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 401–500 of 549
Sensing the Sensor: Estimating Camera Properties with Minimal Information
Ghosh, P., Liu, X., Qiu, H., Vieira, M. A. M., Sukhatme, G. S., & Govindan, R. (2022) · ACM Transactions on Sensor Networks · doi:10.1145/3508393 · Cited by 2
“Subsequent work has extended this in several ways: adding a Bayesian model to capture location uncertainty [30]; adding additional pooling layers [29]; exploring different loss functions in training [31, 38]; or using multi-task learning [53, 61].”
MethodQ-Learning Scheduler for Multi Task Learning Through the use of Histogram of Task Uncertainty
Meshgi, K., Mirzaei, M. S., & Sekine, S. (2022) · Workshop on Representation Learning for NLP (RepL4NLP) · doi:10.18653/v1/2022.repl4nlp-1.2 · Cited by 2
“The model uncertainty can be obtained via Monte Carlo dropout sampling (Kendall et al., 2015), as a function of the samples’ variance to be used as an estimation of the error (Kendall and Cipolla, 2016).”
MethodTowards Explainability in Modular Autonomous System Software
Zheng, H., Zang, Z., Yang, S., & Mangharam, R. (2023) · IEEE Intelligent Vehicles Symposium · doi:10.1109/iv55152.2023.10186720 · Cited by 2
High-accuracy 3D locators tracking in real time using monocular vision
Kulanesan, C. E., Vacher, P., Charleux, L., & Roux, É. (2024) · Machine Vision and Applications · doi:10.1007/s00138-023-01498-9 · Cited by 2
Camera Relocalization in Shadow-free Neural Radiance Fields
Xu, S., Liu, C., Chen, Y., Zhu, Z., Yan, Z., Shi, Y., Zhao, H., & Zhou, G. (2024) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra57147.2024.10611228 · Cited by 2
“Previous methods utilize discriminative networks [3]–[5] that perform implicit image feature matching and regress the absolute poses on the given images.”
MethodExploring Matching Rates: From Keypoint Selection to Camera Relocalization
Lin, H., Long, C., Fei, Y., Xia, Q., Yin, E., Yin, B., & Yang, X. S. (2024) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3664647.3681628 · Cited by 2
“Compared to the two-step process of extracting features from images and then performing retrieval, the PoseNet series [23–25] utilizes convolutional neural networks to extract features from images and subsequently to directly regress camera poses.”
BackgroundMethodYao, D., Zhu, M., Zhu, H., Cai, W., & Zhou, L. (2024) · ISPRS Journal of Photogrammetry and Remote Sensing · doi:10.1016/j.isprsjprs.2024.10.027 · Cited by 3
Hasan, H. M., García, M. Á., Rashwan, H. A., & Puig, D. (2025) · Image and Vision Computing · doi:10.1016/j.imavis.2025.105674 · Cited by 3
Large-Scale Benchmarking of Vision-Based Indoor Localization Using Absolute Pose Regression
Yu, J. Y., Yang, Y., & Namboodiri, V. (2025) · International Conference on Indoor Positioning and Indoor Navigation · doi:10.1109/ipin66788.2025.11213320 · Cited by 2
“In the Bayesian PoseNet, uncertainty is incorporated only before the layers with randomly initialized weights, following [29].”
Synthetic Datasets for Numeric Uncertainty Quantification: Proposing Datasets for Future Researchers
Kabir, H. M. D., Abdar, M., Khosravi, A., Nahavandi, D., Mondal, S. K., Khanam, S., Mohamed, S. M. K., Srinivasan, D., Nahavandi, S., & Suganthan, P. N. (2023) · IEEE Systems Man and Cybernetics Magazine · doi:10.1109/msmc.2022.3218423 · Cited by 2
“UQ is gaining vast popularity due to its demand in ML and DL methods [6], [7], [8], [9].”
BackgroundImage‐based localisation using shared‐information double stream hourglass networks
Ding, Y., Tao, Q., Wang, L., Li, D., & Zhang, M. (2018) · Electronics Letters · doi:10.1049/el.2017.3205 · Cited by 2
A Method to Build Multi-Scene Datasets for CNN for Camera Pose Regression
Ma, Y., Guo, H., Chen, H., Tian, M., Huo, X., Long, C., Tang, S., Song, X., & Wang, Q. (2018) · IEEE International Conference on Artificial Intelligence and Virtual Reality (AIVR) · doi:10.1109/aivr.2018.00022 · Cited by 2
“It shows that this feature selection is a more powerful regularization technique than dropout used in [13] in the pose regression task.”
