pith. sign in

arxiv: 2606.27729 · v1 · pith:FHL4N2O6new · submitted 2026-06-26 · 💻 cs.CV

Learning 1-Bit LiDAR-based Localization with Auxiliary Objective

Kaijie Yin , Zhiyuan Zhang , Tian Gao , Wentao Zhu , Cheng-zhong Xu , Hui Kong This is my paper

Pith reviewed 2026-06-29 04:40 UTC · model grok-4.3

classification 💻 cs.CV
keywords binary neural networksLiDAR localization6-DoF pose estimationinformation bottleneckautonomous systemsmodel compression
0
0 comments X

The pith

An auxiliary objective lets binary neural networks achieve accurate 6-DoF LiDAR localization.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces BiLoc, the first binary neural network for 6-DoF LiDAR-based localization. It reinterprets BNN training using the information-bottleneck principle to keep only minimal sufficient representations for pose estimation. The key addition is an auxiliary objective that adaptively regulates how much information the binary encoder retains, which reduces the severe loss from 1-bit compression and supplies extra training signals to overcome limited capacity and gradient mismatch. Experiments on large outdoor datasets show this approach sets a new state of the art for binary LiDAR localization. A reader would care because it makes always-on localization feasible on resource-limited autonomous systems.

Core claim

BiLoc establishes the first binary neural network framework for 6-DoF LiDAR localization by introducing an auxiliary objective that adaptively regulates information retention in the binary encoder according to the information-bottleneck principle, thereby mitigating the performance drop from binarization.

What carries the argument

The auxiliary objective, which provides additional optimization signals to compensate for the limited representational capacity and gradient mismatch in BNNs.

If this is right

  • Localization can run with a much smaller portion of on-board compute resources.
  • Binary networks become competitive with full-precision methods for outdoor pose estimation.
  • Information loss from binarization can be mitigated without changing the network architecture.
  • Autonomous systems can maintain continuous localization without heavy hardware.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Similar auxiliary objectives could improve binary networks in other vision tasks like detection or segmentation.
  • The information-bottleneck reinterpretation might guide binarization in other sensor modalities.
  • Deploying such models could enable localization on low-power embedded devices for robotics.

Load-bearing premise

The auxiliary objective can effectively compensate for the information loss and optimization challenges caused by binarizing the network for this localization task.

What would settle it

If experiments without the auxiliary objective show only minor or no improvement over standard BNN training on the same datasets, the claim that it mitigates the loss would not hold.

Figures

Figures reproduced from arXiv: 2606.27729 by Cheng-zhong Xu, Hui Kong, Kaijie Yin, Tian Gao, Wentao Zhu, Zhiyuan Zhang.

Figure 1
Figure 1. Figure 1: (a) Visualization of the information loss of real-valued ViT-S/16 and ReActNet￾based ViT-S/16 on the Oxford Radar RobotCar dataset [1] at the first, middle, and last blocks [12]. Darker colors indicate less discarded information, showing higher impor￾tance for LiDAR localization. (b) Comparison of localization accuracy of full-precision ViT-S/16 and ReActNet-based ViT-S/16 on the 18-14-14-42 test split of … view at source ↗
Figure 2
Figure 2. Figure 2: BiLoc framework: The raw LiDAR point cloud is first projected into a range image that preserves geometric and depth information. The range image is fed into a binarized backbone to extract global features. These features are then passed to a fully binarized PoseDiffusion [62] decoder to predict the global pose. During training, the backbone features are also supervised by an auxiliary objective, which prov… view at source ↗
Figure 3
Figure 3. Figure 3: The inner product of two binary vectors can be efficiently computed using XNOR and PopCount. After the XNOR operation, let p be the number of matched bits (PopCount result) and u be the number of unmatched bits, where n = p+u is the vector length. Since u = n − p, the dot product can be expressed as p − (n − p). 3.2 Information-bottleneck Theory Information-bottleneck (IB) theory provides a unified framewo… view at source ↗
Figure 4
Figure 4. Figure 4: LiDAR localization results on the Oxford dataset (18-14-14-42 subset). Ground truth and predictions are shown in black and red, respectively, and the star marks the first frame. Each subfigure reports the mean position and orientation errors [PITH_FULL_IMAGE:figures/full_fig_p013_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: LiDAR localization results on the NCLT dataset (2012-03-31 subset). Ground truth and predictions are shown in black and red, respectively, and the star marks the first frame. Each subfigure reports the mean position and orientation errors. 0.00 0.40 0.80 1.20 value of 1 0 0.05 0.1 v alu e o f 2 5.86 5.41 5.27 5.40 5.54 5.18 4.89 5.12 5.65 5.26 5.03 5.21 4.5 5.0 5.5 6.0 [PITH_FULL_IMAGE:figures/full_fig_p0… view at source ↗
Figure 6
Figure 6. Figure 6: The hyper-parameters λ1 and λ2 in our BiLoc framework are selected through empirical evaluation on the 18-14-14-42 subset of the Oxford dataset. head. Due to the limited range of 1-bit operators supported by TC-BNN, the achieved inference speedup does not reach the theoretical upper bound. We ex￾pect that latency can be further reduced with the development of hardware accelerators specifically designed for… view at source ↗
read the original abstract

