Pith. sign in

REVIEW 4 major objections 6 minor 1 cited by

Enhancing Scene Coordinate Regression with Efficient Keypoint Detection and Sequential Information

T0 review · 4 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read The paper claims that saliency-guided patch selection and a cross-frame reprojection loss lift scene coordinate regression relocalization recall by 11 percentage points over the ACE baseline and raise single-frame speed to 90 Hz.

desk verdict Solid single-frame SCR gains, but the sequence-mode update equations are inconsistent as printed, so the headline sequence result isn't reproducible from the text. read the letter →

arxiv 2412.06488 v2 pith:A5P4YBGU submitted 2024-12-09 cs.RO cs.CV

Kuan Xu , Zeyu Jiang , Haozhi Cao , Shenghai Yuan , Chen Wang , Lihua Xie This is my paper · ORCID
classification cs.ROcs.CV
keywords scenecoordinateregressionvisualrelocalizationkeypointdetectionsaliency-guidedsamplingimplicittriangulationmulti-viewcross-lossrepetitivetexturesequence-based
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

Scene coordinate regression (SCR) locates a camera by having a network predict 3D scene coordinates for image pixels, then solving for pose. The paper argues that SCR's weak spots—textureless regions and repetitive textures—come from treating all pixels equally and from relying on implicit triangulation across views. It proposes a single scene-agnostic network that both encodes patches and scores their saliency, so mapping and relocalization can focus on the most informative, repeatable regions. A second ingredient, a cross-loss that re-projects predicted scene points onto keyframes, injects explicit multi-view constraints during mapping, which strengthens triangulation where textures repeat. On indoor benchmarks the full system lifts recall by 11 percentage points over the ACE baseline and runs single-frame relocalization at 90 Hz.

What carries the argument

The load-bearing mechanism is the unified scene-agnostic module $f_S$ that outputs both a patch feature vector and a saliency score for every image patch, together with the cross-loss $L_{C,i}$ that supervises a predicted scene point through its reprojection onto a matched keyframe. The saliency head (three convolutional layers over the shared backbone, with a 65-channel output for $8\times8$ cells plus a dustbin) ranks patches so that the same top-1,000 regions are used in mapping and relocalization. The cross-loss operationalizes multi-view triangulation: for each patch matched to a keyframe, the predicted 3D point is projected into the keyframe and penalized by the reprojection error, so the scene-specific head is trained to be consistent across viewpoints instead of only within a single image. Sequence relocalization then maintains a set of scene points whose reprojection errors stay small and refines them with inverse projection updates, so the multi-view consistency learned at mapping time is reused at query time.

What would settle it

A concrete test: take the full system on a target scene and replace the keypoint-head saliency ranking with random patch selection at the same 1,000-patch budget while keeping the cross-loss and sequence relocalization unchanged; if recall stays within the reported 11-percentage-point improvement, the saliency component is not doing the claimed work. A second test would measure the viewpoint repeatability of the top-1,000 patches (e.g., ground-truth 3D overlap across frames) in a scene whose texture statistics differ strongly from ScanNet, and check whether the recall gain tracks that repeatability.

Watch

Extended reading notes

Core claim

The central claim is that SCR can be made both faster and more accurate by redirecting compute toward salient, repeatable image patches and by supervising scene-point predictions from multiple viewpoints rather than one. Concretely, the paper extends the ACE backbone with a lightweight keypoint detection head that outputs a per-pixel saliency heatmap, trains it once on ScanNet using a learned keypoint detector as teacher, and then uses the top 1,000 saliency-ranked patches for buffer sampling and for keypoint selection at query time. During mapping, a self-loss supervises each patch's predicted 3D point from its own frame, and a cross-loss re-projects that point into a matched keyframe and penalizes the reprojection error, explicitly tying together observations of the same region. At relocalization, a single-frame mode applies PnP with RANSAC to the saliency-selected correspondences, and a sequence mode tracks maintained scene points by optical flow, estimates the pose from tracked correspondences, then fuses a network-predicted pose by inlier-weighted averaging on the manifold and refines scene points by inverse projection. The reported result is that this design improves recall over the ACE baseline by 6.4 percentage points in single-frame mode and 11 percentage points in sequence mode on 7-Scenes, while increasing single-frame throughput from 56 Hz to 90 Hz.

