The DOI matches its Crossref record; 30 of the 30 references it cites were found in a registry or an index.
Modelling uncertainty in deep learning for camera relocalization
Open accessAbstract
We present a robust and real-time monocular six degree of freedom visual relocalization system. We use a Bayesian convolutional neural network to regress the 6-DOF camera pose from a single RGB image. It is trained in an end-to-end manner with no need of additional engineering or graph optimisation. The algorithm can operate indoors and outdoors in real time, taking under 6ms to compute. It obtains approximately 2m and 6° accuracy for very large scale outdoor scenes and 0.5m and 10° accuracy indoors. Using a Bayesian convolutional neural network implementation we obtain an estimate of the model's relocalization uncertainty and improve state of the art localization accuracy on a large scale outdoor dataset. We leverage the uncertainty measure to estimate metric relocalization error and to detect the presence or absence of the scene in the input image. We show that the model's uncertainty is caused by images being dissimilar to the training dataset in either pose or appearance.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“We adopted the ResNet architectures with pre-trained weights as the network backbone, due to its low number of pooling layers, and good accuracy-complexity trade-off [35].”
From Deep Learning for Spacecraft Pose Estimation from Photorealistic Rendering (2020) · cites this paper as [35]
Citations
Cites
ImageNet classification with deep convolutional neural networks
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017) · Communications of the ACM · doi:10.1145/3065386 · Cited by 112,155 · cited 1×
“Convolutional neural networks have recently been shown to be extremely powerful for generating image representations for classification [12], [13].”
BackgroundDistinctive Image Features from Scale-Invariant Keypoints
Lowe, D. G. (2004) · International Journal of Computer Vision · doi:10.1023/b:visi.0000029664.99615.94 · Cited by 55,589 · cited 3×
“This has been achieved using SIFT [2] features with FAB-MAP [6] or using convolutional neural network features [7].”
BackgroundMethodGoing deeper with convolutions
Szegedy, C., Liu, W., Jia, Y., Sermanet, P., Reed, S., Anguelov, D., Erhan, D., Vanhoucke, V., & Rabinovich, A. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298594 · Cited by 48,562 · cited 4×
“Additionally, in green we show the output from the first auxiliary pose regressor from the GoogLeNet architecture (see Figure 3 of [13]).”
BackgroundMethodInfluential(2014) · Cited by 33,419
Visualizing and understanding convolutional networks
Zeiler, M. D., & Fergus, R. (2014) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-10590-1_53 · Cited by 15,839 · cited 1×
“These higher level contours and shape cues [14] have been shown to be increasingly invariant to object instantiation parameters.”
BackgroundCaffe: Convolutional Architecture for Fast Feature Embedding
Jia, Y., Shelhamer, E., Donahue, J., Karayev, S., Long, J., Girshick, R., Guadarrama, S., & Darrell, T. (2014) · Proceedings of the 22nd ACM international conference on Multimedia · doi:10.1145/2647868.2654889 · Cited by 11,077 · cited 1×
“Using standard libraries such as [25] we can obtain an efficient implementation of a Bernoulli Bayesian convolutional neural network.”
MethodORB: An efficient alternative to SIFT or SURF
Rublee, E., Rabaud, V., Konolige, K., & Bradski, G. (2011) · 2011 International Conference on Computer Vision · doi:10.1109/iccv.2011.6126544 · Cited by 11,037 · cited 2×
“Visual SLAM [1], [4], [5], [9], and relocalization research [10], [11], has largely focused on registering viewpoints using point-based landmark features, such as SIFT [2] or ORB [3].”
