The DOI matches its Crossref record; 29 of the 29 references it cites were found in a registry or an index.
DeepVO: Towards end-to-end visual odometry with deep Recurrent Convolutional Neural Networks
Open accessAbstract
This paper studies monocular visual odometry (VO) problem. Most of existing VO algorithms are developed under a standard pipeline including feature extraction, feature matching, motion estimation, local optimisation, etc. Although some of them have demonstrated superior performance, they usually need to be carefully designed and specifically fine-tuned to work well in different environments. Some prior knowledge is also required to recover an absolute scale for monocular VO. This paper presents a novel end-to-end framework for monocular VO by using deep Recurrent Convolutional Neural Networks (RCNNs). Since it is trained and deployed in an end-to-end manner, it infers poses directly from a sequence of raw RGB images (videos) without adopting any module in the conventional VO pipeline. Based on the RCNNs, it not only automatically learns effective feature representation for the VO problem through Convolutional Neural Networks, but also implicitly models sequential dynamics and relations using deep Recurrent Neural Networks. Extensive experiments on the KITTI VO dataset show competitive performance to state-of-the-art methods, verifying that the end-to-end Deep Learning technique can be a viable complement to the traditional VO systems.
Overview
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
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?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“More surprisingly, they show that the welltrained networks enable scale-consistent predictions over long videos, while the accuracy is still inferior to traditional methods because of ignoring geometric information.”
From DF-VO: What Should Be Learnt for Visual Odometry? (2021) · cites this paper as Wang et al. 2017
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“On the opposing side, end-to-end approaches can estimate the 6 degrees of freedom (DoF) poses directly from a sequence of RGB images, that is, it does not need engineering effort to design the modules in the traditional VO pipeline [4].”
From Transformer-Based Model for Monocular Visual Odometry: A Video Understanding Approach (2023) · cites this paper as [4]
The cited source says what the claim attributes to it.
In the citing paper
“This problem is called monocular visual odometry and often relies on geometric approaches that require considerable engineering effort for a specific scenario.”
From Transformer-Based Model for Monocular Visual Odometry: A Video Understanding Approach (2023) · cites this paper as [4]
The check wasn’t confident either way, so it gives no verdict — never disputed.
In the citing paper
“the amount of crater detections that ODN-detector produces is robust to disturbances or modifications of the image quality due to Gaussian noise, shot noise, brightness levels, as well as different illumination conditions.”
From Optical navigation for Lunar landing based on Convolutional Neural Network crater detector (2022) · cites this paper as [16]
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Very Deep Convolutional Networks for Large-Scale Image Recognition
Simonyan, K., & Zisserman, A. (2014) · arXiv · doi:10.48550/arxiv.1409.1556 · Cited by 114,148 · cited 1×
“There have been some popular and powerful DNN architectures, such as VGGNet [22] and GoogLeNet [23], developed for computer vision tasks, producing remarkable performance.”
MethodGoing 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
Are we ready for autonomous driving? The KITTI vision benchmark suite
Geiger, A., Lenz, P., & Urtasun, R. (2012) · 2012 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2012.6248074 · Cited by 14,897 · cited 2×
“In this section, we evaluate the proposed end-to-end monocular VO approach on the well-known KITTI VO/SLAM benchmark [3].”
BackgroundMethod(2016) · Cited by 11,679
ORB-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×
“Examples include keyframe base PTAM [9] and ORB-SLAM [10].”
BackgroundParallel Tracking and Mapping for Small AR Workspaces
Klein, G., & Murray, D. (2007) · 2007 6th IEEE and ACM International Symposium on Mixed and Augmented Reality · doi:10.1109/ismar.2007.4538852 · Cited by 4,346 · cited 1×
“Examples include keyframe base PTAM [9] and ORB-SLAM [10].”
BackgroundFlownet: Learning optical flow with convolutional networks
Dosovitskiy, A., Fischer, P., Ilg, E., Hausser, P., Hazirbas, C., Golkov, V., Smagt, P. V. D., Cremers, D., & Brox, T. (2015) · 2015 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2015.316 · Cited by 4,313 · cited 2×
“In order to reduce both the training time and data required to converge, the CNN is based on a pre-trained FlowNet model [24].”
MethodMonoSLAM: Real-time single camera SLAM
Davison, A. J., Reid, I. D., Molton, N. D., & Stasse, O. (2007) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2007.1049 · Cited by 3,959 · cited 1×
“To mitigate this problem, visual Simultaneous Localisation and Mapping (SLAM) or Structure from Motion (SfM) can be adopted to maintain a feature map for drift correction along with pose estimation [8].”