Load-bearing premise

The load-bearing premise is that the saliency ranking learned once on ScanNet by imitating a generic keypoint detector transfers to every target scene and that the top 1,000 patches it picks are repeatable across viewpoints; if that prior misranks pixels in a new environment, both mapping-time sampling and relocalization-time keypoint selection lose their benefit, and the paper gives no scene-specific calibration or failure analysis for that transfer.

Editorial extensions

If this is right

  • If the saliency prior transfers, SCR systems can sample fewer patches per image without losing accuracy, which directly explains the observed speedup from 56 Hz to 90 Hz in single-frame mode.
  • Explicit cross-frame supervision should reduce visual aliasing: scenes with repetitive textures, such as the Stairs sequence, show the largest gains (7.2% single-frame and 13.6% sequence-mode recall over ACE).
  • The method is not tied to one baseline: applying the same modules to GLACE improves recall by 8.1% on 7-Scenes and 3.1% on 12-Scenes, suggesting the components are transferable enhancements.
  • Sequence-mode relocalization depends on temporal continuity, so on datasets with large frame-to-frame motion (Cambridge Landmarks) only the single-frame mode is used and the gains shrink to a 3 cm median error reduction.

Reading between the lines

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

  • The paper does not test whether per-scene adaptation of the saliency head would improve repeatability; its ablation attributes 5.1 of the 11 percentage points on 7-Scenes to the keypoint head alone, so online adaptation is a natural experiment.
  • Because the cross-loss depends on feature matching, scenes with low match counts should weaken the multi-view supervision; measuring per-scene inlier match rates would predict where the sequence-mapping gain concentrates.
  • The paper itself suggests a vision-transformer backbone as future work; replacing the patch-limited receptive field with global context would test whether long-range dependencies further help repetitive-texture disambiguation.
  • The 90 Hz single-frame figure measures inference with saliency-selected keypoints; adding optical-flow tracking and pose fusion in sequence mode changes the cost profile, so the two modes' efficiency should not be compared on the same clock.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper presents SeqACE, an extension of the ACE scene-coordinate-regression pipeline. A shared backbone is augmented with a keypoint detection head (KDH) trained by distilling SuperPoint on ScanNet; the top-scoring patches are retained for mapping and relocalization. Sequence information is injected both during mapping, via a cross-loss that supervises predicted 3D points from keyframe observations, and during relocalization, via a sequence mode that tracks keypoints with optical flow, fuses PnP poses, and updates tracked 3D points. Experiments on 7-Scenes, 12-Scenes, and Cambridge Landmarks report recall and speed improvements over ACE, and a generalization experiment applies the method to GLACE.

Significance. If the reported results hold, the paper offers a practical improvement to ACE: a compact-map SCR system with faster single-frame relocalization and a sequence mode that improves recall in repetitive-texture environments. The generalization experiment in Table IV is a valuable check that the contributions are not ACE-specific, and the authors state that source code will be released. The main reservations are that the sequence-mode point-update equations are internally inconsistent as printed, the sequence-mode and speed measurement protocols are underspecified, and no variance estimates are given for the central accuracy claims. These issues are fixable, but they currently prevent full verification of the headline results.

