seeds, and state the evaluation protocol used for the baseline numbers.
  3. [§6.4 and §6.1] The architecture split in Table 5 is selected using the Google Map test set, because no validation split is described; this test-set selection can inflate the reported numbers and weaken the claim that the chosen Encoder(1,2)/Projector(3) split is optimal. In addition, Table 2 omits the 'Alternating without MARL' condition, so the ablation cannot separate the contribution of alternating optimization from the contribution of the MARL phase. Please add the missing condition and perform the ablations on a held-out validation split.
minor comments (3)
  1. [§5.4] The statement that the remaining gap to supervised methods is 'negligible in practical applications' is not supported by any perceptual or downstream-task evaluation; a roughly three-fold larger corner error can be material in stitching or fusion. Either soften the claim or add a quantitative downstream evaluation.
  2. [§4.2, Eq. (5)] The notation for the feature tensors is inconsistent: the text describes f_A as 'feature vectors' in R^{N×D×H×W}, and the definition 'subtracting the spatial mean from each unit' is ambiguous about which axis is centered. Please clarify by defining the normalization explicitly over h,w for each (n,i).
  3. [§6.2] The explanation that GAP prevents the trivial solution because 'local features are forced to be similar' is speculative; a more direct diagnostic, such as feature statistics before and after the projector with and without GAP, would make the claim more convincing.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the unsupervised training objective is not fitted to the held-out MACE metric, and no load-bearing self-citation or definitional reduction is present.

full rationale

The paper's derivation is self-contained. The two-phase training objective is defined directly by Eq. (5) (Geometry Barlow Twins) and Eq. (6) (modality loss), and the registration network is optimized against those losses without any ground-truth homography. The reported MACE (Eq. 7) is computed from held-out ground-truth homographies and is used only for evaluation, so the central claim is not a fitted input renamed as a prediction. The only hyperparameters (lambda = 0.005, encoder/projector split, alternating schedule) are adopted from the external Barlow Twins work or justified by ablations on Google Map; none is tuned to the target benchmark result. All citations are to external prior work (Barlow Twins, biHomE, UDHN, IHN, RHWF), with no load-bearing self-citation chain and no imported uniqueness theorem. The trivial-solution concern in Section 3.2 is an optimization-collapse risk, and the alternating-schedule/GAP ablations in Tables 2-3 provide empirical evidence against that collapse; even if one doubts the strength of that evidence, that is a correctness or robustness concern, not circularity of the derivation. The limitation section explicitly acknowledges a performance gap versus supervised training, which further supports that the comparison is empirical rather than constructed. Therefore the appropriate finding is no significant circularity, with score 0.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claim rests on standard homography/DLT math, a domain assumption that planar homographies relate the paired images, and two design assumptions of AltO: that GBT gradients align geometry, and that alternating optimization avoids trivial collapse. No new physical entities are introduced.

free parameters (3)
  • lambda (redundancy reduction weight) = 0.005
    Set to the default from Barlow Twins [12], Section 5.2; no sensitivity analysis or tuning on the target benchmarks is reported.
  • Encoder/projector split of ResNet-34 = Encoder: stem + stages 1,2; Projector: stage 3 + GAP
    Selected by ablation on Google Map (Section 6.4, Table 5); the choice affects MACE from 3.78 to 9.76 depending on configuration.
  • Training schedule hyperparameters = lr 3e-4, weight decay 1e-5, 200 epochs, batch 16, gradient clip 1.0
    Reported in Section 5.2 without sensitivity analysis; these are choices that could affect the comparison.
assumptions (4)
  • domain assumption Planar homography assumption for all benchmark image pairs.
    Evaluation and training assume a ground-truth homography exists; datasets are constructed or pre-registered to satisfy this (Sections 3.1 and 5.1).
  • domain assumption Feature alignment in a learned space is a valid proxy for geometric alignment.
    The GL phase minimizes GBT between warped and fixed feature maps; this assumes similarity in feature space tracks spatial correspondence (Section 4.2, Eq. 5).
  • ad hoc to paper Alternating GL and MARL avoids the trivial solution collapse.
    No formal guarantee is given; evidence is the ablation in Section 6.1, Table 2 on Google Map only.
  • standard math Standard DLT and homography parameterization math.
    Used to convert predicted offsets or flow into homography matrices (Section 3.1, Appendix A.2).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Unsupervised Homography Estimation on Multimodal Image Pair via Alternating Optimization." pith.science (2026). https://pith.science/paper/O264Y2VI