6-DoF LiDAR-based localization is a fundamental capability for autonomous systems operating in large-scale outdoor environments. Many deep-learning-based localization methods have achieved promising performance so far. However, as one of the always-on modules competing for limited on-board computational resources, the localization module is expected to consume only a small portion of the overall compute budget. Most existing learning-based methods are still too heavy for this purpose. In contrast, binary neural networks (BNNs) offer an appealing solution, but the 1-bit compression causes severe information loss and performance drop. In this paper, we address this challenge by proposing Binarized LiDAR-based Localization (BiLoc), the first binary neural network framework for 6-DoF LiDAR localization. Specifically, we reinterpret the training of BNNs from the perspective of the information-bottleneck principle, aiming at retaining minimal yet sufficient representations for pose estimation while suppressing redundant variations. And we introduce an auxiliary objective that adaptively regulates information retention in the binary encoder, effectively mitigating the information loss caused by binarization. This auxiliary objective provides additional optimization signals that compensate for the limited representational capacity and the gradient mismatch inherent in BNNs. Extensive experiments on large-scale outdoor LiDAR datasets demonstrate that BiLoc establishes a new state of the art for LiDAR localization with BNNs.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 0 minor

Summary. The manuscript proposes BiLoc, the first binary neural network (BNN) framework for 6-DoF LiDAR-based localization. It reinterprets BNN training through the information-bottleneck principle to retain minimal yet sufficient representations for pose estimation, and introduces an auxiliary objective that adaptively regulates information retention in the binary encoder. This objective is claimed to supply additional optimization signals that compensate for limited representational capacity and gradient mismatch in BNNs, thereby mitigating binarization-induced information loss. The paper asserts that extensive experiments on large-scale outdoor LiDAR datasets establish a new state of the art for LiDAR localization with BNNs.

Significance. If the experimental claims hold with rigorous validation, the work would be significant for enabling low-compute, always-on localization in autonomous systems. The information-bottleneck reinterpretation and auxiliary objective represent a targeted attempt to address known BNN limitations in a robotics context, with potential for broader application to other 1-bit perception tasks.

major comments (3)
  1. [Abstract] Abstract: The central claim that 'extensive experiments demonstrate that BiLoc establishes a new state of the art' is unsupported by any reported metrics, baselines, error bars, dataset details, or ablation results, rendering the effectiveness of the auxiliary objective unevaluable.
  2. [Abstract] Abstract, paragraph on the auxiliary objective: No loss formulation, weighting hyperparameter schedule, or derivation is provided showing how the auxiliary objective alters gradient flow through the binarization operator or compensates for the straight-through estimator mismatch; this is load-bearing for the claim that it mitigates information loss.
  3. [Abstract] Abstract: The assertion that the auxiliary objective 'provides additional optimization signals that compensate for the limited representational capacity and the gradient mismatch inherent in BNNs' lacks any supporting analysis of training dynamics or gradient statistics, which is required to substantiate the information-bottleneck reinterpretation.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive feedback focused on the abstract. We will revise the abstract to include key quantitative results supporting the SOTA claim. Detailed formulations, derivations, and analyses are already present in the main text and will not be duplicated in the abstract due to length constraints.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The central claim that 'extensive experiments demonstrate that BiLoc establishes a new state of the art' is unsupported by any reported metrics, baselines, error bars, dataset details, or ablation results, rendering the effectiveness of the auxiliary objective unevaluable.