MethodORB-SLAM: A Versatile and Accurate Monocular SLAM System
Mur-Artal, R., Montiel, J. M. M., & Tardos, J. D. (2015) · Zaguan (Universidad de Zaragoza) · Cited by 3,153
Multiple view geometry in computer vision
Opower, H. (2002) · Optics and Lasers in Engineering · doi:10.1016/s0143-8166(01)00145-2 · Cited by 3,044
Flagged first, then by citations
Research works printed but not linked: 1
- zaremba, 2014, Learning to execute
Cited by
Sorted by citations
Items 101–200 of 969
DytanVO: Joint Refinement of Visual Odometry and Motion Segmentation in Dynamic Environments
Shen, S., Cai, Y., Wang, W., & Scherer, S. (2023) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48891.2023.10161306 · Cited by 68
“DeepVO [4], TartanVO and TrianFlow [42] are three learning-based VO methods not targeting dynamic scenes while CC [11] is an”
BackgroundMethodInfluentialDetecting Deficient Coverage in Colonoscopies
Freedman, D., Blau, Y., Katzir, L., Aides, A., Shimshoni, I., Veikherman, D., Golany, T., Gordon, A., Corrado, G. S., Matias, Y., et al. (2020) · IEEE Transactions on Medical Imaging · doi:10.1109/tmi.2020.2994221 · Cited by 68
“[9], [10]; however, this entails the need for domain adaptation, which we avoid.”
BackgroundMethodEfficient 3D Deep LiDAR Odometry
Wang, G., Wu, X., Jiang, S., Liu, Z., & Wang, H. (2022) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2022.3207015 · Cited by 66
Learning to Solve Nonlinear Least Squares for Monocular Stereo
Clark, R., Bloesch, M., Czarnowski, J., Leutenegger, S., & Davison, A. J. (2018) · Lecture notes in computer science · doi:10.1007/978-3-030-01237-3_18 · Cited by 65
Self-citationOptical navigation for Lunar landing based on Convolutional Neural Network crater detector
Silvestrini, S., Piccinin, M., Zanotti, G., Brandonisio, A., Bloise, I., Feruglio, L., Lunghi, P., Lavagna, M., & Varile, M. (2022) · Aerospace Science and Technology · doi:10.1016/j.ast.2022.107503 · Cited by 65
Improving GNSS Positioning Using Neural-Network-Based Corrections
Kanhere, A. V., Gupta, S., Shetty, A., & Gao, G. (2022) · NAVIGATION Journal of the Institute of Navigation · doi:10.33012/navi.548 · Cited by 64
Global Pose Estimation with an Attention-Based Recurrent Network
Parisotto, E., Chaplot, D. S., Zhang, J., & Salakhutdinov, R. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw.2018.00061 · Cited by 88
A Survey on Visual Navigation and Positioning for Autonomous UUVs
Qin, J., Li, M., Li, D., Zhong, J., & Yang, K. (2022) · Remote Sensing · doi:10.3390/rs14153794 · Cited by 65
“The generative adversarial network plays an important role in navigation, positioning, and mapping, especially supervised depth estimation, due to its feature of automatically generating low-cost training data.”
BackgroundPointLoc: Deep Pose Regressor for LiDAR Point Cloud Localization
Wang, W., Wang, B., Zhao, P., Chen, C., Clark, R., Yang, B., Markham, A., & Trigoni, N. (2021) · IEEE Sensors Journal · doi:10.1109/jsen.2021.3128683 · Cited by 62
Self-citationLocal Supports Global: Deep Camera Relocalization With Sequence Enhancement
Xue, F., Wang, X., Yan, Z., Wang, Q., Wang, J., & Zha, H. (2019) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2019.00293 · Cited by 62
“Since images are high-dimension data with much redundant information, learning information from raw data is ineffective [35].”
BackgroundMonocular Visual Odometry Based on Depth and Optical Flow Using Deep Learning
Ban, X., Wang, H., Chen, T., Wang, Y., & Xiao, Y. (2020) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2020.3024011 · Cited by 62
“The application of DL methods to VSLAM is a landmark development, and it makes full use of the advantages of DL to process image sequence [9].”
BackgroundMethodActive Neural LocalizationPreprint
Chaplot, D. S., Parisotto, E., & Salakhutdinov, R. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1801.08214 · Cited by 59
“…et al., 1960; Smith et al., 1990) geometry-based visual odometry methods (Nist´er et al., 2006) and most recently, learning-based visual odometry methods which learn to predict motion between consecutive frames using recurrent convolutional neural networks (Clark et al., 2017b; Wang et al., 2017).”
BackgroundMethodGuided Feature Selection for Deep Visual Odometry
Xue, F., Wang, Q., Wang, X., Dong, W., Wang, J., & Zha, H. (2019) · Lecture notes in computer science · doi:10.1007/978-3-030-20876-9_19 · Cited by 58
A Practical Survey on Visual Odometry for Autonomous Driving in Challenging Scenarios and Conditions
Agostinho, L. R., Ricardo, N. M., Pereira, M. I., Hiolle, A., & Pinto, A. M. (2022) · IEEE Access · doi:10.1109/access.2022.3188990 · Cited by 57
Learning Monocular Visual Odometry through Geometry-Aware Curriculum Learning
Saputra, M. R. U., de Gusmão, P. P. B., Wang, S., Markham, A., & Trigoni, N. (2019) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2019.8793581 · Cited by 56
“We also use δ and ζ to weigh the translation and rotation terms in relative loss as seen in [22], [4].”