@misc{pith2026241113036,
  author       = {Pith},
  title        = {Pith review of: Unsupervised Homography Estimation on Multimodal Image Pair via Alternating Optimization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/O264Y2VI}},
  note         = {Machine review of arXiv:2411.13036}
}
read the original abstract

Estimating the homography between two images is crucial for mid- or high-level vision tasks, such as image stitching and fusion. However, using supervised learning methods is often challenging or costly due to the difficulty of collecting ground-truth data. In response, unsupervised learning approaches have emerged. Most early methods, though, assume that the given image pairs are from the same camera or have minor lighting differences. Consequently, while these methods perform effectively under such conditions, they generally fail when input image pairs come from different domains, referred to as multimodal image pairs. To address these limitations, we propose AltO, an unsupervised learning framework for estimating homography in multimodal image pairs. Our method employs a two-phase alternating optimization framework, similar to Expectation-Maximization (EM), where one phase reduces the geometry gap and the other addresses the modality gap. To handle these gaps, we use Barlow Twins loss for the modality gap and propose an extended version, Geometry Barlow Twins, for the geometry gap. As a result, we demonstrate that our method, AltO, can be trained on multimodal datasets without any ground-truth data. It not only outperforms other unsupervised methods but is also compatible with various architectures of homography estimators. The source code can be found at:~\url{https://github.com/songsang7/AltO}

Figures

Figures reproduced from arXiv: 2411.13036 by the authors.

Figure 1
Figure 1. Examples of types of gaps. This paper will address both geometry and modality gaps [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Conceptual Diagram of the Barlow Twins Method [12] Barlow Twins, which we adopt, requires further explanation for the subsequent sections. Fig￾ure 2 illustrates the conceptual diagram of the Barlow Twins method. The encoder extracts the representations r A and r B from inputs x A and x B, then produces embedding vectors v A and v B by passing these r A and r B through the projector. At this point, the dimension of b… view at source ↗
Figure 3
Figure 3. Overview of architecture. Upper diagram shows static view and lower diagrams illustrate [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Examples of image pair for each datasets. Google Map and Google Earth are introduced by [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Visualization of homography estimation using center box. The first row shows the state [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: Moving, fixed, and warped images. The first row displays the moving image ( [PITH_FULL_IMAGE:figures/full_fig_p015_6.png]
Figure 7
Figure 7. Figure 7: Visualization of the iterative registration process in AltO. For RAFT [ [PITH_FULL_IMAGE:figures/full_fig_p016_7.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. Collaborative Learning for Unsupervised Multimodal Remote Sensing Image Registration: Integrating Self-Supervision and MIM-Guided Diffusion-Based Image Translation

    eess.IV 2025-05 conditional novelty 6.0 of 10

    An unsupervised collaborative training framework with MIM-guided diffusion translation and pseudo-label distillation reaches competitive or better cross-modal registration accuracy on five remote sensing datasets than...

Reference graph

Works this paper leans on

40 extracted references · 16 canonical work pages · cited by 1 Pith paper

  1. [1]

    Matthew Brown and David G. Lowe. Automatic panoramic image stitching using invariant features. Int. J. Comput. Vis., 74(1):59–73, 2007. doi: 10.1007/S11263-006-0002-3. URL https://doi.org/10.1007/s11263-006-0002-3

    doi:10.1007/s11263-006-0002-3 2007

  2. [2]

    Yuan Zhou, Anand Rangarajan, and Paul D. Gader. An integrated approach to registration and fusion of hyperspectral and multispectral images. IEEE Trans. Geosci. Remote. Sens., 58(5): 3020–3033, 2020. doi: 10.1109/TGRS.2019.2946803. URL https://doi.org/10.1109/ TGRS.2019.2946803

    work page arXiv 2020

  3. [3]

    3d reconstruction based on homography mapping

    Zhongfei Zhang and Allen R Hanson. 3d reconstruction based on homography mapping. Proc. ARPA96, pages 1007–1012, 1996

    work page 1996

  4. [4]