MethodQuickest Detection of Abnormal Vehicle Movements on Highways
Wang, J., Dhanapal, R. K., Ramakrishnan, P., & Balasingam, B. (2019) · International Conference on Information Fusion (FUSION) · doi:10.23919/fusion43075.2019.9011374 · Cited by 2
“Other approaches in closely related research experimented swarm intelligence [35], hidden Markov models [36], and other Bayesian approaches [37], [15].”
MethodRnR: Retrieval and Reprojection Learning Model for Camera Localization
Yang, S., & Shi, D. (2021) · IEEE Access · doi:10.1109/access.2021.3061634 · Cited by 2
“Many studies [13]–[15] have been carried out on the basis of it.”
BackgroundMethodHandcrafted Features Can Help End-to-End Pose Estimation Using CNN
Hirakawa, Y., & KURODA, Y. (2022) · IEEE/SICE International Symposium on System Integration (SII) · doi:10.1109/sii52469.2022.9708741 · Cited by 2
Direct pose estimation from RGB images using 3D objects
Dede, M. A., & Genç, Y. (2022) · Pamukkale University Journal of Engineering Sciences · doi:10.5505/pajes.2021.08566 · Cited by 2
EpiLoc: Deep Camera Localization Under Epipolar Constraint
Xu, L., Guan, T., Luo, Y., Wang, Y., Chen, Z., & Liu, W. (2022) · KSII Transactions on Internet and Information Systems · doi:10.3837/tiis.2022.06.014 · Cited by 2
Camera localization with Siamese neural networks using iterative relative pose estimation
Kim, D., & Ko, K. H. (2022) · Journal of Computational Design and Engineering · doi:10.1093/jcde/qwac066 · Cited by 2
DarkLoc: Attention-based Indoor Localization Method for Dark Environments Using Thermal Images
Zhou, B., Pan, L., Li, Q., Liu, G., Xiong, A., & Li, Q. (2022) · International Conference on Indoor Positioning and Indoor Navigation · doi:10.1109/ipin54987.2022.9918100 · Cited by 2
“They further developed an uncertainty model for pose estimation called Bayesian PoseNet[7] and automatically balance location and orientation weights with constraints between 2D images and 3D models in PoseNet2[8].”
MethodOoD-Pose: Camera Pose Regression From Out-of-Distribution Synthetic Views
Ng, T., Lopez-Rodriguez, A., Balntas, V., & Mikolajczyk, K. (2022) · International Conference on 3D Vision (3DV) · doi:10.1109/3dv57658.2022.00082 · Cited by 2
“Further improvements have been made by exploring geometric constraints [8, 28], uncertainty [27] and sequences [84, 88].”
BackgroundEnergy-Based Models for Cross-Modal Localization using Convolutional Transformers
Wu, A., & Ryoo, M. S. (2023) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48891.2023.10160267 · Cited by 2
“Several extensions have improved upon performance by modeling pose uncertainty [26][27], and introducing geometry-aware loss functions [28].”
BackgroundMS-EBDL: Reliable Glaucoma Assessment via Sufficient Epistemic Uncertainty
Zhao, R., Ma, R., Zhou, L., Jia, X., & Liu, J. (2023) · IEEE International Conference on Bioinformatics and Biomedicine (BIBM) · doi:10.1109/bibm58861.2023.10386009 · Cited by 2
“Uncertainty measurement [16], [17] is an effective approach to improve the confidence of model predictions for reliable retinal assessment.”
BackgroundMobileARLoc: On-device Robust Absolute Localisation for Pervasive Markerless Mobile AR
Liu, C., Zhao, Y., & Braud, T. (2024) · IEEE International Conference on Pervasive Computing and Communications Workshops (PerCom Workshops) · doi:10.1109/percomworkshops59983.2024.10503320 · Cited by 2
“Moreover, existing uncertainty estimation methods can be time-consuming [15] and lack extensibility due to the need for specific loss functions and training schemes [16]–[19].”