BackgroundMethodThrun, S. (2002) · CACM · doi:10.1145/504729.504754 · Cited by 7,941 · cited 1×
“nsider localization uncertainty [1], [5], [11], previous probabilistic algorithms have been proposed. Bayesian approaches include extended Kalman filters and particle filter approaches such as FastSLAM [18]. However these approaches estimate uncertainty from sensor noise models, not the uncertainty of the model to repre…”
MethodORB-SLAM: A Versatile and Accurate Monocular SLAM System
Mur-Artal, R., Montiel, J. M. M., & Tardos, J. D. (2015) · IEEE Transactions on Robotics · doi:10.1109/tro.2015.2463671 · Cited by 6,778 · cited 1×
“Visual SLAM [1], [4], [5], [9], and relocalization research [10], [11], has largely focused on registering viewpoints using point-based landmark features, such as SIFT [2] or ORB [3].”
BackgroundDropout as a bayesian approximation: Representing model uncertainty in deep learning
Gal, Y., & Ghahramani, Z. (2015) · arXiv · doi:10.48550/arxiv.1506.02142 · Cited by 6,471 · cited 1×
“Recent work [23] has extended dropout to approximate Bayesian inference over the network’s weights.”
Background
Flagged first, then by citations
Cited by
Sorted by citations
Items 501–549 of 549
Malherbe, E., Remise, M., Zhang, S., & Perrot, M. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-26284-5_1
Towards Extreme Weather Detection: Dark Channel Priori Enhanced Detection with New Labelled Data
Li, Y. (2023) · Highlights in Science Engineering and Technology · doi:10.54097/hset.v44i.7348
Alam, M. S., Mohamed, F., Selamat, A., & Hossain, A. B. (2023) · SSRN Electronic Journal · doi:10.2139/ssrn.4474209
Lamping, C., Kootstra, G., & Derks, M. (2023) · SSRN Electronic Journal · doi:10.2139/ssrn.4505301
Robust Dense Depth Estimation in Adverse Weather ConditionsPreprint
Zhang, H., Wei, H., & Huilan, L. (2023) · SSRN Electronic Journal · doi:10.2139/ssrn.4524934
Improved Bilinear Pooling for Real-Time Pose Event Camera Relocalisation
Tabia, A., Bonardi, F., & Bouchafa, S. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-43148-7_19
Understanding the Effect of Deep Ensembles in LiDAR-Based Place Recognition
Vaghi, M., D’Elia, F., Ballardini, A. L., & Sorrenti, D. G. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-47546-7_20
Progressive Temporal Transformer for Bird’s-Eye-View Camera Pose Estimation
Wu, Z., Cai, J., Huang, R., Liu, X., & Chai, Z. (2023) · Lecture notes in computer science · doi:10.1007/978-981-99-8076-5_10
Holder, C. J., & Shafique, M. (2023) · International Conference on Robotics, Automation and Artificial Intelligence (RAAI) · doi:10.1109/raai59955.2023.10601279
“2D – 2D Implicit Map Localisation, that we refer to in this work as neural mapping, estimates pose via a neural network that has learned an implicit representation of a scene from a training set of image – pose pairs [17] [18] [19].”
BackgroundSMLoc: spatial multilayer perception-guided camera localization
Feng, J., Wang, S., & Sun, H. (2024) · Journal of Electronic Imaging · doi:10.1117/1.jei.33.5.053013
“11 14, 17 To achieve a structured dimensionality reduction and choose the appropriate feature correlations for the task of pose estimation, Walch et al. 12 recommended using long-short-term long-short-term Although the above methods improve the accuracy of the pose regression in terms of model…”
Two Stage Non-Linear Pose estimation of Non-Cooperative satellites using Deep Keypoint Detection
Reddy, V. S., & Rajendran, S. (2024) · International Conference on Computing, Communication and Networking Technologies (ICCCNT) · doi:10.1109/icccnt61001.2024.10724479
Learning Active Re-Localization with Deep Reinforcement Learning
Liu, M., Sun, L., Fan, Z., & Liu, Y. (2024) · International Conference on Robotics and Computer Vision (ICRCV) · doi:10.1109/icrcv62709.2024.10758614
“In [11], Kendall and R. Cipolla show that uncertainty of neural-network-based localization can be estimated and the poses can be sampled from the measurement model.”