    Zhang and Y .F

    B. Zhang and Y .F. Li. An efficient method for dynamic calibration and 3d reconstruction using homographic transformation. Sensors and Actuators A: Physical , 119(2):349–357,

    work page

  5. [5]

    David G. Lowe. Distinctive image features from scale-invariant keypoints. Int. J. Comput. Vis., 60(2):91–110, 2004. doi: 10.1023/B:VISI.0000029664.99615.94. URL https://doi.org/ 10.1023/B:VISI.0000029664.99615.94

    arXiv 2004

  6. [6]

    SURF: speeded up robust features

    Herbert Bay, Tinne Tuytelaars, and Luc Van Gool. SURF: speeded up robust features. In Ales Leonardis, Horst Bischof, and Axel Pinz, editors, Computer Vision - ECCV 2006, 9th European Conference on Computer Vision, Graz, Austria, May 7-13, 2006, Proceedings, Part I, volume 3951 of Lecture Notes in Computer Science, pages 404–417. Springer, 2006. doi: 10.10...

    doi:10.1007/11744023 2006

  7. [7]

    Ethan Rublee, Vincent Rabaud, Kurt Konolige, and Gary R. Bradski. ORB: an efficient alternative to SIFT or SURF. In Dimitris N. Metaxas, Long Quan, Alberto Sanfeliu, and Luc Van Gool, editors, IEEE International Conference on Computer Vision, ICCV 2011, Barcelona, Spain, November 6-13, 2011, pages 2564–2571. IEEE Computer Society, 2011. doi: 10.1109/ICCV ...

    arXiv 2011

  8. [8]

    Deep image homography estimation

    Daniel DeTone, Tomasz Malisiewicz, and Andrew Rabinovich. Deep image homography estimation. CoRR, abs/1606.03798, 2016. URL http://arxiv.org/abs/1606.03798

    arXiv 2016

Show all 40 references
  1. [9]

    Chen, Shreyas S

    Ty Nguyen, Steven W. Chen, Shreyas S. Shivakumar, Camillo Jose Taylor, and Vijay Kumar. Unsupervised deep homography: A fast and robust homography estimation model. IEEE Robotics Autom. Lett., 3(3):2346–2353, 2018. doi: 10.1109/LRA.2018.2809549. URL https: //doi.org/10.1109/LR...

    2018

  2. [10]

    Perceptual loss for robust unsupervised homography estimation

    Daniel Koguciuk, Elahe Arani, and Bahram Zonooz. Perceptual loss for robust unsupervised homography estimation. In IEEE Conference on Computer Vision and Pattern Recog- nition Workshops, CVPR Workshops 2021, virtual, June 19-25, 2021 , pages 4274–4283. Computer Vision Foundati...

    2021

  3. [11]

    Deep lucas-kanade homography for multimodal image alignment

    Yiming Zhao, Xinming Huang, and Ziming Zhang. Deep lucas-kanade homography for multimodal image alignment. In IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2021, virtual, June 19-25, 2021 , pages 15950–15959. Computer Vision Foundation / IEEE, 2021. doi: 10....

    2021

  4. [12]

    Barlow twins: Self- supervised learning via redundancy reduction

    Jure Zbontar, Li Jing, Ishan Misra, Yann LeCun, and Stéphane Deny. Barlow twins: Self- supervised learning via redundancy reduction. In Marina Meila and Tong Zhang, editors, Proceedings of the 38th International Conference on Machine Learning, ICML 2021, 18-24 July 2021, Virtu...

    2021

  5. [13]

    Rift: Multi-modal image matching based on radiation- variation insensitive feature transform

    Jiayuan Li, Qingwu Hu, and Mingyao Ai. Rift: Multi-modal image matching based on radiation- variation insensitive feature transform. IEEE Transactions on Image Processing, 29:3296–3310,

  6. [14]

    Pos-gift: A geometric and intensity-invariant feature transformation for multimodal images.Information Fusion, 102:102027, 2024

    Zhuolu Hou, Yuxuan Liu, and Li Zhang. Pos-gift: A geometric and intensity-invariant feature transformation for multimodal images.Information Fusion, 102:102027, 2024. ISSN 1566-2535. doi: https://doi.org/10.1016/j.inffus.2023.102027. URL https://www.sciencedirect.com/ science/...