major comments (4)
  1. [Section IV-D.2, Eqs. (13a) and (14)] The geometric derivation of the scene-point update is incorrect as printed. With camera center t_w^n and unit ray direction v_nd, the foot of the perpendicular from P to the observation ray is D = t_w^n + ((P - t_w^n)^T v_nd) v_nd. Equation (13a) instead prints ((P - t_w^n)^T v_nd) v_nd - (P - t_w^n), which equals the offset vector D - P, not the coordinates of D. Then Eq. (14) computes P_new = P + beta(D_bar - P). If D_bar is interpreted as the offset D - P from Eq. (13a), the update moves along D - 2P; if D_bar is interpreted as the foot D, the preceding equation is wrong. Because the sequence mode contributes 4.6 points of recall in Table V, this component is not reproducible from the text. The authors should correct Eq. (13a) to add the camera center, or change Eq. (14) to P_new = P + beta * D_bar, and ideally verify the released code against the corrected equations.
  2. [Section IV-D.2 and Section V-A] The sequence-mode evaluation protocol is underspecified. The paper does not state whether test frames are processed strictly in chronological order with only previously estimated poses, how the first frame is initialized, what happens when optical-flow tracking or PnP/RANSAC fails, how many RANSAC iterations are used, or which feature-tracking parameters are chosen. Since the sequence mode provides a headline gain (+4.6% on 7-Scenes in Table V) and is evaluated on video datasets, this protocol must be described in detail to rule out temporal leakage and to allow independent replication.
  3. [Section V-A, Tables I and II] The runtime comparison underlying the efficiency claim is not backed by a measurement protocol. The FPS values (56 Hz for ACE, 90 Hz for Ours-Single, 59 Hz for Ours-Sequence) are reported without specifying warmup, input resolution at inference, batch size, the exact timing method, the number of repeated runs, or whether the feature-cache used during mapping is included. This matters because 'increases the running speed from 56 Hz to 90 Hz' is a central contribution; please provide a precise measurement setup and, ideally, per-stage timings.
  4. [Tables I, II, and V] All accuracy numbers appear to be single runs with no error bars or multiple-seed experiments. The ablations in Table V include a 1.3% contribution from sequence-based mapping on 7-Scenes, and several per-scene differences are only a few points. Given the randomness in patch sampling, RANSAC, and neural-network training, mean-plus-std over at least three runs is needed to assess whether the reported gains are significant and to support the quantitative attribution in Table V.
minor comments (6)
  1. [Section V-A, Stairs paragraph] The sentence 'contributing recall rate improvements of 8.9% and 4.8%, respectively' is not consistent with Table V (KDH +5.1%, SM +1.3%, SR +4.6% on 7-Scenes) or with the Stairs numbers (ACE 3.8, single 11.0, sequence 17.4). Please clarify how the 8.9% figure is obtained.
  2. [Section IV-C] The keyframe selection criterion and the inlier definition used for Eq. (9) are vague: 'detected features' needs a precise reference, and the matching threshold for deciding whether a pair (p_k,j, p_i) is an inlier should be stated, along with the LightGlue settings used for patch matching.
  3. [Section IV-B] The KDH distillation procedure needs more details for reproducibility: the ScanNet split used, the number of training iterations, the learning rate, and whether SuperPoint is applied to grayscale or RGB inputs.
  4. [Tables I and II] The map-size column should state the unit explicitly (parameters vs. bytes) and whether it includes the backbone or only the scene-specific head; the text alternates between 4.1M and 4MB.
  5. [Section V-B and Table IV] The notation 'Ours' is ambiguous in the generalization table: Table IV uses 'ACE + Ours' and 'GLACE + Ours', but on 7-Scenes and 12-Scenes this appears to be the sequence mode while on Cambridge Landmarks it is the single mode. Please define this clearly.
  6. [Conclusion] The stated limitation that the CNN backbone uses only local patch information is acknowledged, but no diagnostic is provided to indicate how much of the remaining error is attributable to this issue; a short experiment or analysis would be helpful.

Circularity Check

0 steps flagged · score 1.0 of 10

No circularity: the reported gains are empirical comparisons against external baselines; the only self-citations are inspirational and not load-bearing.

full rationale

The paper's central derivation chain is empirical and externally anchored: the keypoint detection head is distilled from SuperPoint (an external teacher) on ScanNet, the mapping head is trained with self-loss and a multi-view cross-loss using LightGlue matches, and relocalization uses PnP/RANSAC and optical flow against the ACE or GLACE baselines. No equation defines the claimed output in terms of the quantity being predicted, and no fitted parameter is renamed as a prediction: the improved recall numbers are measured on held-out test sequences of 7-Scenes, 12-Scenes, and Cambridge Landmarks against external baselines. The KDH saliency prior is transferred from ScanNet, not fit to the target scenes, so the 'top-1,000 patches' selection is not a self-fulfilling fit. The authors do cite their own prior work (AirSLAM [6] and AirVO [38]), but only for architectural inspiration and classical visual-odometry context; these citations do not supply a uniqueness theorem, do not forbid alternatives, and are not the basis for the central claims. A separate correctness concern exists in Section IV-D.2: Eq. (13a) defines the perpendicular-foot offset rather than the foot coordinates, and Eq. (14) then subtracts the old point again, so the printed update vector is not the described projection toward the observation ray. This is a reproducibility/correctness issue, not a circularity, because it is not equivalent by construction to the experimental result. Overall, no circular step can be exhibited from the text.