    Authors: The abstract is a concise summary and omits specific numbers for brevity. The full manuscript (Sections 4 and 5) reports all requested details: metrics with error bars, baselines, dataset specifications, and ablations isolating the auxiliary objective. We will revise the abstract to incorporate 1-2 key quantitative results (e.g., translation/rotation errors on the primary datasets) to make the SOTA claim self-contained. revision: yes

  2. Referee: [Abstract] Abstract, paragraph on the auxiliary objective: No loss formulation, weighting hyperparameter schedule, or derivation is provided showing how the auxiliary objective alters gradient flow through the binarization operator or compensates for the straight-through estimator mismatch; this is load-bearing for the claim that it mitigates information loss.

    Authors: The abstract intentionally remains at a conceptual level. The explicit loss formulation, adaptive weighting schedule, and derivation linking the auxiliary term to gradient flow through the binarization operator (including compensation for straight-through estimator mismatch) appear in Section 3.2, grounded in the information-bottleneck reinterpretation. Standard practice places such technical detail in the body rather than the abstract. revision: no

  3. Referee: [Abstract] Abstract: The assertion that the auxiliary objective 'provides additional optimization signals that compensate for the limited representational capacity and the gradient mismatch inherent in BNNs' lacks any supporting analysis of training dynamics or gradient statistics, which is required to substantiate the information-bottleneck reinterpretation.

    Authors: The supporting analysis of training dynamics, gradient statistics, and information retention is contained in the experimental results and appendix of the full manuscript. The abstract states the high-level claim on the basis of those analyses; we do not believe the abstract itself must reproduce the gradient histograms or dynamics plots. revision: no

Circularity Check

0 steps flagged

No circularity: empirical auxiliary objective with no self-referential derivation chain

full rationale

The paper presents BiLoc as an empirical framework that adds an auxiliary training objective to standard BNN training for LiDAR pose estimation. The abstract describes reinterpreting BNN training via the information-bottleneck principle and introducing an auxiliary loss to regulate information retention, but supplies no equations, loss formulations, or derivation steps that reduce the claimed performance gain to a fitted parameter, self-citation, or input by construction. No uniqueness theorems, ansatzes smuggled via prior self-work, or renamed empirical patterns are invoked in the provided text. The approach is framed as a practical training modification whose validity rests on experimental results rather than a closed mathematical loop. This is the normal case of a non-circular empirical contribution.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claim rests on the effectiveness of the auxiliary objective in compensating for binarization limits; this is introduced without independent evidence beyond the claim of SOTA results.

free parameters (1)
  • auxiliary objective weighting hyperparameter
    The auxiliary objective is described as adaptively regulating retention, implying at least one balancing hyperparameter whose value is not stated.
axioms (1)
  • domain assumption The information-bottleneck principle can be directly reinterpreted as a training objective for binary networks in pose estimation
    The abstract states the training of BNNs is reinterpreted from this perspective to retain minimal yet sufficient representations.

pith-pipeline@v0.9.1-grok · 5778 in / 1239 out tokens · 32587 ms · 2026-06-29T04:40:48.591303+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

80 extracted references · 6 canonical work pages · 6 internal anchors

  1. [1]

    In: IEEE in- ternational conference on robotics and automation

    Barnes, D., Gadd, M., Murcutt, P., Newman, P., Posner, I.: The oxford radar robotcar dataset: A radar extension to the oxford robotcar dataset. In: IEEE in- ternational conference on robotics and automation. pp. 6433–6438. IEEE (2020)

    2020

  2. [2]

    In: International Conference on Machine Learning

    Belilovsky, E., Eickenberg, M., Oyallon, E.: Decoupled greedy learning of cnns. In: International Conference on Machine Learning. pp. 736–745. PMLR (2020)

    2020

  3. [3]

    Estimating or Propagating Gradients Through Stochastic Neurons for Conditional Computation

    Bengio, Y., Léonard, N., Courville, A.C.: Estimating or propagating gradients through stochastic neurons for conditional computation. ArXivabs/1308.3432 (2013)

    work page internal anchor Pith review Pith/arXiv arXiv 2013

  4. [4]

    Berger, T.: Rate Distortion Theory and Data Compression, pp. 1–39. Springer Vienna, Vienna (1975)