BackgroundData-Driven Protection Levels for Camera and 3D Map-based Safe Urban Localization
Gupta, S., & Gao, G. (2020) · Proceedings of the Satellite Division's International Technical Meeting (Online)/Proceedings of the Satellite Division's International Technical Meeting (CD-ROM) · doi:10.33012/2020.17698 · Cited by 1
“Kendall and Cipolla [24] estimate the uncertainty in DNN-based camera pose estimation from images, by evaluating the network multiple times through dropout [25].”
BackgroundExploiting Points and Lines in Regression Forests for RGB-D Camera RelocalizationPreprint
Meng, L., Tung, F., Little, J. J., Valentin, J., & de Silva, C. W. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1710.10519 · Cited by 2
“We also provide results in terms of median translation error and rotation error against Bayesian PoseNet [11], PoseNet+Geometric [12], CNN+LSTM [13], Active Search [17], SCRF [2], BTBRF [8].”
BackgroundMethodInfluentialRisk-Aware Planning by Extracting Uncertainty from Deep Learning-Based Perception
Toubeh, M. (2018) · VTechWorks (Virginia Tech) · Cited by 2
Hybrid Scene Compression for Visual LocalizationPreprint
Camposeco, F., Cohen, A., Pollefeys, M., & Sattler, T. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1807.07512 · Cited by 2
“To this end, we compare to state-ofthe-art feature-based methods [8, 18, 34], as well as recent deep-learning-based methods [16, 17, 20, 21], as they also reduce the memory consumption of the localization database w.”
MethodInfluentialRisk-Aware Planning by Confidence Estimation using Deep Learning-Based PerceptionPreprint
Toubeh, M., & Tokekar, P. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1910.00101 · Cited by 2
“In a similar work, the same approximation approach is applied to assist in camera relocalization for landmark detection in a SLAM problem [13].”
MethodAdversarial Networks for Camera Pose Regression and RefinementPreprint
Bui, M., Baur, C., Navab, N., Ilić, S., & Albarqouni, S. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1903.06646 · Cited by 2
“Thus, several extensions and modifications of this method have been proposed, including uncertainty estimation [20, 21, 10], LSTM units [42], frame-to-frame information [2, 24, 13, 7] and previous pose fusion [40, 30].”
BackgroundMethodNotable Site Recognition using Deep Learning on Mobile and Crowd-sourced Imagery
Tan, J., Noulas, A., Sáez-Trumper, D., & Schifanella, R. (2020) · IEEE International Conference on Mobile Data Management (MDM) · doi:10.1109/mdm48529.2020.00036 · Cited by 2
“These environmental conditions effectively translate to a noisy environment that the system in place has to cope with [9], [10].”
Epistemic-Uncertainty-Based Divide-and-Conquer Network for Single-Image Super-Resolution
Yang, J., Chen, S., Li, Q., Jiang, T., Chen, Y., & Wang, J. (2022) · Electronics · doi:10.3390/electronics11223809 · Cited by 1
“Lately, the successful adoptions of Bayesian uncertainty in classification [16], segmentation [17], and camera re-localization problems [18] have shown the power of uncertainty in vision tasks.”
BackgroundPose and Focal Length Estimation Using a Point and a Line With Known Camera Position
Guo, K., Cao, R., Zhang, Z., Zhou, X. J., Yue, C., & Liu, B. (2023) · IEEE Access · doi:10.1109/access.2023.3317050 · Cited by 1
“ing more and more popular [1], [2], [3], [4], [5].”
BackgroundLocalization and Change Detection Through Aerial Environment Perception
Chiciudean, V., Florea, H., Oniga, F., & Nedevschi, S. (2024) · IEEE International Conference on Intelligent Computer Communication and Processing (ICCP) · doi:10.1109/iccp63557.2024.10792993 · Cited by 1
“PoseNet [28] [29], on the other hand, approaches 6DoF pose estimation as a regression problem.”
Centra-Net: A Centralized Network for Visual Localization Spanning Multiple Scenes
Jiang, Z., Dai, K., Wang, K., Xie, T., Fan, Z., Li, R., Kang, P., & Zhao, L. (2025) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2025.3535388 · Cited by 1
“…advancements of deep learning, learning-based works for visual localization are predominantly segmented into three primary parts: absolute pose regression (APR) [17], [18], [19], relative pose regression (RPR) [20], [21], and scene coordinate regression (SCoRe) [22], [23], [24], [25], [26], [27].”