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

The method relies on several hand-set hyperparameters and domain assumptions about the reliability of pre-trained keypoint detectors and matchers. None of these are fitted to the target test data in a way that would invalidate the empirical claims, but they are load-bearing engineering choices that affect the reported results.

free parameters (5)
  • top_k_patches = 1000
    Number of highest-saliency patches selected per image during mapping and relocalization; chosen to align with ACE's 1024 patch budget.
  • cross_loss_weight_lambda = 0.5
    Weight for the cross-loss when a match is an inlier; set by hand, not tuned per scene.
  • scene_point_update_beta = 1/N_P
    Scale factor for the scene point update along the inverse projection direction; decreases as the number of observations grows.
  • pseudo_gt_depth = 10m
    Fixed depth used for inverse projection to compute pseudo ground truth 3D points in the 'otherwise' branch of the losses.
  • keyframe_threshold = 0.5
    A frame is selected as a keyframe when matched features with the previous keyframe are less than half of the detected features.
assumptions (5)
  • domain assumption SuperPoint keypoint predictions provide reliable saliency supervision that transfers from ScanNet to target scenes.
    Section IV-B: KDH is trained by distillation from SuperPoint on ScanNet; the method assumes this generalizes to 7-Scenes, 12-Scenes, and Cambridge without fine-tuning.
  • domain assumption LightGlue feature matches between keyframe and current frame are correct often enough to serve as supervision for the cross-loss.
    Section IV-C: the cross-loss uses matched pixels p_{k,j} as additional observations; incorrect matches would inject wrong supervision, though lambda is zeroed for non-inlier matches.
  • domain assumption Optical flow accurately tracks selected keypoints between adjacent frames in the sequence relocalization mode.
    Section IV-D: tracked points are used to compute the pose via PnP; this is only valid under small inter-frame motion, which is why Cambridge Landmarks is excluded.
  • ad hoc to paper The fixed-depth pseudo ground truth provides a useful training signal for invalid predictions.
    Section IV-C, Eq. 6-8: inverse projection at 10m depth is used in the 'otherwise' branch; this is a heuristic not grounded in scene geometry.
  • ad hoc to paper The keyframe selection rule ensures varying baselines without manual tuning.
    Section IV-C: the threshold (less than half of detected features matched) is chosen by hand and its sensitivity is not analyzed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Enhancing Scene Coordinate Regression with Efficient Keypoint Detection and Sequential Information." pith.science (2026). https://pith.science/paper/A5P4YBGU

@misc{pith2026241206488,
  author       = {Pith},
  title        = {Pith review of: Enhancing Scene Coordinate Regression with Efficient Keypoint Detection and Sequential Information},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/A5P4YBGU}},
  note         = {Machine review of arXiv:2412.06488}
}
read the original abstract

Scene Coordinate Regression (SCR) is a visual localization technique that utilizes deep neural networks (DNN) to directly regress 2D-3D correspondences for camera pose estimation. However, current SCR methods often face challenges in handling repetitive textures and meaningless areas due to their reliance on implicit triangulation. In this paper, we propose an efficient and accurate SCR system. Compared to existing SCR methods, we propose a unified architecture for both scene encoding and salient keypoint detection, allowing our system to prioritize the encoding of informative regions. This design significantly improves computational efficiency. Additionally, we introduce a mechanism that utilizes sequential information during both mapping and relocalization. The proposed method enhances the implicit triangulation, especially in environments with repetitive textures. Comprehensive experiments conducted across indoor and outdoor datasets demonstrate that the proposed system outperforms state-of-the-art (SOTA) SCR methods. Our single-frame relocalization mode improves the recall rate of our baseline by 6.4% and increases the running speed from 56Hz to 90Hz. Furthermore, our sequence-based mode increases the recall rate by 11% while maintaining the original efficiency.

Figures

Figures reproduced from arXiv: 2412.06488 by the authors.

Figure 1
Figure 1. In this paper, we propose an efficient SCR system that [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Our system comprises a scene-agnostic module and a scene-specific module. The scene-agnostic module identifies informative [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Sampling strategies comparison. ACE [8] randomly samples [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: In our sequence-based relocalization, the predicted 3D scene [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: The recall rates in repetitive texture environments as the [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: The relocalization trajectories of different SCR systems in two scenarios (top: Apt1_kitchen, bottom: Chess). We also report the [PITH_FULL_IMAGE:figures/full_fig_p007_6.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. Autonomous 3D Moving Target Encirclement and Interception with Range measurement

    cs.RO 2025-06 conditional novelty 5.0 of 10

    A two-drone range-only estimator and anti-synchronization controller lets autonomous guardians encircle and intercept a moving hostile UAV in 3D.