BackgroundMethodResultInfluentialSelf-citationDeep Keypoint-Based Camera Pose Estimation with Geometric Constraints
Jau, Y.-Y., Zhu, R., Su, H., & Chandraker, M. K. (2020) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros45743.2020.9341229 · Cited by 60
“To take advantage of temporal information, DeepVO [22], [30] proposes to use recurrent neural networks (RNN) to predict poses along the sequence.”
MethodFutureMapping: The Computational Structure of Spatial AI SystemsPreprint
Davison, A. J. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1803.11288 · Cited by 70
“[55] on estimating incremental visual odometry from video using an RNN whose training was via a supervised known pose signal.”
MethodZhao, C., Tang, Y. S., Sun, Q., & Vasilakos, A. V. (2021) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2021.3071886 · Cited by 55
Aslan, M. F., Durdu, A., Yusefi, A., & Yılmaz, A. (2022) · Neural Networks · doi:10.1016/j.neunet.2022.09.001 · Cited by 58
Li, Y., Shen, H., Fu, Y., & Wang, K. (2024) · Expert Systems with Applications · doi:10.1016/j.eswa.2024.124918 · Cited by 56
Who Let the Dogs Out? Modeling Dog Behavior from Visual Data
Ehsani, K., Bagherinezhad, H., Redmon, J., Mottaghi, R., & Farhadi, A. (2018) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2018.00426 · Cited by 64
Self-Supervised deep homography estimation with invertibility constraints
Wang, C., Wang, X., Bai, X. P., Liu, Y., & Zhou, J. W. (2019) · Pattern Recognition Letters · doi:10.1016/j.patrec.2019.09.021 · Cited by 53
Generalizing to the Open World: Deep Visual Odometry with Online Adaptation
Li, S., Wu, X., Cao, Y., & Zha, H. (2021) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr46437.2021.01298 · Cited by 53
“Learning-based VO has been widely studied in recent years and shown impressive results [35, 37, 45].”
Hierarchical Metric Learning and Matching for 2D and 3D Geometric Correspondences
Fathy, M. E., Tran, Q.-H., Zia, M. Z., Vernaza, P., & Chandraker, M. K. (2018) · Lecture notes in computer science · doi:10.1007/978-3-030-01267-0_49 · Cited by 52
Near-Field Perception for Low-Speed Vehicle Automation Using Surround-View Fisheye Cameras
Eising, C., Horgan, J., & Yogamani, S. K. (2021) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2021.3127646 · Cited by 58
“In computer vision, deep learning is playing a dominant role in various recognition tasks and gradually for geometric tasks, like depth [17] and motion estimation [49].”
BackgroundRelPose++: Recovering 6D Poses from Sparse-view Observations
Lin, A., Zhang, J. Y., Ramanan, D., & Tulsiani, S. (2024) · International Conference on 3D Vision (3DV) · doi:10.1109/3dv62453.2024.00126 · Cited by 52
“For instance, poses can directly be predicted using an RNN [45, 50] or auto-regressively [53] for SLAM applications.”
BackgroundLipson, L., Teed, Z., & Deng, J. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-72627-9_24 · Cited by 52
End-to-end navigation for Autonomous Underwater Vehicle with Hybrid Recurrent Neural Networks
Mu, X., He, B., Zhang, X., Song, Y., Shen, Y., & Feng, C. (2019) · Ocean Engineering · doi:10.1016/j.oceaneng.2019.106602 · Cited by 52
Learning Monocular Visual Odometry with Dense 3D Mapping from Dense 3D Flow
Zhao, C., Sun, L., Purkait, P., Duckett, T., & Stolkin, R. (2018) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2018.8594151 · Cited by 50
Efficient Deep Visual and Inertial Odometry with Adaptive Visual Modality Selection
Yang, M., Chen, Y., & Kim, H.-S. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-19839-7_14 · Cited by 56
Ye, W., Lan, X., Chen, S., Ming, Y., Yu, X., Bao, H., Cui, Z., & Zhang, G. (2023) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52729.2023.00924 · Cited by 51
Uncertainty Estimation for Data-Driven Visual Odometry
Costante, G., & Mancini, M. (2020) · IEEE Transactions on Robotics · doi:10.1109/tro.2020.3001674 · Cited by 51
“3) LS-VO and DeepVO: Asdata-drivenVObaselines,we opt for LS-VO [20] and DeepVO [11].”