    1975

  5. [5]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

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

    2023

  6. [6]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Brachmann, E., Krull, A., Nowozin, S., Shotton, J., Michel, F., Gumhold, S., Rother, C.: Dsac-differentiable ransac for camera localization. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 6684–6692 (2017)

    2017

  7. [7]

    The International Journal of Robotics Research35(9), 1023–1035 (2016)

    Carlevaris-Bianco, N., Ushani, A.K., Eustice, R.M.: University of michigan north campus long-term vision and lidar dataset. The International Journal of Robotics Research35(9), 1023–1035 (2016)

    2016

  8. [8]

    In: Proceedings of the IEEE/CVF conference on com- puter vision and pattern recognition

    Chen, L., Wang, D., Gan, Z., Liu, J., Henao, R., Carin, L.: Wasserstein contrastive representation distillation. In: Proceedings of the IEEE/CVF conference on com- puter vision and pattern recognition. pp. 16296–16305 (2021)

    2021

  9. [9]

    In: Proceedings of the European Conference on Computer Vision (2022)

    Chen, S., Li, X., Wang, Z., Prisacariu, V.: Dfnet: Enhance absolute pose regres- sion with direct feature matching. In: Proceedings of the European Conference on Computer Vision (2022)

    2022

  10. [10]

    Advances in Neural Information Processing Systems37, 30651–30669 (2024)

    Chen, Z., Qin, H., Guo, Y., Su, X., Yuan, X., Kong, L., Zhang, Y.: Binarized diffu- sion model for image super-resolution. Advances in Neural Information Processing Systems37, 30651–30669 (2024)

    2024

  11. [11]

    IEEE transac- tions on pattern analysis and machine intelligence45(10), 12358–12376 (2023)

    Chen, Z., Sun, K., Yang, F., Guo, L., Tao, W.: Sc2-pcr++: Rethinking the gener- ation and selection for efficient and robust point cloud registration. IEEE transac- tions on pattern analysis and machine intelligence45(10), 12358–12376 (2023)

    2023

  12. [12]

    In: Proceedings of the IEEE/CVF conference on com- puter vision and pattern recognition

    Cheng, X., Rao, Z., Chen, Y., Zhang, Q.: Explaining knowledge distillation by quantifying the knowledge. In: Proceedings of the IEEE/CVF conference on com- puter vision and pattern recognition. pp. 12925–12935 (2020)

    2020

  13. [13]

    cuDNN: Efficient Primitives for Deep Learning

    Chetlur, S., Woolley, C., Vandermersch, P., Cohen, J., Tran, J., Catanzaro, B., Shelhamer, E.: cudnn: Efficient primitives for deep learning. arXiv preprint arXiv:1410.0759 (2014)

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  14. [14]

    In: Proceed- ings of the IEEE/CVF international conference on computer vision

    Cho, J.H., Hariharan, B.: On the efficacy of knowledge distillation. In: Proceed- ings of the IEEE/CVF international conference on computer vision. pp. 4794–4802 (2019) 16 Yin et al

    2019

  15. [15]

    Ad- vances in neural information processing systems26(2013)

    Cuturi, M.: Sinkhorn distances: Lightspeed computation of optimal transport. Ad- vances in neural information processing systems26(2013)

    2013

  16. [16]

    In: International Confer- ence on Machine Learning

    Deng, S., Lin, H., Li, Y., Gu, S.: Surrogate module learning: Reduce the gradient error accumulation in training spiking neural networks. In: International Confer- ence on Machine Learning. pp. 7645–7657. PMLR (2023)

    2023

  17. [17]

    In: International Conference on Learning Representations (2021)

    Dosovitskiy, A.: An image is worth 16x16 words: Transformers for image recogni- tion at scale. In: International Conference on Learning Representations (2021)

    2021

  18. [18]

    IEEE Computational Intelligence Magazine17(4), 39–51 (2022)

    Duan, S., Principe, J.C.: Training deep architectures without end-to-end back- propagation: A survey on the provably optimal methods. IEEE Computational Intelligence Magazine17(4), 39–51 (2022)

    2022

  19. [19]

    Fischler, M.A., Bolles, R.C.: Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography. Commun. ACM24, 381–395 (1981)

    1981

  20. [20]

    In: IEEE International Conference on Robotics and Automa- tion

    Frickenstein, A., Vemparala, M.R., Mayr, J., Nagaraja, N.S., Unger, C., Tombari, F., Stechele, W.: Binary dad-net: Binarized driveable area detection network for autonomous driving. In: IEEE International Conference on Robotics and Automa- tion. pp. 2295–2301 (2020)