Reference graph

Works this paper leans on

42 extracted references · 20 canonical work pages · cited by 1 Pith paper

  1. [1]

    Visual camera re-localization from rgb and rgb-d images using dsac,

    E. Brachmann and C. Rother, “Visual camera re-localization from rgb and rgb-d images using dsac,” IEEE transactions on pattern analysis and machine intelligence , vol. 44, no. 9, pp. 5847–5865, 2021

    2021

  2. [2]

    maplab: An open framework for research in visual-inertial mapping and localization,

    T. Schneider, M. Dymczyk, M. Fehr, K. Egger, S. Lynen, I. Gilitschen- ski, and R. Siegwart, “maplab: An open framework for research in visual-inertial mapping and localization,” IEEE Robotics and Automa- tion Letters, vol. 3, no. 3, pp. 1418–1425, 2018

    work page 2018

  3. [3]

    From coarse to fine: Robust hierarchical localization at large scale,

    P.-E. Sarlin, C. Cadena, R. Siegwart, and M. Dymczyk, “From coarse to fine: Robust hierarchical localization at large scale,” in CVPR, 2019

    2019

  4. [4]

    Geometric loss functions for camera pose regression with deep learning,

    A. Kendall and R. Cipolla, “Geometric loss functions for camera pose regression with deep learning,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2017, pp. 5974–5983

    2017

  5. [5]

    Structure-from-motion revisited,

    J. L. Schönberger and J.-M. Frahm, “Structure-from-motion revisited,” in Conference on Computer Vision and Pattern Recognition (CVPR) , 2016

    work page 2016

  6. [6]

    Airslam: An efficient and illumination-robust point-line visual slam system,

    K. Xu, Y . Hao, S. Yuan, C. Wang, and L. Xie, “Airslam: An efficient and illumination-robust point-line visual slam system,” IEEE Transactions on Robotics , 2025

    2025

  7. [7]

    Visual localization via few-shot scene region classification,

    S. Dong, S. Wang, Y . Zhuang, J. Kannala, M. Pollefeys, and B. Chen, “Visual localization via few-shot scene region classification,” in 2022 International Conference on 3D Vision (3DV). IEEE, 2022, pp. 393– 402

    2022

  8. [8]

    Accelerated coordi- nate encoding: Learning to relocalize in minutes using rgb and poses,

    E. Brachmann, T. Cavallari, and V . A. Prisacariu, “Accelerated coordi- nate encoding: Learning to relocalize in minutes using rgb and poses,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 5044–5053

    2023

Show all 42 references
  1. [9]

    Focustune: Tun- ing visual localization through focus-guided sampling,

    S. T. Nguyen, A. Fontan, M. Milford, and T. Fischer, “Focustune: Tun- ing visual localization through focus-guided sampling,” in Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, 2024, pp. 3606–3615

    2024

  2. [10]

    Glace: Global local accelerated coordinate encoding,

    F. Wang, X. Jiang, S. Galliani, C. V ogel, and M. Pollefeys, “Glace: Global local accelerated coordinate encoding,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp. 21 562–21 571

    2024

  3. [11]

    Map- based visual-inertial localization: Consistency and complexity,

    Z. Zhang, Y . Jiao, S. Huang, R. Xiong, and Y . Wang, “Map- based visual-inertial localization: Consistency and complexity,” IEEE Robotics and Automation Letters , vol. 8, no. 3, pp. 1407–1414, 2023

    2023

  4. [12]

    Pe-vins: Accurate monocular visual-inertial slam with point-edge features,

    C. Liu, H. Yu, P. Cheng, W. Sun, J. Civera, and X. Chen, “Pe-vins: Accurate monocular visual-inertial slam with point-edge features,” IEEE Transactions on Intelligent Vehicles , 2024

    2024

  5. [13]

    Is geometry enough for matching in visual localization?