Leveraging GANs and depth-guided diffusion for improved 3d reconstruction and depth estimation
Laribi, S. S., Nini, B., & Abdelhak, B. (2024) · STUDIES IN ENGINEERING AND EXACT SCIENCES · doi:10.54021/seesv5n2-775
Quantifying Epistemic Uncertainty in Absolute Pose Regression
Zangeneh, F., Dekel, A., Pieropan, A., & Jensfelt, P. (2025) · Lecture notes in computer science · doi:10.1007/978-3-031-95918-9_13
Indoor Dark Light Video Positioning Algorithm Based on Backbone Structure From Motion
He, Y., Ren, Y., Yue, M., Zhang, L., Liu, C., & Li, C. (2025) · IEEE Sensors Journal · doi:10.1109/jsen.2025.3581284
A multi-level composite attention-guided network for indoor visual localization
Song, X., Yang, H., Tang, J., Li, X., Lu, X., & Hei, X. (2025) · Digital Signal Processing · doi:10.1016/j.dsp.2025.105458
6-DOF Pose Estimation For Event Cameras Using A Transformer-Based Approach
Evry, U., Tabia, A., Bonardi, F., Evry, U., Bouchaa, S., & Evry, U. (2025) · Computer Science Research Notes · doi:10.24132/csrn.2025-13
“PoseNet Table 2: Comparison between our method results and the results of PoseNet [12], Bayesian PoseNet [11] and SP-LSTM [25].”
InfluentialA Pyramid Convolution-Based Scene Coordinate Regression Network for AR-GIS
Xu, H., Zhu, C., Wang, Y., Zhu, H., & Ma, W. (2025) · ISPRS International Journal of Geo-Information · doi:10.3390/ijgi14080311
“Later extensions incorporated uncertainty estimation [47] and transformer-based architectures [48] to improve adaptability across different scenes.”
Song, Y., Zhuang, H., Jiang, J., Liu, Z., & Fang, Z. (2025) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra55743.2025.11127875
Shahzadi, T., Tarashima, S., Tagawa, N., & Bokuda, K. (2025) · IEEE International Ultrasonics Symposium · doi:10.1109/ius62464.2025.11201500
“The use of point estimation to optimize non-linear big rotations might lead to incorrect motion estimation in textureless environments [5], [6].”
Fülöp, T., & Katona, J. (2025) · Innovations in Systems and Software Engineering · doi:10.1007/s11334-025-00621-5 · Cited by 1
Liu, Y., Meng, Y., Pan, X., Jiang, J., & Chen, C. (2025) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros60139.2025.11247099
“Kendall et al. [12] pioneered the use of CNNs for end-to-end pose estimation, bypassing traditional feature matching steps.”
A Review of Learning Based Visual Relocalization Methods
Lin, H., Wang, Y., Yin, B., & Yang, X. S. (2025) · Computational Visual Media · doi:10.26599/cvm.2025.9450492
Probabilistic Direct Structure from Motion via Hierarchical Reparameterization Trick
Wang, Y., & Tagawa, N. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-14495-9_6
Visual Localization Using Implicit Representations and Particle Filtering-Based Pose Refinement
Cheimariotis, G., & Zarpalas, D. (2026) · IEEE Access · doi:10.1109/access.2026.3659994
“Several APR methods are deep-learning pose regression methods [3], [5], [6], [7], [8], [10], [11], which aim to estimate 6-DoF poses directly from RGB inputs.”
SemiFlow: Website Fingerprinting Using Semi-Supervised Learning
Lv, S., Yang, Y., & Huang, Y. (2025) · IEEE International Conference on Trust, Security and Privacy in Computing and Communications (TrustCom) · doi:10.1109/trustcom66490.2025.00294
“Quantifying uncertainty in DNNs has become a crucial research direction [29].”