HAL-NeRF: High Accuracy Localization Leveraging Neural Radiance Fields
Reppas, A., Cheimariotis, G., Papadopoulos, P. K., Frasiolas, P., & Zarpalas, D. (2025) · IEEE International Conference on Artificial Intelligence and eXtended and Virtual Reality (AIxVR) · doi:10.1109/aixvr63409.2025.00024 · Cited by 1
“A prominent group of techniques is Absolute Pose Regression (APR) [3], [4], [5], [6], which directly predicts camera poses from input images using deep learning models.”
A Survey of Visual SLAM Based on RGB-D Images Using Deep Learning and Comparative Study for VOE
Le, V.-H., & Nguyen, T.-H.-P. (2025) · Algorithms · doi:10.3390/a18070394 · Cited by 1
Wu, Q., & Campbell, M. (2025) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra55743.2025.11127943 · Cited by 1
“Pose regressor learns the pose in an end-to-end (image-to-pose) neural network [8], [9], which can be resource-intensive and unable to deal with dynamic scenes as well as generating to unseen images.”
MVL-Loc: Leveraging Vision-Language Model for Generalizable Multi-Scene Camera Relocalization
Xiao, Z., Yang, S., Ji, S., Yin, J., Wen, Z., & Wu, W. (2025) · Applied Sciences · doi:10.3390/app152312642 · Cited by 1
Multi-scene camera relocalization via modulated coordinate regression and low-rank adaptation
Sarıgül, M., Karacan, L., & Ata, B. (2025) · Scientific Reports · doi:10.1038/s41598-025-29429-w · Cited by 1
“Later improvements incorporated uncertainty modeling 10 , geometric reprojection loss 36 , and recurrent structures to capture spatial correlations in feature maps 37 .”
Active 6D pose estimation for textureless objects using multi-view RGB frames
Yang, J., Xue, W.-J., Ghavidel, S., & Waslander, S. L. (2025) · The international journal of robotics research · doi:10.1177/02783649251411922 · Cited by 8
“For example, in robot localization, the authors in (Zhang and Scaramuzza 2018, 2019) use the Fisher information maximization to find highly informative trajectories and achieve high localization accuracy.”
Huang, X., Zhu, Z., Zhang, X., Cao, W., & Yu, S. (2025) · International Conference on Electronic Information Engineering and Computer Communication (EIECC) · doi:10.1109/eiecc67963.2025.11409644 · Cited by 1
“Recent advances in deep learning have enabled learning-based camera pose estimation methods that directly regress camera poses from images using convolutional neural networks, avoiding explicit feature matching and geometric optimization[1].”
Wall-Object Multimodal Encoding for Robust Coarse-to-Fine Relocalization Using LiDAR Signals
Liu, D., Yan, W., & Wang, W. (2026) · IEEE Signal Processing Letters · doi:10.1109/lsp.2026.3709534 · Cited by 1
“Particle filtering [13], [14] distributes numerous particles across the map and evaluates their likelihood based on sensor observations.”
Deep Learning and Geometry-based Image Localization Enhanced by Bluetooth Signals
Ishihara, T., Kitani, K., Asakawa, C., & Hirose, M. (2018) · Journal of Information Processing · doi:10.2197/ipsjjip.26.707 · Cited by 1
On the role of geometry in geo-localization
Kadosh, M., Moses, Y., & Shamir, A. (2021) · Computational Visual Media · doi:10.1007/s41095-020-0196-2 · Cited by 2
“Kendall & Cipolla [10] improved their earlier work [12] by applying an uncertainty framework to the CNN pose regressor.”
MethodToward Accurate Motion Estimation with Affine Correspondences
Alyousefi, K. (2020) · Digital Collections of Colorado (Colorado State University) · Cited by 1
Optical Flow Prediction in Auto Driving from Single Image via Conditional Variational Auto-Encoder
Yan, J., Xu, L., & Mei, K. (2019) · IEEE International Conference on Unmanned Systems and Artificial Intelligence (ICUSAI) · doi:10.1109/icusai47366.2019.9124852 · Cited by 1
“These algorithms [1, 2, 11, 16, 18, 19, 20] mainly use the powerful fitting ability of the traditional convolutional neural network to extract the features of the two images and match features, which can get good results in some existing data sets.”
