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
Research works printed but not linked: 1
- zaremba, 2014, Learning to execute
Cited by
Sorted by citations
Items 501–600 of 969
Xiao, X., Zhang, Z., Xia, G.-S., Shao, Z., Gong, J., & Li, D. (2025) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2025.3559983 · Cited by 5
“Various supervised [45], [46], [47] and unsupervised [48], [49], [50] methods have emerged, offering new perspectives.”
Geometry-Constrained Monocular Scale Estimation Using Semantic Segmentation for Dynamic Scenes
Zhang, H., Wu, Z., Li, H., Shangguan, Q., & An, K. (2025) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2025.3625332 · Cited by 5
“The first treats MVO as an end-to-end process [29], [30], directly obtaining pose estimates with absolute scale from monocular frames and trained models.”
Visual Inertial System for ROV Positioning
Marques, F. M., Castro, F., Parente, M. A., & Costa, P. (2020) · Abu Dhabi International Petroleum Exhibition & Conference (ADIPEC) · doi:10.2118/203395-ms · Cited by 5
A Comparative Evaluation of Learned Feature Descriptors on Hybrid Monocular Visual SLAM Methods
Bruno, H. M. S., & Colombini, E. L. (2020) · Latin American Robotics Symposium, Brazilian Symposium on Robotics and Workshop on Robotics in Education (LARS/SBR/WRE) · doi:10.1109/lars/sbr/wre51543.2020.9307033 · Cited by 5
“Wang et. al. proposed one of the most notable deep-learning end-to-end VO approaches, called DeepVO [2].”
Unsupervised Deep Learning of Depth, Ego-Motion, and Optical Flow from Stereo Images
Yang, D., Luo, Z., Shang, P., & Hu, Z. (2021) · International Conference on Traffic and Logistic Engineering (ICTLE) · doi:10.1109/ictle53360.2021.9525746 · Cited by 5
Multi-view Monocular Depth and Uncertainty Prediction with Deep SfM in Dynamic Environments
Homeyer, C., Lange, O. E. D., & Schnörr, C. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-09037-0_31 · Cited by 5
Liu, X., Shen, F., Zhao, J., & Nie, C. (2023) · The Visual Computer · doi:10.1007/s00371-023-02840-y · Cited by 5
“Supervised learning of depth estimation, VO, and optical flow estimation With recent development of deep learning, great progress has been made in many tasks of 3D geometry understanding, including depth estimation [6, 7, 31], VO [9, 10, 32–37], optical flow estimation [11–14].”
The Effective 3D MRI Reconstruction Method Driven by the Fusion Strategy in NSST Domain
Huang, J., Wang, L., Tahir, M. D., Cheng, T., Guo, X., Wang, Y., & Liu, C. (2023) · International Journal of Advanced Computer Science and Applications · doi:10.14569/ijacsa.2023.0140477 · Cited by 5
“proposed the famous DeepVO model [15], where the recurrent convolutional neural network is employed to infer the pose directly from the source videos without any traditional visual odometry, improving the visual odometry loop in 3D reconstruction [16].”
MethodStereoVO: Learning Stereo Visual Odometry Approach Based on Optical Flow and Depth Information
Duan, C., Junginger, S., Thurow, K., & Liu, H. (2023) · Applied Sciences · doi:10.3390/app13105842 · Cited by 5
Image based and Point Cloud based Methods for 3D View Reconstruction in Real-time Environment
Agrawal, A., Sharma, T., Verma, N. K., & Vasikarla, S. (2024) · IEEE Annual Computing and Communication Workshop and Conference (CCWC) · doi:10.1109/ccwc60891.2024.10427868 · Cited by 5
“Algorithms such as PointNet [11], Faster R-CNN [28], [29], DeepVO [30], and MV3D [31] leverage this dataset [26] for tasks like point cloud processing, object detection, visual odometry, and 3D object detection in autonomous driving scenarios.”
MethodStereo-RIVO: Stereo-Robust Indirect Visual Odometry
Salehi, E., Aghagolzadeh, A., & Hosseini, R. (2024) · Journal of Intelligent & Robotic Systems · doi:10.1007/s10846-024-02116-0 · Cited by 5
“In model-based VO/VSLAM methods, supervised [14, 28, 29] and self-supervised [15, 30] learning architectures are commonly used.”