    2020

  21. [21]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Gao, T., Zhang, Y., Zhang, Z., Liu, H., Yin, K., Xu, C., Kong, H.: Bhvit: Bina- rized hybrid vision transformer. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 3563–3572 (2025)

    2025

  22. [22]

    Gibson, J.D.: Rate Distortion Theory, pp. 77–92. Springer Nature Switzerland, Cham (2025)

    2025

  23. [23]

    IEEE Robotics and Automation Letters9(6), 5070–5077 (2024)

    Grainge, O., Milford, M., Bodala, I., Ramchurn, S.D., Ehsan, S.: Design space exploration of low-bit quantized neural networks for visual place recognition. IEEE Robotics and Automation Letters9(6), 5070–5077 (2024)

    2024

  24. [24]

    IEEE Robotics and Automation Letters (2024)

    Grainge, O., Milford, M., Bodala, I., Ramchurn, S.D., Ehsan, S.: Structured prun- ing for efficient visual place recognition. IEEE Robotics and Automation Letters (2024)

    2024

  25. [25]

    He, Y., Lou, Z., Zhang, L., Liu, J., Wu, W., Zhou, H., Zhuang, B.: Bivit: Extremely compressedbinaryvisiontransformers.In:IEEE/CVFInternationalConferenceon Computer Vision. pp. 5651–5663 (2023)

    2023

  26. [26]

    In: Proceedings of the IEEE international conference on computer vision

    He, Y., Zhang, X., Sun, J.: Channel pruning for accelerating very deep neural networks. In: Proceedings of the IEEE international conference on computer vision. pp. 1389–1397 (2017)

    2017

  27. [27]

    Distilling the Knowledge in a Neural Network

    Hinton, G., Vinyals, O., Dean, J.: Distilling the knowledge in a neural network. arXiv preprint arXiv:1503.02531 (2015)

    work page internal anchor Pith review Pith/arXiv arXiv 2015

  28. [28]

    Advances in neural information processing systems29(2016)

    Hubara,I.,Courbariaux,M.,Soudry,D.,El-Yaniv,R.,Bengio,Y.:Binarizedneural networks. Advances in neural information processing systems29(2016)

    2016

  29. [29]

    In: Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition

    Jacob, B., Kligys, S., Chen, B., Zhu, M., Tang, M., Howard, A., Adam, H., Kalenichenko,D.:Quantizationandtrainingofneuralnetworksforefficientinteger- arithmetic-only inference. In: Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition. pp. 2704–2713 (2018)

    2018

  30. [30]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

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

    2017

  31. [31]

    In: Proceedings of the IEEE international con- ference on computer vision

    Kendall, A., Grimes, M., Cipolla, R.: Posenet: A convolutional network for real- time 6-dof camera relocalization. In: Proceedings of the IEEE international con- ference on computer vision. pp. 2938–2946 (2015)

    2015

  32. [32]

    In: Proceedings of the IEEE/CVF winter conference on applications of computer vi- sion

    Komorowski, J.: Minkloc3d: Point cloud based large-scale place recognition. In: Proceedings of the IEEE/CVF winter conference on applications of computer vi- sion. pp. 1790–1799 (2021) BiLoc 17

    2021

  33. [33]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops

    Le, P.H.C., Li, X.: Binaryvit: Pushing binary vision transformers towards convolu- tional models. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops. pp. 4665–4674 (June 2023)

    2023

  34. [34]

    IEEE Transactions on Parallel and Distributed Systems32(7), 1878–1891 (2020)

    Li, A., Su, S.: Accelerating binarized neural networks via bit-tensor-cores in turing gpus. IEEE Transactions on Parallel and Distributed Systems32(7), 1878–1891 (2020)

    2020

  35. [35]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Li, W., Liu, C., Yu, S., Liu, D., Zhou, Y., Shen, S., Wen, C., Wang, C.: Light- loc: Learning outdoor lidar localization at light speed. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 6680– 6689 (2025)

    2025

  36. [36]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Li, W., Yang, Y., Yu, S., Hu, G., Wen, C., Cheng, M., Wang, C.: Diffloc: Diffusion model for outdoor lidar localization. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 15045–15054 (2024)