Deep Learning for Robot Vision
Mamilla, K. R., & Anto, S. (2021) · Smart innovation, systems and technologies · doi:10.1007/978-981-33-4443-3_34 · Cited by 1
Improving Robot Localisation by Ignoring Visual Distraction
Méndez, O., Vowels, M. J., & Bowden, R. (2021) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros51168.2021.9636595 · Cited by 1
“In this area, one of the key applications is camera pose regression and the most common family of approaches is PoseNet [12], [10], [11], [3] and its derivatives.”
BackgroundFusion of Visual Odometry Information for Enhanced Camera Pose Estimation
Wang, J., Wang, Y., Guo, C., Xing, S., & Ye, X. (2024) · International Conference on Control, Robotics and Cybernetics (CRC) · doi:10.1109/crc60659.2023.10488546 · Cited by 1
“A comparison is made with deep learning-based camera pose estimation models, including PoseNet[9], Bayesian PoseNet[10], PoseNet[11], VlocNet[12], DeepVO[13] and CnnBspp[14].”
MethodBack to the Context: Tuning Visual Localization Using Structural and Edge Context Within Image
Istighfarin, N. F., Lee, S., & Jo, H. (2024) · IEEE Access · doi:10.1109/access.2024.3483963 · Cited by 1
“PoseNet [5] and its variants [6], [7] employ CNN to map the global image into a camera pose directly commonly known as the absolute pose regression (APR) method.”
MethodMap-Free Visual Relocalization Enhanced by Instance Knowledge and Depth Knowledge
Xiao, M., Chen, R., Luo, H., Zhao, F., Wu, F., Xiong, H., Ma, X., & Wang, J. (2025) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp49660.2025.10888917 · Cited by 1
“We leverage instance knowledge to enhance the feature point matching and depth knowledge to facilitate scale recovery.”
UAN: Unified Attention Network for Convolutional Neural Networks.
Joseph, T., Derpanis, K. G., & Qureshi, F. Z. (2019) · Cited by 1
Augmenting Deep Learning Performance in an Evidential Multiple Classifier System
Vandoni, J., Hégarat‐Mascle, S. L., & Aldea, E. (2019) · Sensors · doi:10.3390/s19214664 · Cited by 1
“MC-dropout has been successfully applied in different applications, from segmentation for scene understanding [32] to camera re-localization [33], allowing to model the predictive uncertainty through standard deviation of the stochastic realizations obtained with dropout.”
MethodIndoor Visual Re-localization Based on Confidence Score Using Omni-Directional Camera
Takahashi, T., Fukuda, H., Kobayashi, Y., & Kuno, Y. (2020) · Communications in computer and information science · doi:10.1007/978-981-15-4818-5_15 · Cited by 1
ExMaps: Long-Term Localization in Dynamic Scenes using Exponential Decay
Rotsidis, A., Lutteroth, C., Hall, P. M., & Richardt, C. (2021) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv48630.2021.00291 · Cited by 1
“Learning-based methods are more recent and may directly regress camera pose from an input image [19, 20, 49] by representing the scene as the weights of a neural network.”
MethodGeometry-Aware Keypoint Network: Accurate Prediction of Point Features in Challenging Scenario
Nowak, T., & Skrzypczyński, P. (2022) · Annals of Computer Science and Information Systems · doi:10.15439/2022f145 · Cited by 1
“Among the examples considering uncertainty, [27] uses Bayesian neural networks to obtain localisation uncertainty, and a recent approach [21] learns the deep neural network uncertainty guided by the geometric uncertainty.”
BackgroundKurenkov, M., Kalinov, I., & Tsetserukou, D. (2022) · IEEE International Conference on Robotics and Biomimetics (ROBIO) · doi:10.1109/robio55434.2022.10011660 · Cited by 1
“In the following work, [16], the authors applied Bayesian inference to the neural network for computing the uncertainty of the final PoseNet coordinate.”