Visual Odometry with Deep Bidirectional Recurrent Neural Networks
Xue, F., Wang, X., Wang, Q., Wang, J., & Zha, H. (2019) · Lecture notes in computer science · doi:10.1007/978-3-030-31726-3_20 · Cited by 5
Semi-Direct Visual Odometry Based on Monocular Depth Estimation
Guo, S., Guo, J., & Bai, C. (2019) · IEEE International Conference on Unmanned Systems (ICUS) · doi:10.1109/icus48101.2019.8996049 · Cited by 5
“However, this learning-based method is not going to be a replacement to classic geometric method [14], their system is not guaranteed to keep in high performance in untrained, unfamiliar environments.”
BackgroundSelfSphNet: Motion Estimation of a Spherical Camera via Self-Supervised Learning
Kim, D., Pathak, S., Moro, A., Yamashita, A., & Asama, H. (2020) · IEEE Access · doi:10.1109/access.2020.2977109 · Cited by 5
“Learning-based approaches to estimate the motion of spherical cameras have recently explored the trajectories of cameras [10], [11].”
BackgroundA Global Occlusion-Aware Approach to Self-Supervised Monocular Visual Odometry
Lu, Y., Xu, X., Ding, M., Lu, Z., & Xiang, T. (2021) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v35i3.16325 · Cited by 5
“Adopting a deep encoder-decoder architecture, most existing deep SSM-VO methods (Zhou et al. 2017; Wang et al. 2017; Bian et al. 2019; Luo et al. 2018) use skip connections to concatenate multi-level features computed over different layers.”
MethodA benchmark analysis of data‐driven and geometric approaches for robot ego‐motion estimation
Legittimo, M., Felicioni, S., Bagni, F., Tagliavini, A., Dionigi, A., Gatti, F., Verucchi, M., Costante, G., & Bertogna, M. (2023) · Journal of Field Robotics · doi:10.1002/rob.22151 · Cited by 5
“The deep learning paradigm has been further exploited by LS ‐ VO (Costante & Ciarfuglia, 2018), DeepVO (Wang et al., 2017), and UA ‐ VO (Costante & Mancini, 2020).”
MethodImproving visual odometry pipeline with feedback from forward and backward motion estimates
Sardana, R., Karar, V., & Poddar, S. (2023) · Machine Vision and Applications · doi:10.1007/s00138-023-01370-w · Cited by 5
“The supervised schemes for VO proposed in [15–17] require the ground truth poses of each frame for learning, which are usually acquired through a GPS/ IMU system or a motion capture system.”
MethodInsights into Batch Selection for Event-Camera Motion Estimation
Valerdi, J. L., Bartolozzi, C., & Glover, A. (2023) · Sensors · doi:10.3390/s23073699 · Cited by 5
“Visual changes have been estimated in event-driven deep networks by providing image sequences [2], by explicitly storing the state used for comparison to individual images [3], or by inherently storing the state in a recursive neural network [4].”
MethodDEEP LEARNING TO SUPPORT 3D MAPPING CAPABILITIES OF A PORTABLE VSLAM-BASED SYSTEM
Padkan, N., Battisti, R., Menna, F., & Fabio Remondino, A. (2023) · 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-1-w1-2023-363-2023 · Cited by 5
“Learning-based vSLAM methods rely on deep learning-based feature descriptors (Bruno and Colombini, 2021), hybrid methods (Tang et al., 2019) or complete end-to-end approaches (Wang et al., 2017).”
MethodDeep Homography Estimation with Pairwise Invertibility Constraint
Wang, X., Wang, C., Bai, X. P., Liu, Y., & Zhou, J. W. (2018) · Lecture notes in computer science · doi:10.1007/978-3-319-97785-0_20 · Cited by 5
Visual-Inertial Odometry for Unmanned Aerial Vehicle using Deep Learning
Lee, H., McCrink, M. H., & Gregory, J. (2019) · AIAA SciTech Forum · doi:10.2514/6.2019-1410 · Cited by 5
Kim, D., Pathak, S., Moro, A., Komatsu, R., Yamashita, A., & Asama, H. (2019) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp.2019.8682203 · Cited by 5
“Recently, Convolutional Neural Networks (CNNs) have shown powerful abilities even in complex regression problems such as camera motion estimation [5, 6, 7] and camera relocalization [8].”