BackgroundMethodGraph-Based Thermal–Inertial SLAM With Probabilistic Neural Networks
Saputra, M. R. U., Lu, C. X., de Gusmão, P. P. B., Wang, B., Markham, A., & Trigoni, N. (2021) · IEEE Transactions on Robotics · doi:10.1109/tro.2021.3120036 · Cited by 49
Self-citationVOLDOR: Visual Odometry From Log-Logistic Dense Optical Flow Residuals
Min, Z., Yang, Y., & Dunn, E. (2020) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr42600.2020.00495 · Cited by 47
“Further adding recurrent neural networks for learning temporal information is presented in [81, 82, 87, 49].”
BackgroundGeometric Consistency for Self-Supervised End-to-End Visual Odometry
Iyer, G., Jatavallabhula, K. M., Gupta, G., Krishna, K. M., & Paull, L. (2018) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw.2018.00064 · Cited by 49
Tightly-Coupled Single-Anchor Ultra-wideband-Aided Monocular Visual Odometry System
Nguyen, T. H., Nguyen, T., & Xie, L. (2020) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra40945.2020.9196794 · Cited by 46
“Depth learning-based methods [19], [20] are applicable in a wide range of scenarios, but requires large amount of data and high computing power platform for training and deployment.”
MethodSensors and Sensor Fusion Methodologies for Indoor Odometry: A Review
Yang, M., Sun, X. S., Jia, F., Rushworth, A. G. A., Dong, X., Zhang, S., Fang, Z., Yang, G., & Liu, B. (2022) · Polymers · doi:10.3390/polym14102019 · Cited by 50
“To enumerate some, for visual odometry, DeepVO [242] leverages deep Recurrent Convolu-tional Neural Networks (RCNNs) to estimate pose in an end-to-end fashion; D3VO [243] formulates the deep prediction of depth, pose, and uncertainty into direct visual odometry.”
MethodRupprecht, T., & Wang, Y. (2022) · Neural Networks · doi:10.1016/j.neunet.2022.05.013 · Cited by 48
Wang, K., Zhao, G., & Lǚ, J. (2024) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2024.3379993 · Cited by 48
“In response to these challenges, recent advancements in deep learning within computer vision have led to the proposition of learning-based VSLAM methods [16], [17], [18], [19].”
BackgroundUnderwater SLAM Meets Deep Learning: Challenges, Multi-Sensor Integration, and Future Directions
Heshmat, M., Saoud, L. S., Abujabal, M., Sultan, A., Elmezain, M., Seneviratne, L. D., & Hussain, I. (2025) · Italian National Conference on Sensors · doi:10.3390/s25113258 · Cited by 48
Wide-Baseline Relative Camera Pose Estimation with Directional Learning
Chen, K., Snavely, N., & Makadia, A. (2021) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr46437.2021.00327 · Cited by 45
“In general, these systems make design choices specific for small-baselines and video sequences, such as using frame-to-frame image reconstruction losses [65, 63, 61], or training with more than two frames [60, 12, 65].”
Overview of deep learning application on visual SLAM
Li, S., Zhang, D., Xian, Y., Li, B., Zhang, T., & Zhong, C. (2022) · Displays · doi:10.1016/j.displa.2022.102298 · Cited by 47
A comprehensive overview of dynamic visual SLAM and deep learning: concepts, methods and challenges
Beghdadi, A., & Mallem, M. (2022) · Machine Vision and Applications · doi:10.1007/s00138-022-01306-w · Cited by 49
“Also, mainstream research focuses on developing some end-to-end methods [101,109,110] to track the camera motion by working on deep learning methods providing better interpretability and generalization.”
BackgroundEDPLVO: Efficient Direct Point-Line Visual Odometry
Zhou, L., Huang, G. P., Mao, Y., Wang, S., & Kaess, M. (2022) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra46639.2022.9812133 · Cited by 42
“In terms of VO, the learning-based methods have achieved significant progress in recent years [4]–[8].”
Self-citationNavnet: AUV Navigation Through Deep Sequential Learning
Zhang, X., He, B., Li, G., Mu, X., Zhou, Y., & Mang, T. (2020) · IEEE Access · doi:10.1109/access.2020.2982272 · Cited by 44
“Wang et al. [29] presented a novel end-to-end framework for monocular VO by utilizing deep Recurrent Convolutional Neural Network (RCNN).”