    2024

  37. [37]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Li,W.,Yu,S.,Wang,C.,Hu,G.,Shen,S.,Wen,C.:Sgloc:Scenegeometryencoding for outdoor lidar localization. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 9286–9295 (2023)

    2023

  38. [38]

    In: AAAI Conference on Artificial Intelligence

    Li, Y., Xu, S., Lin, M., Cao, X., Liu, C., Sun, X., Zhang, B.: Bi-vit: Pushing the limit of vision transformer quantization. In: AAAI Conference on Artificial Intelligence. vol. 38, pp. 3243–3251 (2024)

    2024

  39. [39]

    In: International conference on machine learning

    Liu, S.Y., Liu, Z., Cheng, K.T.: Oscillation-free quantization for low-bit vision transformers. In: International conference on machine learning. pp. 21813–21824. PMLR (2023)

    2023

  40. [40]

    In: European conference on computer vision

    Liu, Z., Shen, Z., Savvides, M., Cheng, K.T.: Reactnet: Towards precise binary neural network with generalized activation functions. In: European conference on computer vision. pp. 143–159. Springer (2020)

    2020

  41. [41]

    In: Proceedings of the European conference on com- puter vision

    Liu, Z., Wu, B., Luo, W., Yang, X., Liu, W., Cheng, K.T.: Bi-real net: Enhanc- ing the performance of 1-bit cnns with improved representational capability and advanced training algorithm. In: Proceedings of the European conference on com- puter vision. pp. 722–737 (2018)

    2018

  42. [42]

    IEEE Transactions on Robotics (2025)

    Luo, L., Cao, S.Y., Li, X., Xu, J., Ai, R., Yu, Z., Chen, X.: Bevplace++: Fast, robust, and lightweight lidar global localization for unmanned ground vehicles. IEEE Transactions on Robotics (2025)

    2025

  43. [43]

    In: Conference on Robot Learning

    Moreau, A., Piasco, N., Tsishkou, D., Stanciulescu, B., de La Fortelle, A.: Lens: Localization enhanced by nerf synthesis. In: Conference on Robot Learning. pp. 1347–1356. PMLR (2022)

    2022

  44. [44]

    DINOv2: Learning Robust Visual Features without Supervision

    Oquab, M., Darcet, T., Moutakanni, T., Vo, H., Szafraniec, M., Khalidov, V., Fernandez, P., Haziza, D., Massa, F., El-Nouby, A., et al.: Dinov2: Learning robust visual features without supervision. arXiv preprint arXiv:2304.07193 (2023)

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  45. [45]

    In: Proceed- ings of the IEEE/CVF conference on computer vision and pattern recognition

    Qin, H., Gong, R., Liu, X., Shen, M., Wei, Z., Yu, F., Song, J.: Forward and backward information retention for accurate binary neural networks. In: Proceed- ings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 2250–2259 (2020)

    2020

  46. [46]

    In: European conference on computer vision

    Rastegari, M., Ordonez, V., Redmon, J., Farhadi, A.: Xnor-net: Imagenet classi- fication using binary convolutional neural networks. In: European conference on computer vision. pp. 525–542. Springer (2016)

    2016

  47. [47]

    In: Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition

    Sandler, M., Howard, A., Zhu, M., Zhmoginov, A., Chen, L.C.: Mobilenetv2: In- verted residuals and linear bottlenecks. In: Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition. pp. 4510–4520 (2018)

    2018

  48. [48]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Sarlin, P.E., DeTone, D., Malisiewicz, T., Rabinovich, A.: Superglue: Learning feature matching with graph neural networks. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 4938–4947 (2020) 18 Yin et al

    2020

  49. [49]

    IEEE transactions on pattern analysis and machine intelligence 45(12), 14222–14233 (2023)

    Shavit, Y., Ferens, R., Keller, Y.: Coarse-to-fine multi-scene pose regression with transformers. IEEE transactions on pattern analysis and machine intelligence 45(12), 14222–14233 (2023)

    2023

  50. [50]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Shen, Z., Liu, Z., Qin, J., Huang, L., Cheng, K.T., Savvides, M.: S2-bnn: Bridging the gap between self-supervised real and 1-bit neural networks via guided distribu- tion calibration. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 2165–2174 (2021)

    2021

  51. [51]

    ISPRS journal of photogrammetry and remote sensing184, 177–188 (2022)