BackgroundDeep Monocular Visual Odometry for Ground Vehicle
Wang, X., & Zhang, H. (2020) · IEEE Access · doi:10.1109/access.2020.3025557 · Cited by 5
“The models of Zhan et al. [13], DeepVO [8], and our method are trained with the absolute scale, so no alignment is required.”
BackgroundMethodInfluentialMonocular visual odometry for underground railway scenarios
Etxeberria-Garcia, M., Labayen, M., Eizaguirre, F., Zamalloa, M., & Arana-Arexolaleiba, N. (2021) · International Conference on Quality Control by Artificial Vision (QCAV) · doi:10.1117/12.2586310 · Cited by 5
Temporal-Aware SfM-Learner: Unsupervised Learning Monocular Depth and Motion from Stereo Video Clips
Zhang, L., Li, G., & Li, T. H. (2020) · IEEE International Conference on Multimedia Information Processing and Retrieval (MIPR) · doi:10.1109/mipr49039.2020.00059 · Cited by 5
“Wang et al. then extended into a new end-to-end architecture [9], resulting in competitive results against geometric VO systems.”
BackgroundSarfi, M. H., & Bisheban, M. (2025) · Journal of Intelligent & Robotic Systems · doi:10.1007/s10846-025-02235-2 · Cited by 4
“DRL techniques are used in many different research areas including decision-making such as visual odometry [17], indoor navigation [18, 19], autonomous navigation with the limited field of view [20], visual object search and localization [21], and many other applications.”
Kang, J. (2025) · Sensors · doi:10.3390/s25196050 · Cited by 4
S 3 E: Self-Supervised State Estimation for Radar-Inertial System
Wang, S., Xie, Y., Liao, Q., & Wang, W. (2025) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51701.2025.02477 · Cited by 4
“Specifically, they convert radar point clouds into a depth map, extracting spatial features by [36] from the depth map and IMU time series, and feeding them into the pose regression network.”
Self-citationMD2VO: Enhancing Monocular Visual Odometry through Minimum Depth Difference
Li, P., Tang, C., Duan, Y., & Li, Z. (2024) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn60899.2024.10649955 · Cited by 4
Park, S. H., Jeong, W.-J., Manawadu, M., & Park, S.-Y. (2025) · Applied Sciences · doi:10.3390/app15031501 · Cited by 4
Deep learning-based visual SLAM for indoor dynamic scenes
Xu, Z., Song, Y. H., Pang, B. J., Xu, Q., & Yuan, X. (2025) · Applied Intelligence · doi:10.1007/s10489-024-05886-0 · Cited by 4
“DeepVO [14] utilizes CNNs for feature extraction and LSTM for continuous pose estimation.”
An Online Adaptation Method for Robust Depth Estimation and Visual Odometry in the Open World
Ji, X., Niu, H., Duan, D., Ying, R., Wen, F., & Liu, P. (2025) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2025.3602581 · Cited by 4
“Meanwhile, these works [22], [23] propose attention mechanisms within the recurrent neural networks (RNNs) to extract spatial and temporal features for pose estimation.”
Zheng, D., Zhao, Y., Tian, W., Zhang, F., Gui, G., & Gui, W. (2025) · IEEE Transactions on Automation Science and Engineering · doi:10.1109/tase.2025.3643622 · Cited by 4
“Wang et al. introduced DeepVO, the first end-to-end supervised visual odometry model using CNNs for geometric feature learning and RNNs for temporal modeling [22].”
Ren, H., Li, C., Zhang, X., Ding, C., Man, C., & Yu, H. (2021) · WRC Symposium on Advanced Robotics and Automation (WRC SARA) · doi:10.1109/wrcsara53879.2021.9612673 · Cited by 4
Eliminating Scale Ambiguity of Unsupervised Monocular Visual Odometry
Wang, Z., Shen, M., & Chen, Q. (2023) · Neural Processing Letters · doi:10.1007/s11063-023-11224-1 · Cited by 5
A comparison of deep learning-based visual odometry algorithms in challenging scenarios
Bruno, H. M. S., Cabral, K., Colombini, E. L., & Givigi, S. N. (2024) · IEEE International Systems Conference (SysCon) · doi:10.1109/syscon61195.2024.10553552 · Cited by 5
“These systems can replace the entire classical VO pipeline [6]–[8].”