MethodDynamic Attention-based Visual Odometry
Kuo, X.-Y., Liu, C., Lin, K.-C., & Lee, C. (2020) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw50498.2020.00026 · Cited by 43
“These baseline methods include non-DCNN based visual SLAM framework ORB-SLAM2 [18] (without enabling the loop closing for a fair comparison with the baseline VO algorithms), the open-source visual odometry library VISO2 [50], as well as the related learning-based VO works discussed in Section 2, such as CL-VO [11], De…”
BackgroundMethodInfluentialMonocular Visual-Inertial-Wheel Odometry Using Low-Grade IMU in Urban Areas
Jung, J. H., Cha, J., Chung, J. Y., Kim, T. I., Seo, M. H., Park, S. Y., Yeo, J. Y., & Park, C. G. (2020) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2020.3018167 · Cited by 43
DeepAVO: Efficient pose refining with feature distilling for deep Visual Odometry
Zhu, R., Yang, M., Liu, W., Song, R., Yan, B., & Xiao, Z. (2021) · Neurocomputing · doi:10.1016/j.neucom.2021.09.029 · Cited by 45
A multi-task learning framework for gas detection and concentration estimation
Liu, H., Li, Q., & Gu, Y. (2020) · Neurocomputing · doi:10.1016/j.neucom.2020.01.051 · Cited by 42
Learning Selective Sensor Fusion for State Estimation
Chen, C., Rosa, S., Lu, C. X., Wang, B., Trigoni, N., & Markham, A. (2022) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2022.3176677 · Cited by 46
“Recently, there are growing interests in applying deep neural networks (DNNs) for learning to estimate system states in an end-to-end manner, for example, solving visual odometry (VO) [20], [45], [53], visual-inertial odometry (VIO) [7], [35], or camera relocalization [6], [22].”
Self-citationVisual SLAM for Unmanned Aerial Vehicles: Localization and Perception
Zhuang, L., Zhong, X., Xu, L., Tian, C., & Yu, W.-S. (2024) · Italian National Conference on Sensors · doi:10.3390/s24102980 · Cited by 46
“More recently, learning-based methods have been used for end-to-end estimating of the camera pose [20].”
MethodDeep Orientation Uncertainty Learning based on a Bingham Loss
Glitschenski, I., Schwarting, W., Sahoo, R., Amini, A., & Karaman, S. (2020) · International Conference on Learning Representations · Cited by 41
The Surprising Effectiveness of Visual Odometry Techniques for Embodied PointGoal Navigation
Zhao, X., Agrawal, H., Batra, D., & Schwing, A. G. (2021) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv48922.2021.01582 · Cited by 42
SBAS: Salient Bundle Adjustment for Visual SLAM
Wang, K., Ma, S., Ren, F., & Lǚ, J. (2021) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2021.3105243 · Cited by 41
Improving Monocular Visual Odometry Using Learned Depth
Sun, L., Yin, W., Xie, E., Li, Z., Sun, C., & Shen, C. (2022) · IEEE Transactions on Robotics · doi:10.1109/tro.2022.3164834 · Cited by 42
“VO methods [50], [25], [1], [2], [4], [51] have been proposed.”
DiffPoseNet: Direct Differentiable Camera Pose Estimation
Parameshwara, C. M., Hari, G., Fermuller, C. M., Sanket, N. J., & Aloimonos, Y. (2022) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52688.2022.00672 · Cited by 40
“In Table 4 we compare our model (DiffPoseNet) with (a) pure deep learning models (TartanVO [45], GeoNet [50], UnDeepVO [29], DeepVO [44], Wang et al. [47]), (b) pure geometric constraint based approaches (ORB-SLAM [32], VISO2-M [41]) and with (c) BiLevelOpt [25], a method implementing the epipolar constraint via a dif…”
BackgroundMethodInfluentialLearning to Communicate and Correct Pose ErrorsPreprint
Vadivelu, N., Ren, M., Tu, J., Wang, J., & Urtasun, R. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2011.05289 · Cited by 40
“More recently, approaches based on RCNN [23, 24] learn this task end-to-end.”
MethodDeep Learning for Visual SLAM: The State-of-the-Art and Future Trends
Favorskaya, M. N. (2023) · Electronics · doi:10.3390/electronics12092006 · Cited by 43
UnDEMoN: Unsupervised Deep Network for Depth and Ego-Motion Estimation
Babu, V. M., Das, K., Majumdar, A., & Kumar, S. (2018) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2018.8593864 · Cited by 39
MagicVO: An End-to-End Hybrid CNN and Bi-LSTM Method for Monocular Visual Odometry
Jiao, J., Jiao, J., Mo, Y., Liu, W., & Deng, Z. (2019) · IEEE Access · doi:10.1109/access.2019.2926350 · Cited by 52
“In order to compensate for the shortcomings of CNN in processing successive image sequences, Wang and his colleges propose a way that introduces LSTM to process the VO problems, which allows the model to learn the information between successive images [8].”
BackgroundYao, S., Zhao, Y., Zhang, A., Su, L., & Abdelzaher, T. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1706.01215 · Cited by 39
“ese models have shown signicant improvements on audio sensing [31], tracking and localization [8, 38, 52, 56], human activity recognition [37, 56], and user identication [31, 56].”
BackgroundDeep Learning for Spacecraft Pose Estimation from Photorealistic Rendering
Proenca, P. F., & Gao, Y. (2019) · arXiv · Cited by 251
“DL has also been successfully applied to visual odometry [24, 25].”