MethodLong-Distance Loop Closure Using General Object Landmarks.Preprint
Holliday, A. G. C., & Dudek, G. (2017) · arXiv (Cornell University) · Cited by 1
Joint Spatial and Layer Attention for Convolutional NetworksPreprint
Joseph, T., Derpanis, K. G., & Qureshi, F. Z. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1901.05376 · Cited by 1
“For image-based camera localization we compare our work with previous approaches proposed in [21], [18] and [37] on 7-Scenes, Cambridge Landmarks, and TUM-LSI.”
MethodResultInfluentialOn Localizing a Camera from a Single ImagePreprint
Ghosh, P., Liu, X., Qiu, H., Vieira, M. A. M., Sukhatme, G. S., & Govindan, R. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2003.10664 · Cited by 4
“Kendall and Cipolla [43] extends PoseNet to incorporate a Bayesian model that determines the localization’s uncertainty.”
BackgroundDeep auxiliary learning for visual localization using colorization taskPreprint
Tian, M., Nie, Q., Shen, H., & Xia, X. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2107.00222 · Cited by 1
“[17] utilized Bayesian CNNs to estimate the uncertainty of predicted pose.”
BackgroundMethodEpistemic uncertainty estimation for object detection in open-set conditions
Miller, D. (2021) · Queensland University of Technology · doi:10.5204/thesis.eprints.213588 · Cited by 1
Polarization based urban scenes understanding
Blanchon, M. (2021) · theses.fr (ABES) · doi:10.70675/9eb0b15bzd1bcz473bza318zfc8c58cc92f1 · Cited by 1
Xia, J. W., Guan, M., & Yang, X. (2017) · International Conference on Computer Communication and Informatics (ICCCI) · doi:10.1109/iccci.2017.8117704
“A strategy referred to as a dropout is ordinarily utilized as a regularizer as a part of convolutional neural systems to avoid over fitting and co-adaption of components [13].”
BackgroundRetractedOptimization Algorithm Toward Deep Features Based Camera Pose Estimation
Chen, H., Guo, F., Lin, Y., & Ji, R. (2017) · Lecture notes in computer science · doi:10.1007/978-3-319-71598-8_45
“[10] present a visual re-localization system that uses a Bayesian convolutional neural network to regress camera pose.”
BackgroundMethodVisual Relocalization Using Long-Short Term Memory Fully Convolutional Network
Zhou, L. (2017) · IEEE International Conference on Tools with Artificial Intelligence (ICTAI) · doi:10.1109/ictai.2017.00097
“We do not enforce the constraint ||q||2=1 during the training, and normalize the predicted quaternion to satisfy the length one condition during the prediction as [1-3].”
BackgroundMethodInfluentialUsing Monte Carlo dropout for non-stationary noise reduction from speechPreprint
Nazreen, P. M., & Ramakrishnan, A. G. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1808.09432
“[25] uses the term MC droput to refer to this technique.”
InfluentialComputing the Relations Among Three Views Based on Artificial Neural Network
Yu, Y. K., Wong, K. H., & Or, S. H. (2018) · IEEE Region 10 Symposium (TENSYMP) · doi:10.1109/tenconspring.2018.8691900
“in [7], [8] adopted a Bayesian convolutional neural network to regress the 6-degree-offreedom camera pose relative to the world coordinate frame from a single RGB image.”
MethodSASSE: Scalable and Adaptable 6-DOF Pose EstimationPreprint
Le, H., Hoang, T., Zhang, Q., Do, T.-T., Eriksson, A. P., & Milford, M. J. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1902.01549
“Scene #Train #Test Active Search [27] PoseNet [16] Bayesian PoseNet [14] PoseNet-LSTM [34] PoseNet17 [15] SASSE SASSE + PGO”
MethodFast Bayesian Uncertainty Estimation of Batch Normalized Single Image Super-Resolution Network.Preprint
Kar, A., & Biswas, P. K. (2019) · arXiv (Cornell University)
FishNet: A Camera Localizer using Deep Recurrent NetworksPreprint
Chen, H., Agethen, S., Wu, C.-M., Hsu, W. H., & Chen, B. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1904.09722
“Scene 2D-3D Matching [28] PoseNet [14] Bayesian PoseNet [13] LSTM LSTM (Reg.”