BackgroundAutonomous Location and Obstacle Avoidance of Inspection Robots Based on Multi-modal Information
Xu, M., Li, C., Liu, C., Wang, S., & Tian, Y. (2024) · International Conference on Energy, Power and Electrical Technology (ICEPET) · doi:10.1109/icepet61938.2024.10626217 · Cited by 4
“Wang et al. [5] proposed a visual odometry composed of a convolutional neural network (CNN) and a recurrent neural network (RNN).”
BackgroundSelf-citationNavigating the Unknown: Uncertainty-Aware Compute-in-Memory Autonomy of Edge Robotics
Darabi, N., Shukla, P., Jayasuriya, D., Kumar, D., Stutts, A. C., & Trivedi, A. R. (2024) · Design, Automation, and Test in Europe · doi:10.23919/date58400.2024.10546628 · Cited by 4
RWKV-VIO: An Efficient and Low-Drift Visual–Inertial Odometry Using an End-to-End Deep Network
Yang, J., Xu, X., Xu, Z., Wu, Z., & Chu, W. (2025) · Sensors · doi:10.3390/s25185737 · Cited by 5
“DeepVO [30] used deep recurrent convolutional neural networks (DRCNNs) to learn motion directly from image sequences, eliminating manual feature engineering.”
Deep Learning Approaches Assessment for Underwater Scene Understanding and Egomotion Estimation
Teixeira, B., Silva, H. M., Matos, A. C., & Silva, E. (2019) · MTS/IEEE OCEANS Conference (OCEANS) · doi:10.23919/oceans40490.2019.8962872 · Cited by 4
“After training with unlabeled stereo images, UnDeepVO can simultaneously perform visual odometry and depth estimation with monocular images.”
BackgroundMethodInfluentialZhang, X., He, Z., Ma, Z., Jun, P., & Yang, K. (2021) · Sensors · doi:10.3390/s21186302 · Cited by 5
Camera pose estimation framework for array‐structured images
Shin, M.-J., Park, W., Kim, J. H., Kim, J., Yun, K., & Kang, S. (2022) · ETRI Journal · doi:10.4218/etrij.2021-0303 · Cited by 4
Pal, S., Gupta, S., Das, N., & Ghosh, K. (2022) · Journal of Autonomous Vehicles and Systems · doi:10.1115/1.4055161 · Cited by 4
A siamese network-based approach for vehicle pose estimation
Zhao, H., Tao, B., Huang, L., & Chen, B. (2022) · Frontiers in Bioengineering and Biotechnology · doi:10.3389/fbioe.2022.948726 · Cited by 4
Zhuang, Y., Jiang, X., Gao, Y., Fang, Z., & Fujita, H. (2022) · Sensors · doi:10.3390/s22249647 · Cited by 4
“As shown in Table 5, in that scene, the deep learning methods including SFM-Learner [19], DeepVO-Feat [44], CC [21] and SC-SfMLearner [5] had a more serious trajectory drift than in the previous medium-amplitude-motion scene.”
MethodInfluentialRecent Advances in Mobile Robot Localization in Complex Scenarios
Zhang, H., Xu, X., Liu, C., Wang, W., Ran, Y., & Tan, Z. (2023) · Lecture notes in electrical engineering · doi:10.1007/978-981-19-6613-2_328 · Cited by 4
A FRAMEWORK FOR MULTI-SENSOR POSITIONING AND MAPPING FOR AUTONOMOUS VEHICLES
Noureldin, A. M., & Elhabiby, M. (2023) · 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-1-w1-2023-339-2023 · Cited by 4
“Moreover, GNSS has other limitations in urban areas, such as multipath and signal blockage, where backup systems are necessary [14,15].”