BackgroundMethodInfluentialReview on SLAM algorithms for Augmented Reality
Sheng, X., Mao, S., Yan, Y., & Yang, X. (2024) · Displays · doi:10.1016/j.displa.2024.102806 · Cited by 41
Maciel-Pearson, B. G., Akçay, S., Atapour–Abarghouei, A., Holder, C. J., & Breckon, T. P. (2019) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2019.2930496 · Cited by 39
“Further enhancements in predicting camera position have been made possible by integrating a Long-Short Term Memory (LSTM) network into the process [12].”
BackgroundMethodInfluentialA Survey of Simultaneous Localization and Mapping.
Huang, B., Zhao, J., & Liu, J. (2019) · Cited by 38
Autonomous Driving with Deep Learning: A Survey of State-of-Art TechnologiesPreprint
Huang, Y., & Chen, Y. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2006.06091 · Cited by 84
Lai, T. (2022) · Sensors · doi:10.3390/s22197265 · Cited by 48
“For example, VINet is a sensor fusion network [101] that can use the estimated pose from DeepVO [102] along with the inertial sensor readings with an LSTM.”
MethodSparsePose: Sparse-View Camera Pose Regression and Refinement
Sinha, S., Zhang, J. Y., Tagliasacchi, A., Gilitschenski, I., & Lindell, D. B. (2023) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52729.2023.02045 · Cited by 41
“In visual odometry, we find techniques that use neural networks to auto-regressively estimate pose [65, 72], but these methods assume a small baseline between subsequent image pairs, rendering them unsuitable to the problem at hand.”
MethodSWformer-VO: A Monocular Visual Odometry Model Based on Swin Transformer
Wu, Z., & Zhu, Y. (2024) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2024.3384911 · Cited by 38
“Integrating deep learning into Visual Odometry (VO), exemplified by DeepVO, utilizes Recurrent Convolutional Neural Networks (RCNN) for improved performance [3].”
BackgroundMethodInfluentialKlenk, S., Motzet, M., Koestler, L., & Cremers, D. (2024) · International Conference on 3D Vision (3DV) · doi:10.1109/3dv62453.2024.00036 · Cited by 42
“Earlier approaches predict the pose end-to-end from images [54, 58, 67].”
Zhao, Z., Wu, C. W., Kong, X., Li, Q., Guo, Z., Lv, Z., & Du, X. (2025) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2025.3562319 · Cited by 44
“Wang et al. [10] proposed a novel end-to-end monocular VO framework using deep recurrent convolutional neural networks (RCNN) called DeepVO, the first end-to-end approach for monocular VO based on DNN.”
MethodEMA-VIO: Deep Visual–Inertial Odometry With External Memory Attention
Tu, Z., Chen, C., Pan, X., Liu, R., Cui, J., & Mao, J. (2022) · IEEE Sensors Journal · doi:10.1109/jsen.2022.3208200 · Cited by 36
“DeepVO [10] proposes an end-toend visual odometry based on deep neural networks, which uses Convolutional Neural Networks (ConvNets) to encode image sensor data into visual features, and a two-layer LSTM network to update visual features to provide pose in an end-toend manner.”
VIUNet: Deep Visual–Inertial–UWB Fusion for Indoor UAV Localization
Kao, P.-Y., Chang, H.-J., Tseng, K.-W., Chen, T., Luo, H.-L., & Hung, Y. (2023) · IEEE Access · doi:10.1109/access.2023.3279292 · Cited by 35
4DRVO-Net: Deep 4D Radar–Visual Odometry Using Multi-Modal and Multi-Scale Adaptive Fusion
Zhuo, G., Lu, S., Zhou, H., Zheng, L., Zhou, M., & Xiong, L. (2023) · IEEE Transactions on Intelligent Vehicles · doi:10.1109/tiv.2023.3330956 · Cited by 35
“DeepVO [7] adopts deep recurrent neural networks to capture the temporal dynamics and interdependencies among sequences, enabling ego-motion estimation.”
BackgroundMethodInfluentialDeep Learning for Underwater Visual Odometry Estimation
Teixeira, B., Silva, H. M., Matos, A. C., & Silva, E. (2020) · IEEE Access · doi:10.1109/access.2020.2978406 · Cited by 37
“VINet [77] is a proposed framework that fuses pose estimates from DeepVO [64] with inertial data, showing comparable performance to traditional fusion systems.”