    2024

  7. [15]

    Dalal and B

    N. Dalal and B. Triggs. Histograms of oriented gradients for human detection. In 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR’05), volume 1, pages 886–893 vol. 1, 2005. doi: 10.1109/CVPR.2005.177

    2005 doi

  8. [16]

    Very deep convolutional networks for large-scale image recognition

    Karen Simonyan and Andrew Zisserman. Very deep convolutional networks for large-scale image recognition. In Yoshua Bengio and Yann LeCun, editors, 3rd International Conference on Learning Representations, ICLR 2015, San Diego, CA, USA, May 7-9, 2015, Conference Track Proceedin...

    2015 arXiv

  9. [17]

    Multiple view geometry in computer vision (2

    Andrew Harltey and Andrew Zisserman. Multiple view geometry in computer vision (2. ed.). Cambridge University Press, 2006. ISBN 978-0-521-54051-3

    2006

  10. [18]

    Iterative deep homography estimation

    Si-Yuan Cao, Jianxin Hu, Ze-Hua Sheng, and Hui-Liang Shen. Iterative deep homography estimation. In IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2022, New Orleans, LA, USA, June 18-24, 2022, pages 1869–1878. IEEE, 2022. doi: 10.1109/ CVPR52688.2022.0019...

    2022

  11. [19]

    Recurrent homography estimation using homography-guided image warping and focus transformer

    Si-Yuan Cao, Runmin Zhang, Lun Luo, Beinan Yu, Zehua Sheng, Junwei Li, and Hui-Liang Shen. Recurrent homography estimation using homography-guided image warping and focus transformer. In IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2023, Vancouver, BC, ...

    2023

  12. [20]

    Gomez, Lukasz Kaiser, and Illia Polosukhin

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser, and Illia Polosukhin. Attention is all you need. In Isabelle Guyon, Ulrike von Luxburg, Samy Bengio, Hanna M. Wallach, Rob Fergus, S. V . N. Vishwanathan, and Roman Garnett,...

    2017

  13. [21]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In 2016 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2016, Las Vegas, NV , USA, June 27-30, 2016, pages 770–778. IEEE Computer Society, 2016. doi: 10.1109/C...

    2016 doi

  14. [22]

    Imagenet: A large- scale hierarchical image database

    Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. Imagenet: A large- scale hierarchical image database. In 2009 IEEE conference on computer vision and pattern recognition, pages 248–255. Ieee, 2009. 12

    2009

  15. [23]

    A multiscale framework with unsupervised learning for remote sensing image registration.IEEE Transactions on Geoscience and Remote Sensing, 60:1–15, 2022

    Yuanxin Ye, Tengfeng Tang, Bai Zhu, Chao Yang, Bo Li, and Siyuan Hao. A multiscale framework with unsupervised learning for remote sensing image registration.IEEE Transactions on Geoscience and Remote Sensing, 60:1–15, 2022. doi: 10.1109/TGRS.2022.3167644

    2022

  16. [24]

    Fast and robust matching for multimodal remote sensing image registration

    Yuanxin Ye, Lorenzo Bruzzone, Jie Shan, Francesca Bovolo, and Qing Zhu. Fast and robust matching for multimodal remote sensing image registration. IEEE Transactions on Geoscience and Remote Sensing, 57(11):9059–9070, 2019. doi: 10.1109/TGRS.2019.2924684

    2019

  17. [25]

    Yu, and Dahua Lin

    Zhirong Wu, Yuanjun Xiong, Stella X. Yu, and Dahua Lin. Unsupervised feature learning via non-parametric instance discrimination. In 2018 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2018, Salt Lake City, UT, USA, June 18-22, 2018 , pages 3733–3742. Compute...

    2018

  18. [26]

    Ting Chen, Simon Kornblith, Mohammad Norouzi, and Geoffrey E. Hinton. A simple framework for contrastive learning of visual representations. In Proceedings of the 37th International Conference on Machine Learning, ICML 2020, 13-18 July 2020, Virtual Event, volume 119 of Procee...

    2020

  19. [27]

    Exploring simple siamese representation learning

    Xinlei Chen and Kaiming He. Exploring simple siamese representation learning. In IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2021, virtual, June 19-25, 2021, pages 15750–15758. Computer Vision Foundation / IEEE, 2021. doi: 10.1109/CVPR46437. 2021.01549. UR...