BackgroundLearning agriculture keypoint descriptors with triplet loss for visual SLAM
Tanco, M. M., Tejera, G., & Martino, J. M. D. (2023) · Journal of Ambient Intelligence and Humanized Computing · doi:10.1007/s12652-023-04681-y · Cited by 4
Highly Dynamic Visual SLAM Dense Map Construction Based on Indoor Environments
Mei, J., Zuo, T., & Song, D. H. (2024) · IEEE Access · doi:10.1109/access.2024.3374523 · Cited by 4
“Wang et al. [21] By proposing a monocular end-to-end framework based on deep recurrent convolutional neural networks, it has some resistance to the influence of dynamic targets and is less effective for fast moving objects.”
BackgroundResearch on SLAM Intelligent Robot Based on Visual Laser Fusion
Lin, J., Liang, T., Ma, G., Li, B., & Yu, T. (2024) · International Conference on Control, Robotics and Cybernetics (CRC) · doi:10.1109/crc60659.2023.10488663 · Cited by 4
“In the field of autonomous driving [4,5,6] , the visual odometer function provided by SLAM technology can be integrated with other positioning methods to achieve a more accurate driving experience.”
MethodDecision Tree-Based Direction Detection Using IMU Data in Autonomous Robots
Apaydın, N. N., Kılıç, İ., Apaydın, M., & Yaman, O. (2024) · Batman Üniversitesi Yaşam Bilimleri Dergisi · doi:10.55024/buyasambid.1501521 · Cited by 4
Semantic Visual SLAM Algorithm Based on Improved DeepLabV3+ Model and LK Optical Flow
Li, Y., Wang, Y., Lu, L., Guo, Y., & An, Q. (2024) · Applied Sciences · doi:10.3390/app14135792 · Cited by 4
Euler Angles Based Loss Function for Camera Localization with Deep Learning
Fang, Q., Zhao, K., Tang, D., Zhou, Z., Zhou, Y., Hu, T., & Zhou, H. (2018) · IEEE International Conference on Cyber Technology in Automation, Control, and Intelligent Systems · doi:10.1109/cyber.2018.8688359 · Cited by 4
Classification of Streetsigns Using Gaussian Process Latent Variable Models
Wöber, W., Aburaia, M., & Olaverri-Monreal, C. (2019) · International Conference on Connected Vehicles and Expo · doi:10.1109/iccve45908.2019.8964883 · Cited by 4
“In addition to object detection, movement estimation [22] or segmentation (e.g.: pixel-wise weed detection [23]) are typical applications of CNNs.”
BackgroundScale-Adaptive Neural Dense Features: Learning via Hierarchical Context Aggregation
Spencer, J., Bowden, R., & Hadfield, S. (2019) · Surrey Research Insight Open Access (The University of Surrey) · Cited by 4
Visual Odometry for Self-Driving with Multihypothesis and Network Prediction
Chen, Y., Mei, Y., & Wan, S.-F. (2021) · Mathematical Problems in Engineering · doi:10.1155/2021/1930881 · Cited by 4
“Particularly, the data-driven end-to-end visual odometrymethods [7, 8] are more straightforward and simpler than the model-based methods, while the accuracy and generalization are still inferior to the model-based methods.”
MethodRyad, I., Zidan, M., Rashad, N., Bakr, D., Bakr, N., Yehia, N., Ismail, Y., AbdelSalam, M., & Salem, A. (2021) · IEEE International Conference on Electronics, Circuits and Systems (ICECS) · doi:10.1109/icecs53924.2021.9665583 · Cited by 4
“Recent research work [2, 3, 4, 5, 6, 7, 8] has investigated training and testing AI Agents of self-driving cars, drones, and Automated Guided Vehicles (AGVs) with synthetically generated data in digital twin simulators, which can be produced in bulk with correct labels.”
Multi-Sensor Fusion Self-Supervised Deep Odometry and Depth Estimation
Wan, Y., Zhao, Q., Guo, C., Xu, C., & Fang, L. (2022) · Remote Sensing · doi:10.3390/rs14051228 · Cited by 5
“Table 2 reports the evaluation results of the DeepVIO output poses, and compares them to the previous works, such as ORB-SLAM2 [36], Deep-VO-Feat [52], SFM-Learner [51], SCSfMLearner [60].”