    Shi, C., Li, J., Gong, J., Yang, B., Zhang, G.: An improved lightweight deep neural network with knowledge distillation for local feature extraction and visual localiza- tion using images and lidar point clouds. ISPRS journal of photogrammetry and remote sensing184, 177–188 (2022)

    2022

  52. [52]

    Opening the Black Box of Deep Neural Networks via Information

    Shwartz-Ziv, R., Tishby, N.: Opening the black box of deep neural networks via information. arXiv preprint arXiv:1703.00810 (2017)

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  53. [53]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Sun, P., Kretzschmar, H., Dotiwalla, X., Chouard, A., Patnaik, V., Tsui, P., Guo, J., Zhou, Y., Chai, Y., Caine, B., et al.: Scalability in perception for autonomous driving: Waymo open dataset. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 2446–2454 (2020)

    2020

  54. [54]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

    Szegedy, C., Liu, W., Jia, Y., Sermanet, P., Reed, S., Anguelov, D., Erhan, D., Vanhoucke, V., Rabinovich, A.: Going deeper with convolutions. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 1–9 (2015)

    2015

  55. [55]

    The information bottleneck method

    Tishby, N., Pereira, F.C., Bialek, W.: The information bottleneck method. arXiv preprint physics/0004057 (2000)

    work page internal anchor Pith review Pith/arXiv arXiv 2000

  56. [56]

    In: IEEE Information Theory Workshop

    Tishby, N., Zaslavsky, N.: Deep learning and the information bottleneck principle. In: IEEE Information Theory Workshop. pp. 1–5 (2015)

    2015

  57. [57]

    In: International conference on machine learning

    Touvron, H., Cord, M., Douze, M., Massa, F., Sablayrolles, A., Jégou, H.: Training data-efficient image transformers & distillation through attention. In: International conference on machine learning. pp. 10347–10357. PMLR (2021)

    2021

  58. [58]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

    Uy, M.A., Lee, G.H.: Pointnetvlad: Deep point cloud based retrieval for large-scale place recognition. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 4470–4479 (2018)

    2018

  59. [59]

    Victor, T., Kusano, K., Gode, T., Chen, R., Schwall, M.: Safety performance of the waymo rider-only automated driving system at one million miles. Tech. Rep. (2023)

    2023

  60. [60]

    Villani, C., et al.: Optimal transport: old and new, vol. 338. Springer (2008)

    2008

  61. [61]

    In: Proceedings of the AAAI Conference on Artificial Intelligence

    Wang, B., Chen, C., Lu, C.X., Zhao, P., Trigoni, N., Markham, A.: Atloc: Attention guided camera localization. In: Proceedings of the AAAI Conference on Artificial Intelligence. pp. 10393–10401 (2020)

    2020

  62. [62]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Wang, J., Rupprecht, C., Novotny, D.: Posediffusion: Solving pose estimation via diffusion-aided bundle adjustment. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 9773–9783 (2023)

    2023

  63. [63]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Wang, S., Kang, Q., She, R., Wang, W., Zhao, K., Song, Y., Tay, W.P.: Hypliloc: Towards effective lidar pose regression with hyperbolic fusion. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 5176–5185 (2023)

    2023

  64. [64]

    IEEE Sensors Journal22(1), 959–968 (2021)

    Wang, W., Wang, B., Zhao, P., Chen, C., Clark, R., Yang, B., Markham, A., Trigoni, N.: Pointloc: Deep pose regressor for lidar point cloud localization. IEEE Sensors Journal22(1), 959–968 (2021)

    2021

  65. [65]

    In: Proceedings of the IEEE/CVF international conference on computer vision

    Wang, Y., Solomon, J.M.: Deep closest point: Learning representations for point cloud registration. In: Proceedings of the IEEE/CVF international conference on computer vision. pp. 3523–3532 (2019) BiLoc 19

    2019

  66. [66]

    International Journal of Computer Vision133(5), 2752–2782 (2025)

    Wang, Y., Ni, Z., Pu, Y., Zhou, C., Ying, J., Song, S., Huang, G.: Infopro: Locally supervised deep learning by maximizing information propagation. International Journal of Computer Vision133(5), 2752–2782 (2025)

    2025

  67. [67]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recog- nition pp

    Wang, Z., Wu, Z., Lu, J., Zhou, J.: Bidet: An efficient binarized object detector. Proceedings of the IEEE/CVF conference on computer vision and pattern recog- nition pp. 2046–2055 (2020)