    2021

  20. [28]

    Vicreg: Variance-invariance-covariance regu- larization for self-supervised learning

    Adrien Bardes, Jean Ponce, and Yann LeCun. Vicreg: Variance-invariance-covariance regu- larization for self-supervised learning. In The Tenth International Conference on Learning Representations, ICLR 2022, Virtual Event, April 25-29, 2022. OpenReview.net, 2022. URL https://op...

    2022

  21. [29]

    Dempster, Nan M

    Arthur P. Dempster, Nan M. Laird, and Donald B. Rubin. Maximum likelihood from in- complete data via the em - algorithm plus discussions on the paper. 1977. URL https: //api.semanticscholar.org/CorpusID:4193919

    1977

  22. [30]

    Network in network

    Min Lin, Qiang Chen, and Shuicheng Yan. Network in network. CoRR, abs/1312.4400, 2013. URL https://api.semanticscholar.org/CorpusID:16636683

    2013 arXiv

  23. [31]

    MacDonald

    Inkyu Sa, Jong Yoon Lim, Ho Seok Ahn, and Bruce A. MacDonald. deepnir: Datasets for generating synthetic NIR images and improved fruit detection system using deep learning techniques. Sensors, 22(13):4721, 2022. doi: 10.3390/S22134721. URL https://doi.org/ 10.3390/s22134721

    2022 doi

  24. [32]

    Multi-spectral sift for scene category recognition

    Matthew Brown and Sabine Süsstrunk. Multi-spectral sift for scene category recognition. In CVPR 2011, pages 177–184, 2011. doi: 10.1109/CVPR.2011.5995637

    2011

  25. [33]

    RAFT: recurrent all-pairs field transforms for optical flow

    Zachary Teed and Jia Deng. RAFT: recurrent all-pairs field transforms for optical flow. In Andrea Vedaldi, Horst Bischof, Thomas Brox, and Jan-Michael Frahm, editors,Computer Vision - ECCV 2020 - 16th European Conference, Glasgow, UK, August 23-28, 2020, Proceedings, Part II, ...

    2020 doi

  26. [34]

    Decoupled weight decay regularization

    Ilya Loshchilov and Frank Hutter. Decoupled weight decay regularization. In 7th International Conference on Learning Representations, ICLR 2019, New Orleans, LA, USA, May 6-9, 2019. OpenReview.net, 2019. URL https://openreview.net/forum?id=Bkg6RiCqY7

    2019

  27. [35]

    Super-convergence: Very fast training of neural networks using large learning rates

    Leslie N Smith and Nicholay Topin. Super-convergence: Very fast training of neural networks using large learning rates. In Artificial intelligence and machine learning for multi-domain operations applications, volume 11006, pages 369–386. SPIE, 2019. 13

    2019

  28. [36]

    Content-aware unsupervised deep homography estimation

    Jirong Zhang, Chuan Wang, Shuaicheng Liu, Lanpeng Jia, Nianjin Ye, Jue Wang, Ji Zhou, and Jian Sun. Content-aware unsupervised deep homography estimation. In Andrea Vedaldi, Horst Bischof, Thomas Brox, and Jan-Michael Frahm, editors, Computer Vision - ECCV 2020 - 16th European...

    2020 doi

  29. [37]

    Representation learning with contrastive pre- dictive coding

    Aäron van den Oord, Yazhe Li, and Oriol Vinyals. Representation learning with contrastive pre- dictive coding. CoRR, abs/1807.03748, 2018. URL http://arxiv.org/abs/1807.03748

    2018 arXiv

  30. [38]

    Efros, Richard Zhang, and Jun-Yan Zhu

    Taesung Park, Alexei A. Efros, Richard Zhang, and Jun-Yan Zhu. Contrastive learning for unpaired image-to-image translation. In Andrea Vedaldi, Horst Bischof, Thomas Brox, and Jan-Michael Frahm, editors, Computer Vision - ECCV 2020 - 16th European Conference, Glasgow, UK, Augu...

    2020 doi

  31. [2005]

    doi: https://doi.org/10.1016/j.sna.2004.10.013

    ISSN 0924-4247. doi: https://doi.org/10.1016/j.sna.2004.10.013. URL https://www. sciencedirect.com/science/article/pii/S0924424704007630

    2004 doi

  32. [2020]

    doi: 10.1109/TIP.2019.2959244

    2019

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.