MethodInfluentialLim, H., & Myung, H. (2019) · Communications in computer and information science · doi:10.1007/978-981-13-7780-8_12 · Cited by 4
He, X., Zhang, S., & Liu, P. (2019) · International Conference on Algorithms, Computing and Artificial Intelligence (ACAI) · doi:10.1145/3377713.3377731 · Cited by 4
“At the same time, there are some ways to directly use RGB images as input, using a well-trained network to output camera poses, so that the output results only depend on the network, thus eliminating the effects of dynamic objects[17].”
MethodASD-SLAM: A Novel Adaptive-Scale Descriptor Learning for Visual SLAM
Ma, T., Wang, Y., Wang, Z., Liu, X., & Zhang, H. (2020) · IEEE Intelligent Vehicles Symposium · doi:10.1109/iv47402.2020.9304626 · Cited by 4
UnDEMoN 2.0: Improved Depth and Ego Motion Estimation through Deep Image SamplingPreprint
Babu, V. M., Kumar, S., Majumder, A., & Das, K. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1811.10884 · Cited by 4
“Most of the pioneering works in this direction learn depth from images in a supervised manner by using ground-truth data from depth sensors [26, 3, 15, 20, 11, 27].”
BackgroundDeep Model-Based 6D Pose Refinement in RGBPreprint
Manhardt, F., Kehl, W., Navab, N., & Tombari, F. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1810.03065 · Cited by 4
WGANVO: Monocular Visual Odometry based on Generative Adversarial NetworksPreprint
Cremona, J., Uzal, L. C., & Pire, T. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2007.13704 · Cited by 4
“DeepVO [15] addresses the problem in an end-to-end manner.”
Kinematics-Guided Reinforcement Learning for Object-Aware 3D Ego-Pose EstimationPreprint
Luo, Z., Hachiuma, R., Yuan, Y., Iwase, S., & Kitani, K. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2011.04837 · Cited by 5
“We use a monocular camera tracking technique, such as Visual Inertial Odometry (VIO) [22, 2], to extract the camera motion and fine-tune our learned policy to match against it.”
MethodA Survey on 3D LiDAR Localization for Autonomous VehiclesPreprint
Elhousni, M., & Huang, X. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2006.00648 · Cited by 4
“2D camera images were first used to try and predict the odometry between a pair of images [5], [6], [7], [8] with results that were more or less acceptable, but still not outperforming the state of the art.”
Eising, C., Horgan, J., & Yogamani, S. K. (2021) · arXiv (Cornell University) · Cited by 4
Unsupervised Joint Learning of Depth, Optical Flow, Ego-motion from VideoPreprint
Li, J., Zhao, J., Song, S., & Feng, T. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2105.14520 · Cited by 4
“PoseNet returns the global pose, and then DeepVO [12] proposed a method to estimate the relative pose using a neural network.”
Non-Cooperative Spacecraft Pose Estimation Using Convolutional Neural Networks
Despond, F. (2021) · doi:10.22215/etd/2021-14867 · Cited by 4
Review of augmented reality maintenance and registration in aerospace industry
Zhao, X. L., Liu, K., Fu, H., & Wang, Y. (2025) · Journal of Manufacturing Processes · doi:10.1016/j.jmapro.2025.07.063 · Cited by 4
Robust Visual Odometry Using Rigidly-Bundled Arbitrarily-Arranged Multi-Cameras
Yu, H., Wang, J., He, Y., Yang, W. T., & Xia, G.-S. (2025) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2025.3621975 · Cited by 3
“Additionally, learning-based methods for feature extraction and tracking [11] [12] have been proposed to enhance feature robustness while reducing the computational burden.”
Machine learning techniques enabling advanced autonomous vehicle navigation
Shrivastava, A., Saxena, G., Ranjan, A., & Pradhan, A. (2026) · Proceedings of the Institution of Mechanical Engineers Part D Journal of Automobile Engineering · doi:10.1177/09544070251408897 · Cited by 3
Pereira, T., Viegas, C. X. P., Soares, S., & Ferreira, N. (2026) · Robotics · doi:10.3390/robotics15020035 · Cited by 6
“Based on [117].”
InfluentialInertialNet: Inertial Measurement Learning for Simultaneous Localization and Mapping
Lin, H., Liu, T.-A., & Lin, W.-Y. (2023) · Sensors · doi:10.3390/s23249812 · Cited by 3
“The objective of the optical-flow-based approaches (such as VINet [27] and DeepVO [26]) aims to improve the generalization capability of the networks.”