Semantic-Enhanced Scene Coordinate Regression Network for Visual Localization
Ma, W., Zhu, C., Dai, H., Wang, C., Zhou, B., Ge, L., Hu, B., Pan, J., & Li, Q. (2026) · IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing · doi:10.1109/jstars.2026.3663700
“Based on this, they further introduced a Bayesian deep learning framework to address uncertainties in predictions[41],aswellasgeometriclossfunctionsanddynamic weightadjustmentmechanismstoimprovetheperformance androbustnessofthenetwork[42].”
Shavit, Y., & Keller, Y. (2025) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw69036.2025.00251
“Early research explored various encoder and decoder (regressor) architectures [16, 18, 23, 30, 32], loss formulations [13, 23], and regularization techniques [12] to enhance performance and mitigate overfitting, aiming to narrow the accuracy gap with structure-based methods.”
Localization Based on End-to-End Learning Methods
Qin, J., Li, M., & Tang, S. (2026) · Springer eBooks · doi:10.1007/978-981-99-1650-4_39-1
Absolute Localization by Mono-camera for a Vehicle in Urban Area using Street View
Yu, L. (2018) · theses.fr (ABES) · doi:10.70675/8b314205zb09ez4208za110zd1ebb59a823c
Hybrid Machine Learning and Geometric Approaches for Single RGB Camera Relocalization
Duong, N.-D. (2019) · theses.fr (ABES) · doi:10.70675/8e4ecee4z215fz435az93aez8cb46cfd16fa
Ensemble Methods for Pedestrian Detection in Dense Crowds
Vandoni, J. (2019) · theses.fr (ABES) · doi:10.70675/fa92f628z9f47z4c57za333z57ccf4b6fb10
“MC-dropout has been successfully applied in different applications, from segmentation for scene understanding [134] to camera re-localization [135], allowing to model the predictive uncertainty through standard deviation of the stochastic realizations obtained with dropout.”
MethodLearning-based algorithms for real-time visual localization of vehicles
Moreau, A. (2023) · theses.fr (ABES) · doi:10.70675/bb1e6237zdce0z4b99z9164z6f369988e1db · Cited by 1
Pose estimation with event camera
Tabia, A. (2024) · theses.fr (ABES) · doi:10.70675/c474e599z63cez489az8a37z2610095ad3ea
oKAN: Orthogonal Kolmogorov-Arnold Networks for Accurate and Interpretable Camera Pose Regression
Wang, W., Liu, H., Jiang, P., Li, T., Liu, S., & Ding, R. (2026) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp55912.2026.11460472
Popescu, M. H., Roitero, K., & Mea, V. D. (2026) · JMIR AI · doi:10.2196/78764
“In addition, NNs can be incorporated into other probabilistic models by using good probability estimates [7,8].”
Semantic-Guided Camera Ray Regression for Visual Localization
Zhang, Y., & Zhao, X. (2025) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51701.2025.02378
“In the experiments, we compare our DIMM method and its variation enhanced by RANSAC (DIMM-R) with both APR [15] Table 3.”
Single-image indoor localization using cross-domain learning from BIM models
Ryszko, P., Włodarczyk, D., & Jarząbek-Rychard, M. (2026) · Reports on Geodesy and Geoinformatics · doi:10.2478/rgg-2026-0004
“…Pose Regression (APR) (Kendall et al., 2015; Kendall and Cipolla, 2016) frameworks demonstrated that deep neural networks can directly regress camera pose from a single image, with Kendall and Cipolla (2016) additionally modeling uncertainty in pose predictions using probabilistic approaches.”
Wu, F., Kacete, A., Royan, J., & Moreau, G. (2026) · International Conference on Multimedia Retrieval · doi:10.1145/3805622.3810746
“Subsequent work improved robustness and accuracy through uncertainty modeling [12, 19], scene coordinate regression [18, 59, 60], or hybrid geometric supervision [5, 6].”