BackgroundMethodEvolution of Visual Odometry TechniquesPreprint
Poddar, S., Kottath, R., & Karar, V. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1804.11142 · Cited by 35
ATVIO: Attention Guided Visual-Inertial Odometry
Liu, L., Li, G., & Li, T. H. (2021) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp39728.2021.9413912 · Cited by 35
GSLAM: A General SLAM Framework and Benchmark
Zhao, Y., Xu, S., Bu, S., Jiang, H., & Han, P. (2019) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2019.00120 · Cited by 34
“In recent years, SLAM technology develops rapidly and a variety of SLAM systems have been proposed, including monocular SLAM system (key-point based [12, 37, 49], direct [15, 16, 53] and semi-direct methods [22, 23]), multisensor SLAM (RGBD [7, 36, 68], Stereo [17, 23, 51] and inertial aided methods [45, 56, 66]), lea…”
An Adaptive Visual Dynamic-SLAM Method Based on Fusing the Semantic Information
Jiao, J., Wang, C., Li, N., Deng, Z., & Xu, W. D. (2021) · IEEE Sensors Journal · doi:10.1109/jsen.2021.3051691 · Cited by 34
“Given two 113 continuous pictures, [14] can directly estimate the transforma114 tion between these two frames by combining CNN and SLAM.”
MethodUnsupervised Monocular Visual Odometry Based on Confidence Evaluation
Liu, Y., Wang, H., Wang, J., & Wang, X. (2021) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2021.3053412 · Cited by 34
Li, Z., Shang, T., Xu, P., & Deng, Z. (2025) · Artificial Intelligence Review · doi:10.1007/s10462-025-11367-8 · Cited by 39
“More recently, from 2018 onward, deep learning methods like DeepVO (Wang et al. 2017), VLocNet++ (Rad-wan et al. 2018), and Transformer-based SLAM have emerged (Liu et al. 2023), enabling semantic understanding and multi-sensor fusion (Lidar, IMU, vision) .”
Leveraging Deep Learning for Visual Odometry Using Optical Flow
Pandey, T., Peña, D., Byrne, J., & Moloney, D. (2021) · Sensors · doi:10.3390/s21041313 · Cited by 33
“Wang et al. [24] were the first ones to propose a sequence-based model.”
BackgroundMethodResultInfluentialFalcone, R., Ciaramella, A., Carrabs, F., Strisciuglio, N., & Martinelli, E. (2022) · Structures · doi:10.1016/j.istruc.2022.05.008 · Cited by 37
Adaptive VIO: Deep Visual-Inertial Odometry with Online Continual Learning
Pan, Y., Zhou, W., Cao, Y., & Zha, H. (2024) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52733.2024.01706 · Cited by 34
“End-to-end learning-based VO utilizes pose and depth estimation networks to replace classic modules of track ing and mapping, trained in either a supervised [40, 43] or self-supervised [18, 45] manner.”
Chen, L., Li, G., Xie, W., Tan, J., Li, Y., Pu, J., Chen, L., Gan, D., & Shi, W. (2024) · Energies · doi:10.3390/en17205177 · Cited by 37
An Improved Method Based on Deep Reinforcement Learning for Target Searching
Wei, X., Huang, X. L., Lü, T., & Song, G. (2019) · International Conference on Robotics and Automation Engineering (ICRAE) · doi:10.1109/icrae48301.2019.9043821 · Cited by 32
“Various methods [12][13][14] have been developed and tested on Microsoft’s AirSim, which is an open source High-fidelity system for simulating and testing, that aims to narrow the gap between simulation and reality in order to aid the development of autonomous vehicles.”
MethodTowards End-to-end Learning of Visual Inertial Odometry with an EKF
Li, C., & Waslander, S. L. (2020) · Canadian Conference on Computer and Robot Vision · doi:10.1109/crv50864.2020.00033 · Cited by 33
“This allows state estimation to be learned directly from data in an end-to-end manner, leveraging both local and higher level features, demonstrated in previous works, such as [3], [4], and [5].”
MethodChappellet, K., Murooka, M., Caron, G., Kanehiro, F., & Kheddar, A. (2023) · IEEE Transactions on Automation Science and Engineering · doi:10.1109/tase.2023.3283497 · Cited by 31
“…is mapped, the map can be reused for localization-only, see e.g., [1]. vSLAM systems have been made for various types of cameras, ranging from the most conventional in seminal vSLAM works [29], [30], to stereo vision [31], panoramic [32], 360 [33] and active RGB-D [34], [35], [36], [37] vision.”
BackgroundDeepTAM: Deep Tracking and Mapping with Convolutional Neural Networks
Zhou, H., Ummenhofer, B., & Brox, T. (2019) · International Journal of Computer Vision · doi:10.1007/s11263-019-01221-0 · Cited by 31
“DeepVO (Wang et al. 2017) runs a deep network for visual odometry, i.”
BackgroundInfluentialA Survey of Simultaneous Localization and Mapping with an Envision in 6G Wireless NetworksPreprint
Huang, B., Zhao, J., & Liu, J. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1909.05214 · Cited by 31
“DeepVO (Monocular) [182] presents a novel end-to-end framework for monocular VO by using deep Recurrent Convolutional Neural Networks (RCNNs).”