    Q. Zhou, S. Agostinho, A. Ošep, and L. Leal-Taixé, “Is geometry enough for matching in visual localization?” in European Conference on Computer Vision . Springer, 2022, pp. 407–425

    2022

  6. [14]

    Back to the feature: Learning robust camera localization from pixels to pose,

    P.-E. Sarlin, A. Unagar, M. Larsson, H. Germain, C. Toft, V . Larsson, M. Pollefeys, V . Lepetit, L. Hammarstrand, F. Kahl, et al. , “Back to the feature: Learning robust camera localization from pixels to pose,” in Proceedings of the IEEE/CVF conference on computer vision and...

    2021

  7. [15]

    Posenet: A convolutional network for real-time 6-dof camera relocalization,

    A. Kendall, M. Grimes, and R. Cipolla, “Posenet: A convolutional network for real-time 6-dof camera relocalization,” in Proceedings of the IEEE international conference on computer vision , 2015, pp. 2938–2946

    2015

  8. [16]

    Image-based localization using hourglass networks,

    I. Melekhov, J. Ylioinas, J. Kannala, and E. Rahtu, “Image-based localization using hourglass networks,” in Proceedings of the IEEE international conference on computer vision workshops , 2017, pp. 879–886

    2017

  9. [17]

    Vidloc: A deep spatio-temporal model for 6-dof video-clip relocalization,

    R. Clark, S. Wang, A. Markham, N. Trigoni, and H. Wen, “Vidloc: A deep spatio-temporal model for 6-dof video-clip relocalization,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2017, pp. 6856–6864

    2017

  10. [18]

    Synthetic view generation for absolute pose regression and image synthesis

    P. Purkait, C. Zhao, and C. Zach, “Synthetic view generation for absolute pose regression and image synthesis.” in BMVC, 2018, p. 69

    2018

  11. [19]

    Direct-posenet: Absolute pose regression with photometric consistency,

    S. Chen, Z. Wang, and V . Prisacariu, “Direct-posenet: Absolute pose regression with photometric consistency,” in 2021 International Conference on 3D Vision (3DV) . IEEE, 2021, pp. 1175–1185

    2021

  12. [20]

    Camera re- localization by exploiting multi-view constraints for scene coordinates regression,

    M. Cai, H. Zhan, C. Saroj Weerasekera, K. Li, and I. Reid, “Camera re- localization by exploiting multi-view constraints for scene coordinates regression,” in Proceedings of the IEEE/CVF International Conference on Computer Vision Workshops , 2019, pp. 0–0

    2019

  13. [21]

    Learning camera lo- calization via dense scene matching,

    S. Tang, C. Tang, R. Huang, S. Zhu, and P. Tan, “Learning camera lo- calization via dense scene matching,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2021, pp. 1831–1841

    2021

  14. [22]

    Kfnet: Learning temporal camera relocalization using kalman filtering,

    L. Zhou, Z. Luo, T. Shen, J. Zhang, M. Zhen, Y . Yao, T. Fang, and L. Quan, “Kfnet: Learning temporal camera relocalization using kalman filtering,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2020, pp. 4919–4928

    2020

  15. [23]

    A new approach to linear filtering and prediction problems,

    R. E. Kalman, “A new approach to linear filtering and prediction problems,” 1960

    1960

  16. [24]

    D2s: Representing sparse descriptors and 3d coordinates for camera relocalization,

    B.-T. Bui, H.-H. Bui, D.-T. Tran, and J.-H. Lee, “D2s: Representing sparse descriptors and 3d coordinates for camera relocalization,” IEEE Robotics and Automation Letters , 2024

    2024

  17. [25]

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

    E. Brachmann and C. Rother, “Learning less is more-6d camera localization via 3d surface regression,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2018, pp. 4654–4662

    2018

  18. [26]

    Reprojection errors as prompts for efficient scene coordinate regression,

    T.-R. Liu, H.-K. Yang, J.-M. Liu, C.-W. Huang, T.-C. Chiang, Q. Kong, N. Kobori, and C.-Y . Lee, “Reprojection errors as prompts for efficient scene coordinate regression,” in European Conference on Computer Vision. Springer, 2024, pp. 286–302

    2024

  19. [27]

    Features from accelerated segment test (fast),

    D. G. Viswanathan, “Features from accelerated segment test (fast),” in Proceedings of the 10th workshop on image analysis for multimedia interactive services, London, UK , 2009, pp. 6–8

    2009

  20. [28]