BackgroundMethodSelf-TIO: Thermal-Inertial Odometry via Self-Supervised 16-Bit Feature Extractor and Tracker
Lee, J., Ando, T., Shinozaki, M., Kitajima, T., An, Q., & Yamashita, A. (2024) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2024.3518303 · Cited by 3
Adaptive Visual-Aided 4D Radar Odometry Through Transformer-Based Feature Fusion
Zhang, Y., Xiao, R., Hong, Z., Hu, L., & Liu, J. (2024) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros58592.2024.10802367 · Cited by 3
“…cloud-based (ICP [29], NDT [30]), classical visual (ORB-SLAM3 [31]); • Learning based algorithms: deep supervised learning-based visual (DeepVO [32]), deep unsupervised learning-based visual (SfmLearner [9]), and learning-based 4D radar SOTA (CMFlow [6]) algorithms; • Our method with radar…”
Wang, Z., Yin, M., Zhu, X., Zhang, Y., Bei, S., & Bo, Y. (2025) · IEEE Sensors Journal · doi:10.1109/jsen.2025.3567672 · Cited by 3
“Deep visual odometry (DeepVO) [22] uses CNNs to directly predict camera pose changes from image sequences, eliminating the need for handcrafted features.”
SeeTree - a modular, open-source system for tree detection and orchard localization
Brown, J., Grimm, C. M., & Davidson, J. R. (2025) · Smart Agricultural Technology · doi:10.1016/j.atech.2025.101074 · Cited by 3
Xu, Z., Zhang, J., Haroutounian, S., Liu, H., Cao, Z., Messner, G. R., Alaverdyan, H., Ahuja, S., Koshy, R., Hanns, J., et al. (2025) · Proceedings of the ACM on Interactive Mobile Wearable and Ubiquitous Technologies · doi:10.1145/3729488 · Cited by 3
“These uncertainty measures are essential for assessing the reliability of segmentation results, identifying areas falsely segmented, and guiding human experts in the refinement process [88], allowing clinical professionals to judge whether an automated diagnosis can be trusted [7].”
Li, R., Wang, Y., Xu, X., Li, F., Tang, F., & Pan, X. (2025) · Cognitive Neurodynamics · doi:10.1007/s11571-025-10386-z · Cited by 3
“Further studies show that place cells also support efficient path planning through energy-based coding mechanisms (Wang et al. 2017).”
Predicting Out-of-View Feature Points for Model-Based Camera Pose Estimation
Moolan-Feroze, O., & Calway, A. D. (2018) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2018.8594297 · Cited by 3
Optical Flow Assisted Monocular Visual Odometry
Wan, Y., Gao, W., & Wu, Y. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-41404-7_26 · Cited by 3
WGANVO: odometría visual monocular basada en redes adversarias generativas
Cremona, J., Uzal, L. C., & Pire, T. (2021) · Revista Iberoamericana de Automática e Informática Industrial RIAI · doi:10.4995/riai.2022.16113 · Cited by 3
“En relación a otros sistemas que apliquen CNNs para resolver el problema de VO, resulta de interés una comparación con otro sistema end-to-end monocular que estime la pose de forma online, tal como es el caso de DeepVO (Wang et al., 2017).”
BackgroundInfluentialAn Unsupervised Monocular Visual Odometry Based on Multi-Scale Modeling
Zhi, H., Yin, C., Li, H., & Pang, S. (2022) · Sensors · doi:10.3390/s22145193 · Cited by 3
“Wang et al. proposed DeepVO [7] in 2017, which uses a two-layer LSTM to process sequence information and realizes the learning of image sequence correlation.”