BackgroundDNN Based Speech Enhancement for Unseen Noises Using Monte Carlo DropoutPreprint
Nazreen, P. M., & Ramakrishnan, A. G. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1806.00516
“The uncertainty is computed by taking the trace of the covariance matrix of each frame [25].”
BackgroundMethodResultInfluentialImproving drone localisation around wind turbines using monocular model-based trackingPreprint
Moolan-Feroze, O., Karachalios, K., Nikolaidis, D., & Calway, A. D. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1902.10474
“This work was later expanded on with the inclusion of measures of uncertainty into the output [7] using Bayesian deep learning.”
MethodNotable Site Recognition on Mobile Devices using Deep Learning with Crowd-sourced Imagery.
Tan, J., Noulas, A., Sáez-Trumper, D., & Schifanella, R. (2019)
Mobile Recognition of Wikipedia Featured Sites using Deep Learning and Crowd-sourced ImageryPreprint
Tan, J., Noulas, A., Sáez-Trumper, D., & Schifanella, R. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1910.09705
“These environmental conditions effectively translate to a noisy environment that the system in place has to cope with [17, 38].”
BackgroundDecoupling Features and Coordinates for Few-shot RGB RelocalizationPreprint
Dong, S., Wu, S., Zhuang, Y., Xu, K., Zhang, S., & Chen, B. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1911.11534
“Several improvement have been proposed through modeling uncertainty [17], exploiting structured feature correlation [34] and designing geometric loss functions [18].”
MethodLearning Stochastic Weight Masking to Resist Adversarial Attacks
Kubo, Y. (2019) · DalSpace (Dalhousie University)
Basier, F., McLean, K., & Pelling, R. (2020) · International Petroleum Technology Conference · doi:10.2523/iptc-19804-abstract
MVLoc: Multimodal Variational Geometry-Aware Learning for Visual Localization.Preprint
Zhou, R. H., Chen, C., Wang, B., Markham, A., & Trigoni, N. (2020) · arXiv (Cornell University)
Nazreen, P. M., & Ramakrishnan, A. G. (2019) · IEEE India Conference · doi:10.1109/indicon47234.2019.9030357
“For the uncertainty measurement, the trace of the covariance matrix of the output samples ( V ar ) is used [16].”
BackgroundMethod6D Camera Relocalization in Ambiguous Scenes via Continuous Multimodal InferencePreprint
Bui, M., Birdal, T., Deng, H., Albarqouni, S., Guibas, L. J., Ilić, S., & Navab, N. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2004.04807
“This has led to a paradigm shift towards estimating a range of solutions, in the form of full probability distributions [1,7,8] or at least solutions that estimate the uncertainty in orientation estimates [43, 54].”
BackgroundMethodInfluentialEgo-Motion Estimation Using Affine Correspondences
Alyousefi, K., & Ventura, J. (2020) · Advances in intelligent systems and computing · doi:10.1007/978-3-030-43020-7_70
Adversarial Transfer of Pose Estimation RegressionPreprint
Sadek, B. C. A. (2020) · arXiv (Cornell University)
Uncertainty in Neural Relational Inference Trajectory ReconstructionPreprint
Karavias, V., Day, B., & Lió, P. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2006.13666
“Recent papers have investigated teasing out model uncertainty by using dropout in the network as a Bayesian deep Gaussian process (Gal & Ghahramani, 2016; Kendall & Cipolla, 2016).”
BackgroundMaking Decisions Under High Stakes: Trustworthy and Expressive Bayesian Deep Learning
Yang, W. (2020) · Digital Access to Scholarship at Harvard (DASH) (Harvard University)
Structured feature enhanced deep visual relocalization
Peng, Q., Xiang, Z., & Wang, T. (2020) · IEEE International Conference on Unmanned Systems (ICUS) · doi:10.1109/icus50048.2020.9274860
Probabilistic Localization of Insect-Scale Drones on Floating-Gate Inverter ArraysPreprint
Shukla, P., Muralidhar, A., Iliev, N., Tulabandhula, T., Fuller, S. B., & Trivedi, A. R. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2102.08247
“A probabilistic framework for drone localization models the likelihood of measurement against a hypothesized drone’s pose model [18].”
Hung, C.-M., Sun, L., Wu, Y., Havoutis, I., & Posner, I. (2021) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48506.2021.9561749
“For an uncertainty estimate, following probabilistic PoseNet [29], we have experimented with the trace of covariance matrix of the samples and the maximum of the variance along each axis.”