Uncertainty-Aware Bidirectional Graph Learning for Camera Localization
Cai, H., Li, Z., Yan, C., Yang, F., Li, X., & Zhai, Q. (2026) · Electronics · doi:10.3390/electronics15143109
Decoupled Camera Pose Estimation based on Image Gradient for UAV Landing Navigation
Yu, Z. G., & Wang, W. (2026) · International Conference on Image and Graphics Processing (ICIGP) · doi:10.1109/icigp68997.2026.11619658
“Unlike sub-pixel accurate edges, this pixel-level noise implies uncertainty [19], which ex-acerbates the parameter ambiguity in standard PnP solvers [20].”
Xin’an, Q. I. U., Wang, L., Dong, Z., Xiao, X., Liu, Z., Zhu, L., Wang, J., & Xu, H. (2026) · Sensors · doi:10.3390/s26165070
“Subsequent APR methods explored uncertainty [29], LSTM and video context [30,31], hourglass networks [32], and lightweight decision forests [33].”
Zhang, J., Hu, Z., Pang, Y., Zhao, Y., Liu, Z., & Wan, M. (2026) · International Conference on Computer Communication and Artificial Intelligence (CCAI) · doi:10.1109/ccai69603.2026.11641924
“Furthermore, Kendall et al. [14] attempted to construct an end-to-end network to directly predict the uncertainty of pose estimation from images.”
Motion-Regime-Aware Feature Decoupling for Transformer-Based Monocular Camera Relocalization
Alqaraleh, S., & Hafez, A. H. A. (2026) · Mathematics · doi:10.3390/math14173035
Pan, X., Feng, B., Zhu, B., Liu, Y., & Liu, Q. (2026) · Aerospace · doi:10.3390/aerospace13090846
Hasan, H. M., Garcia, M. A., Rashwan, H. A., & Puig, D. (2026) · Machine Vision and Applications · doi:10.1007/s00138-026-01935-5
“…, Fire , Heads , and Office scenes Method Chess Fire Heads Office PoseNet [ 3] 0.32m 8.12° 0.47m 14.4° 0.29m 12.0° 0.48m 7.68° Bayesian PoseNet [ 20] 0.37m 7.24° 0.43m 13.7° 0.31m 12.0° 0.48m 8.04° LSTM PoseNet [ 21] 0.24m 5.77° 0.34m 11.9° 0.21m 13.7° 0.30m 8.08° BranchNet [45 ] 0.18m 5.17°…”
InfluentialLin, H., Long, C., Zhang, J., Jiang, C., Ge, H., Yin, E., Yin, B., & Yang, X. (2026) · doi:10.1109/cvprf71909.2026.00133
“Some methods utilize convolutional networks to directly regress camera poses [19, 32–34, 58 ], while others employ random forest regression [ 16 , 23”
Metadata
- Authors
- Kendall, A., & Cipolla, R.
- Affiliations
- University of Cambridge (GB)
- ORCID
- Cipolla, Roberto
- Venue
- 2016 IEEE International Conference on Robotics and Automation (ICRA)
- Publisher
- IEEE
- Published
- 2016-05
- Pages
- 4762-4769
- ISSN
- 1050-4729 (ISSN-L)
- OpenAlex
- W2279895976
- Cited by
- 548
- References
- 30
- Citations by year
- 2026: 20 · 2025: 32 · 2024: 44 · 2023: 57 · 2022: 52 · 2021: 68 · 2020: 90 · 2019: 82 · 2018: 57 · 2017: 39 · 2016: 6
- FWCI
- 107.26
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- Proceedings - IEEE International Conference on Robotics and Automation/Proceedings (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- Cambridge University Engineering Department Publications Database (submittedVersion)
- Apollo (University of Cambridge) (submittedVersion)
- arXiv (Cornell University)
- Apollo
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Robotics and Sensor-Based Localization
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Image and Video Retrieval Techniques
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Vision and Imaging
- Keywords
- camera relocalization, Bayesian convolutional neural network, uncertainty estimation, 6D pose estimation, real-time localization, out-of-distribution detection, indoor and outdoor scenes
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
