mlsanityData to 7 Oct 2026

Two-View Geometry Scoring Without Correspondences

arXiv 2306.01596 ↗First posted 2 Jun 2023cs.CVcited by 11

Not on the map. Why?

Abstract (arXiv)

Camera pose estimation for two-view geometry traditionally relies on RANSAC. Normally, a multitude of image correspondences leads to a pool of proposed hypotheses, which are then scored to find a winning model. The inlier count is generally regarded as a reliable indicator of "consensus". We examine this scoring heuristic, and find that it favors disappointing models under certain circumstances. As a remedy, we propose the Fundamental Scoring Network (FSNet), which infers a score for a pair of overlapping images and any proposed fundamental matrix. It does not rely on sparse correspondences, but rather embodies a two-view geometry model through an epipolar attention mechanism that predicts the pose error of the two images. FSNet can be incorporated into traditional RANSAC loops. We evaluate FSNet on fundamental and essential matrix estimation on indoor and outdoor datasets, and establish that FSNet can successfully identify good poses for pairs of images with few or unreliable correspondences. Besides, we show that naively combining FSNet with MAGSAC++ scoring approach achieves state of the art results.

Key facts. Cited by 11 papers in the map (data through 7 Oct 2026). Builds on 49 of its 77 references that are in the map. First posted on arXiv 2 Jun 2023.

Builds on

49 of 77 references.

References · 49

  1. 2017Attention Is All You Needcited by 59,932
  2. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  3. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 29,915
  4. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,604
  5. 2015Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shiftcited by 6,979
  6. 2017ScanNet: Richly-annotated 3D Reconstructions of Indoor Scenescited by 3,088
  7. 2020Transformers are RNNs: Fast Autoregressive Transformers with Linear Attentioncited by 1,979
  8. 2019SuperGlue: Learning Feature Matching with Graph Neural Networkscited by 1,437
  9. 2017SuperPoint: Self-Supervised Interest Point Detection and Descriptioncited by 1,313
  10. 2021LoFTR: Detector-Free Local Feature Matching with Transformerscited by 974
  11. 2018MegaDepth: Learning Single-View Depth Prediction from Internet Photoscited by 731
  12. 2015Empirical Evaluation of Rectified Activations in Convolutional Networkcited by 515
37 more
  1. 2015YFCC100M: The New Data in Multimedia Researchcited by 234
  2. 2019Kornia: an Open Source Differentiable Computer Vision Library for PyTorchcited by 231
  3. 2020Image Matching across Wide Baselines: From Paper to Practicecited by 224
  4. 2019R2D2: Repeatable and Reliable Detector and Descriptorcited by 206
  5. 2020Revisiting Stereo Depth Estimation From a Sequence-to-Sequence Perspective with Transformerscited by 190
  6. 2021COTR: Correspondence Transformer for Matching Across Imagescited by 182
  7. 2017Learning to Find Good Correspondencescited by 176
  8. 2019MAGSAC++, a fast, reliable and accurate robust estimatorcited by 160
  9. 2019Learning Two-View Correspondences and Geometry Using Order-Aware Networkcited by 148
  10. 2017Graph-Cut RANSACcited by 133
  11. 2019SOSNet: Second Order Similarity Regularization for Local Descriptor Learningcited by 131
  12. 2019Neural-Guided RANSAC: Learning Where to Sample Model Hypothesescited by 129
  13. 2018MAGSAC: marginalizing sample consensuscited by 122
  14. 2019Key.Net: Keypoint Detection by Handcrafted and Learned CNN Filterscited by 117
  15. 2020Epipolar Transformerscited by 117
  16. 2021TransMVSNet: Global Context-aware Multi-view Stereo Network with Transformerscited by 117
  17. 2021Learning Accurate Dense Correspondences and When to Trust Themcited by 113
  18. 2022Map-free Visual Relocalization: Metric Pose Relative to a Single Imagecited by 112
  19. 2021Learning to Match Features with Seeded Graph Matching Networkcited by 92
  20. 2022Multi-Frame Self-Supervised Depth with Transformerscited by 64
  21. 2021On the Limits of Pseudo Ground Truth in Visual Camera Re-localisationcited by 63
  22. 2019To Learn or Not to Learn: Visual Localization from Essential Matricescited by 61
  23. 2017Repeatability Is Not Enough: Learning Affine Regions via Discriminabilitycited by 57
  24. 2014TILDE: A Temporally Invariant Learned DEtectorcited by 55
  25. 2022MVSTER: Epipolar Transformer for Efficient Multi-View Stereocited by 50
  26. 2021Extreme Rotation Estimation using Dense Correlation Volumescited by 42
  27. 2022The 8-Point Algorithm as an Inductive Bias for Relative Pose Prediction by ViTscited by 38
  28. 2019NM-Net: Mining Reliable Neighbors for Robust Feature Correspondencescited by 29
  29. 2021VSAC: Efficient and Accurate Estimator for H and Fcited by 17
  30. 2015Learning to Assign Orientations to Feature Pointscited by 16
  31. 2018Deep Fundamental Matrix Estimation without Correspondencescited by 14
  32. 2020AdaLAM: Revisiting Handcrafted Outlier Detectioncited by 14
  33. 2020HDD-Net: Hybrid Detector Descriptor with Mutual Interactive Learningcited by 13
  34. 2021ScaleNet: A Shallow Architecture for Scale Estimationcited by 12
  35. 2021On the Instability of Relative Pose Estimation and RANSAC's Rolecited by 10
  36. 2022NeFSAC: Neurally Filtered Minimal Samplescited by 9
  37. 2017Fundamental Matrix Estimation: A Study of Error Criteriacited by 6

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 1; 2024: 4; 2025: 5; 2026: 1.

11 papers.

  1. 2023Unifying Correspondence, Pose and NeRF for Pose-Free Novel View Synthesis from Stereo Pairscited by 37
  2. 2024Matching 2D Images in 3D: Metric Relative Pose from Metric Correspondencescited by 25
  3. 2025Relative Pose Estimation through Affine Corrections of Monocular Depth Priorscited by 18
  4. 2024FAR: Flexible, Accurate and Robust 6DoF Relative Camera Pose Estimationcited by 12
  5. 2025Deep Learning Reforms Image Matching: A Survey and Outlookcited by 7
  6. 2025A Scene is Worth a Thousand Features: Feed-Forward Camera Localization from a Collection of Image Featurescited by 4
  7. 2024Breaking the Frame: Visual Place Recognition by Overlap Predictioncited by 3
  8. 2025Dense Match Summarization for Faster Two-view Estimationcited by 2
  9. 2025RANSAC Scoring Functions: Analysis and Reality Checkcited by 2
  10. 2024PRAGO: Differentiable Multi-View Pose Optimization From Objectness Detectionscited by 0
  11. 2026Agentic Relative Camera Pose Estimation via Learned Ranking and Verificationcited by 0