MethodNeurAll: Towards a Unified Visual Perception Model for Automated Driving
Sistu, G., Leang, I., Chennupati, S., Yogamani, S. K., Eising, C., Milz, S., & Rawashdeh, S. A. (2019) · IEEE International Conference on Intelligent Transportation Systems (ITSC) · doi:10.1109/itsc.2019.8917043 · Cited by 36
“[19] use a combination of”
Learning Pose Estimation for UAV Autonomous Navigation and Landing Using Visual-Inertial Sensor Data
Baldini, F., Anandkumar, A., & Murray, R. M. (2020) · Proceedings of the ... American Control Conference/Proceedings of the American Control Conference · doi:10.23919/acc45564.2020.9147400 · Cited by 31
DeepCrashTest: Turning Dashcam Videos into Virtual Crash Tests for Automated Driving Systems
Bashetty, S. K., Amor, H. B., & Fainekos, G. (2020) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra40945.2020.9197053 · Cited by 31
Kechagias‐Stamatis, O., Aouf, N., Dubanchet, V., & Richardson, M. A. (2020) · Acta Astronautica · doi:10.1016/j.actaastro.2020.07.034 · Cited by 30
“This is because as the CNN depth increases, the features extracted by each convolutional layer become finer, and thus more channels are required to learn these features [52].”
Transformer guided geometry model for flow-based unsupervised visual odometry
Li, X., Hou, Y., Wang, P., Gao, Z., Xu, M., & Li, W. (2021) · Neural Computing and Applications · doi:10.1007/s00521-020-05545-8 · Cited by 30
“Several attempts [45, 46, 51] have been made to use RNN or long short-term memory (LSTM) to learn and enforce the temporal correlation over a long period.”
BackgroundMethodResultInfluentialSelf-supervised Visual-LiDAR Odometry with Flip Consistency
Li, B., Hu, M. H., Wang, S., Wang, L., & Gong, X. (2021) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv48630.2021.00389 · Cited by 28
“In addition, all proposed components including the data augmentation strategy, the siamese architecture, and the flip consistency loss, gradually boost the performance in both VLOs and VOs, demonstrating their effectiveness.”
BackgroundMethodInfluentialSelf-citationMonoEF: Extrinsic Parameter Free Monocular 3D Object Detection
Zhou, Y., He, Y., Zhu, H., Wang, C., Li, H., & Jiang, Q. (2021) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2021.3136899 · Cited by 31
“With the success of deep neural networks, end-to-end learning-based methods [54], [55], [59], [60] have been proposed to tackle the visual odometry problem.”
BackgroundExploration-Based SLAM (e-SLAM) for the Indoor Mobile Robot Using Lidar
Ismail, H., Roy, R., Sheu, L.-J., Chieng, W.-H., & Tang, L. (2022) · Sensors · doi:10.3390/s22041689 · Cited by 32
Aerial Robots, Visual-Inertial Odometry of
Scaramuzza, D., & Zhang, Z. (2020) · Springer eBooks · doi:10.1007/978-3-642-41610-1_71-1 · Cited by 29
“In this respect, the recent development of deep visual odometry (Costante et al. 2016; Wang et al. 2017; Zhou et al. 2017) has shown promising initial results, especially in addressing open challenges with standard cameras, such as dealing with the aperture problem, motion blur, defocus, and…”
BackgroundDeep geometry-aware camera self-calibration from video
Hagemann, A., Knorr, M., & Stiller, C. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.00318 · Cited by 29
“While classical SfM pipelines generalize well to unseen environments, they typically rely on handcrafted features and thus do not leverage the capabilities of learned features which have proven effective across a variety of 3D computer vision tasks [43, 40, 48, 32].”
Background
Metadata
- Authors
- Wang, S., Clark, R., Wen, H., & Trigoni, N.
- Affiliations
- University of Oxford (GB)
- Venue
- 2017 IEEE International Conference on Robotics and Automation (ICRA)
- Publisher
- IEEE
- Published
- 2017-05
- Pages
- 2043-2050
- ISSN
- 1050-4729 (ISSN-L)
- OpenAlex
- W2598706937
- Cited by
- 970
- References
- 29
- Citations by year
- 2026: 63 · 2025: 120 · 2024: 104 · 2023: 123 · 2022: 130 · 2021: 131 · 2020: 125 · 2019: 116 · 2018: 52 · 2017: 6
- FWCI
- 143.15
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- Proceedings - IEEE International Conference on Robotics and Automation/Proceedings (publishedVersion)
- arXiv (Cornell University) (publishedVersion)
- Warwick Research Archive Portal (University of Warwick) (submittedVersion)
- arXiv (Cornell University)
- Funders
- Engineering and Physical Sciences Research Council (EP/M019918/1)
- 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 Vision and Imaging
- Physical Sciences › Earth and Planetary Sciences › Geology › 3D Surveying and Cultural Heritage
- Keywords
- monocular visual odometry, end-to-end learning, recurrent convolutional neural network, pose estimation, feature representation learning
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