    2046

  68. [68]

    In: 2015 IEEE international conference on robotics and automation (ICRA)

    Wolcott,R.W.,Eustice,R.M.:Fastlidarlocalizationusingmultiresolutiongaussian mixture maps. In: 2015 IEEE international conference on robotics and automation (ICRA). pp. 2814–2821. IEEE (2015)

    2015

  69. [69]

    In: European Conference on Computer Vision

    Xu, S., Li, Y., Zeng, B., Ma, T., Zhang, B., Cao, X., Gao, P., Lü, J.: Ida-det: An information discrepancy-aware distillation for 1-bit detectors. In: European Conference on Computer Vision. pp. 346–361. Springer (2022)

    2022

  70. [70]

    In: Forty-first International Conference on Machine Learning (2024)

    Xu, S., Wang, M., Li, Y., Lin, M., Zhang, B., Doermann, D., Sun, X.: Learning 1-bit tiny object detector with discriminative feature refinement. In: Forty-first International Conference on Machine Learning (2024)

    2024

  71. [71]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Yang, B., Li, Z., Li, W., Cai, Z., Wen, C., Zang, Y., Muller, M., Wang, C.: Lisa: Lidar localization with semantic awareness. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 15271–15280 (2024)

    2024

  72. [72]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Yang, Y., Li, W., Ao, S., Xu, Q., Yu, S., Guo, Y., Zhou, Y., Shen, S., Wang, C.: Raloc: Enhancing outdoor lidar localization via rotation awareness. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 3304–3313 (2025)

    2025

  73. [73]

    International Journal of Computer Vision132, 3139 – 3171 (2024)

    Yin, H., Xu, X., Lu, S., Chen, X., Xiong, R., Shen, S., Stachniss, C., Wang, Y.: A survey on global lidar localization: Challenges, advances and open problems. International Journal of Computer Vision132, 3139 – 3171 (2024)

    2024

  74. [74]

    In: IEEE/RSJ International Conference on Intelligent Robots and Systems

    Yin, K., Gao, T., Kong, H.: Pathfinder for low-altitude aircraft with binary neu- ral network. In: IEEE/RSJ International Conference on Intelligent Robots and Systems. pp. 20692–20699 (2025)

    2025

  75. [75]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Yin, K., Zhang, Z., Kong, S., Gao, T., Xu, C.Z., Kong, H.: Information- bottleneck driven binary neural network for change detection. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 7176–7186 (2025)

    2025

  76. [76]

    In: Proceedings of the 32nd ACM International Conference on Multimedia

    Yin, P., Zhu, X., Song, J., Gao, L., Shen, H.T.: Si-bivit: Binarizing vision trans- formers with spatial interaction. In: Proceedings of the 32nd ACM International Conference on Multimedia. pp. 8169–8178 (2024)

    2024

  77. [77]

    IEEE Transactions on Intelligent Trans- portation Systems24(1), 489–500 (2022)

    Yu, S., Wang, C., Lin, Y., Wen, C., Cheng, M., Hu, G.: Stcloc: Deep lidar local- ization with spatio-temporal constraints. IEEE Transactions on Intelligent Trans- portation Systems24(1), 489–500 (2022)

    2022

  78. [78]

    Pattern Recog- nition128, 108685 (2022)

    Yu, S., Wang, C., Wen, C., Cheng, M., Liu, M., Zhang, Z., Li, X.: Lidar-based lo- calization using universal encoding and memory-aware regression. Pattern Recog- nition128, 108685 (2022)

    2022

  79. [79]

    In: Proceedings of the IEEE/CVF international conference on computer vision

    Zhang, L., Song, J., Gao, A., Chen, J., Bao, C., Ma, K.: Be your own teacher: Improve the performance of convolutional neural networks via self distillation. In: Proceedings of the IEEE/CVF international conference on computer vision. pp. 3713–3722 (2019)

    2019

  80. [80]

    IEEE Transactions on Pattern Analysis and Machine Intelligence45(4), 5099–5113 (2022)

    Zhang, Q., Cheng, X., Chen, Y., Rao, Z.: Quantifying the knowledge in a dnn to explain knowledge distillation for classification. IEEE Transactions on Pattern Analysis and Machine Intelligence45(4), 5099–5113 (2022)

    2022