    ORB: An efficient alternative to sift or surf,

    E. Rublee, V . Rabaud, K. Konolige, and G. Bradski, “ORB: An efficient alternative to sift or surf,” in 2011 International conference on computer vision . IEEE, 2011, pp. 2564–2571

    2011

  21. [29]

    Superpoint: Self- supervised interest point detection and description,

    D. DeTone, T. Malisiewicz, and A. Rabinovich, “Superpoint: Self- supervised interest point detection and description,” in Proceedings of the IEEE conference on computer vision and pattern recognition workshops, 2018, pp. 224–236

    2018

  22. [30]

    R2d2: Reliable and repeatable detector and descriptor,

    J. Revaud, C. De Souza, M. Humenberger, and P. Weinzaepfel, “R2d2: Reliable and repeatable detector and descriptor,” Advances in neural information processing systems , vol. 32, 2019

    2019

  23. [31]

    D2-net: A trainable cnn for joint description and detection of local features,

    M. Dusmanu, I. Rocco, T. Pajdla, M. Pollefeys, J. Sivic, A. Torii, and T. Sattler, “D2-net: A trainable cnn for joint description and detection of local features,” in Proceedings of the ieee/cvf conference on computer vision and pattern recognition , 2019, pp. 8092–8101

    2019

  24. [32]

    Scannet: Richly-annotated 3d reconstructions of indoor scenes,

    A. Dai, A. X. Chang, M. Savva, M. Halber, T. Funkhouser, and M. Nießner, “Scannet: Richly-annotated 3d reconstructions of indoor scenes,” in Proc. Computer Vision and Pattern Recognition (CVPR), IEEE, 2017

    2017

  25. [33]

    Grad- norm: Gradient normalization for adaptive loss balancing in deep multitask networks,

    Z. Chen, V . Badrinarayanan, C.-Y . Lee, and A. Rabinovich, “Grad- norm: Gradient normalization for adaptive loss balancing in deep multitask networks,” in International conference on machine learning. PMLR, 2018, pp. 794–803

    2018

  26. [34]

    Hartley and A

    R. Hartley and A. Zisserman, Multiple view geometry in computer vision. Cambridge university press, 2003

    2003

  27. [35]

    Lightglue: Local feature matching at light speed,

    P. Lindenberger, P.-E. Sarlin, and M. Pollefeys, “Lightglue: Local feature matching at light speed,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2023, pp. 17 627– 17 638

    2023

  28. [36]

    Decoupled weight decay regularization,

    I. Loshchilov and F. Hutter, “Decoupled weight decay regularization,” in International Conference on Learning Representations , 2019

    2019

  29. [37]

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

    L. N. Smith and N. Topin, “Super-convergence: Very fast training of neural networks using large learning rates,” in Artificial intelligence and machine learning for multi-domain operations applications , vol. 11006. SPIE, 2019, pp. 369–386

    2019

  30. [38]

    Airvo: An illumination- robust point-line visual odometry,

    K. Xu, Y . Hao, S. Yuan, C. Wang, and L. Xie, “Airvo: An illumination- robust point-line visual odometry,” in 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) . IEEE, 2023, pp. 3429–3436

    2023

  31. [39]

    An iterative image registration technique with an application to stereo vision,

    B. D. Lucas and T. Kanade, “An iterative image registration technique with an application to stereo vision,” in IJCAI’81: 7th international joint conference on Artificial intelligence , vol. 2, 1981, pp. 674–679

    1981

  32. [40]

    Scene coordinate regression forests for camera relocalization in rgb-d images,

    J. Shotton, B. Glocker, C. Zach, S. Izadi, A. Criminisi, and A. Fitzgib- bon, “Scene coordinate regression forests for camera relocalization in rgb-d images,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2013, pp. 2930–2937

    2013

  33. [41]

    Learning to navigate the energy landscape,

    J. Valentin, A. Dai, M. Nießner, P. Kohli, P. Torr, S. Izadi, and C. Keskin, “Learning to navigate the energy landscape,” in2016 Fourth International Conference on 3D Vision (3DV). IEEE, 2016, pp. 323– 332

    2016

  34. [42]

    An image is worth 16x16 words: Transformers for image recognition at scale,

    A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S. Gelly, J. Uszkoreit, and N. Houlsby, “An image is worth 16x16 words: Transformers for image recognition at scale,” ICLR, 2021

    2021

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.