MethodThe Visual Localization Based On The Semantic Error Image
Xin, X., Jiang, J., & Kang, L. (2020) · International Conference on Big Data and Information Analytics (BigDIA) · doi:10.1109/bigdia51454.2020.00079
SMVNet: Deep Learning Architectures for Accurate and Robust Multi-View Stereopsis
Yao, S., Wang, Y., Aliakbarpour, H., Seetharaman, G. S., Rao, R., & Palaniappan, K. (2020) · IEEE Applied Imagery Pattern Recognition Workshop · doi:10.1109/aipr50011.2020.9425188
“Recently, machine learning based techniques, especially CNN based deep learning methods, have been widely used to solve MVS related problems including 3D semantic segmentation [21], [22], structure-from-motion (SFM) [23], synthetic view generating [24], image-based localization [25], [26], and etc..”
MethodThe brain at risk: Detection and quantification of cerebrovascular disease
van Hespen, K. M. (2021) · Utrecht University Repository (Utrecht University) · doi:10.33540/521
Markov Localisation using Heatmap Regression and Deep Convolutional Odometry
Méndez, O., Hadfield, S., & Bowden, R. (2021) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48506.2021.9562044
“They concluded that PoseNet-like models are only reliable for coarse pose estimation and this calls into question the reliability of learning-based methods that do not employ other sources of information.”
InfluentialFang, J., & Guizhong, L. (2021) · IEEE International Conference on Unmanned Systems (ICUS) · doi:10.1109/icus52573.2021.9641225
“To settle the weakness of sparse hand-crafted features, recent methods introduced the deep learning based scene geometry understanding methods via a lot of densely labeled ground-truth data, such as monocular depth estimation [13],[14], optical flow [16],[17],[18] and camera pose estimation [20],[21].”
MethodContinual Learning for Image-Based Camera LocalizationPreprint
Wang, S., Laskar, Z., Melekhov, I., Li, X., & Kannala, J. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2108.09112
Comparison Between Two Recurrent Multitasking Networks for Robot Relocation
Aragão, D. P., & Nascimento, T. (2021) · Latin American Robotics Symposium, Brazilian Symposium on Robotics and Workshop on Robotics in Education (LARS/SBR/WRE) · doi:10.1109/lars/sbr/wre54079.2021.9605396
DEEP LEARNING APPLIED TO LOCATING FERROMAGNETIC FOREIGN BODIES IN HUMANS
Rogozinski, M. (2021) · doi:10.17771/pucrio.acad.57112
“More recent methods seek to solve the problem using regression from Convolutional Neural Networks to find the position and orientation of the camera directly from the image [34]–[41].”
MethodFINet: Feature Interactions Across Dimensions and Hierarchies for Camera Localization
liao, T., Tang, T., Chen, J., Liang, Y., & Xiao, Y. (2021) · International Conference on Computing and Pattern Recognition (ICCPR) · doi:10.1145/3497623.3497631
“APR methods was first proposed by PoseNet [12] to connect GoogLeNet backbone with the Multilayer Perceptrons (MLPs) for reckoning camera poses, subsequently, Hourglass and Bayesian PoseNet [10, 16] modify the backbone CNNs to improve accuracy.”
BackgroundMethodComputing the Relations Among Three Views Based on Artificial Neural Network
Yu, Y. K., Wong, K. H., & Or, S. H. (2018) · IEEE International Conference on Electro Information Technology · doi:10.1109/eit.2018.8500160
IMAGE ORIENTATION BY EMBEDDING IN A GAN LATENT SPACE
Kniaz, V. V., Knyaz, V. A., Mizginov, V. A., Bordodymov, A. N., Moshkantsev, P. V., Barylnik, S., & Novikov, D. (2022) · The international archives of the photogrammetry, remote sensing and spatial information sciences/International archives of the photogrammetry, remote sensing and spatial information sciences · doi:10.5194/isprs-archives-xlviii-2-w2-2022-149-2022
“For input image the pretrained PoseNet model generates samples by dropping out activation units of convolutional layers with a given probability.”
MethodTu, C.-H., & Lu, E. H.-C. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-21743-2_4
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