InfluentialMapping method of single LiDAR for indoor degraded environment
Zijiang, Z., Wu, J., Cheng, H., Lin, J., & Shuo, Y. (2022) · Computers & Electrical Engineering · doi:10.1016/j.compeleceng.2022.108284 · Cited by 3
A Novel End-to-End Visual Odometry Framework Based on Deep Neural Network
Wang, Y., Cao, R., Guan, Y., & Zhang, X. (2022) · International Conference on Computer Engineering and Artificial Intelligence (ICCEAI) · doi:10.1109/icceai55464.2022.00115 · Cited by 3
Boosting auxiliary task guidance: a probabilistic approach
Hasib, I. M. A., Sultana, S. S., Nyeen, I. Z., & Sabur, M. A. (2022) · IAES International Journal of Artificial Intelligence · doi:10.11591/ijai.v12.i1.pp96-105 · Cited by 3
CertainOdom: Uncertainty Weighted Multi-task Learning Model for LiDAR Odometry Estimation
Sun, L., Ding, G., Yoshiyasu, Y., & Kanehiro, F. (2022) · IEEE International Conference on Robotics and Biomimetics (ROBIO) · doi:10.1109/robio55434.2022.10011808 · Cited by 3
“DeepVO [8] introduces a Recurrent Neural Network (RNN) based model for pose regression.”
BackgroundMethodResultInfluentialPixel-Wise Prediction based Visual Odometry via Uncertainty Estimation
Chen, H., Liao, T.-H., Yang, H.-K., & Lee, C. (2023) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv56688.2023.00255 · Cited by 3
SLAM Methods for Augmented Reality Systems for Flight Simulators
Nwobodo, O. J., Wereszczyński, K., & Cyran, K. A. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-35995-8_46 · Cited by 3
“DeepVo[11].”
Causal calibration: iteratively calibrating LiDAR and camera by considering causality and geometry
Liu, R., Shi, J., Zhang, H., Zhang, J., & Sun, B. (2023) · Complex & Intelligent Systems · doi:10.1007/s40747-023-01140-1 · Cited by 3
“DeepVO [36] is implemented as an end-to-end visual odometry by considering the importance of sequential dependence and complex motion dynamics of an image sequence.”
MethodXu, S., Xiong, H., Wu, Q., Yao, T., Wang, Z., & Wang, Z. (2023) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48891.2023.10161464 · Cited by 3
“Recently, deep learning based SLAM methods [4], [5], [6], [7] have demonstrated great success with superior performance in challenging situ-”
BackgroundMethodGraph attention network-optimized dynamic monocular visual odometry
Hongru, Z., & Qiao, X. (2023) · Applied Intelligence · doi:10.1007/s10489-023-04687-1 · Cited by 3
“DeepVO [7] firstly realizes an end-to-end pose estimation by formulating VO as a sequence-to-sequence learning problem, and achieves competitive performance in terms of accuracy and generalization ability.”
BackgroundMethodInfluentialFederated Meta Learning for Visual Navigation in GPS-denied Urban Airspace
Yüksek, B., Yu, Z., Suri, N., & İnalhan, G. (2023) · IEEE/AIAA Digital Avionics Systems Conference (DASC) · doi:10.1109/dasc58513.2023.10311195 · Cited by 3
“∥ is 2-norm, κ is a scale factor to provide balance between position and orientation errors, and N is number of samples [14].”
BackgroundMethodInfluentialGMIW-Pose: Camera Pose Estimation via Global Matching and Iterative Weighted Eight-Point Algorithm
Chen, F., Wu, Y., Liao, T., Zeng, H., Ouyang, S., & Guan, J. (2023) · Electronics · doi:10.3390/electronics12224689 · Cited by 3
Sudevan, V., Zayer, F., Javed, S., Karki, H., Masi, G., & Dias, J. M. M. (2023) · International Conference on Advanced Robotics (ICAR) · doi:10.1109/icar58858.2023.10436500 · Cited by 3
“Geometry-based VIO uses handcrafted features, involving complex calculations and sensitivity to illumination and image distortion [11].”
BackgroundVisual-LIDAR SLAM Based on Supervised Hierarchical Deep Neural Networks
An, Y., Sun, Z., Zhang, C., Yue, H., Yan, Z., & Xu, H. (2024) · Youth Academic Annual Conference of Chinese Association of Automation (YAC) · doi:10.1109/yac63405.2024.10598571 · Cited by 3
“Wang et al. [15] introduced DeepVO, which initially extracts appearance features from the current frame using a CNN. DeepVO integrates a recurrent neural network (RNN) with the CNN to address the visual odometry challenge.”
MethodVisual-Inertial Features Fusion for Pose Estimation
Wang, Y., & Xu, X. (2024) · Chinese Control Conference (CCC) · doi:10.23919/ccc63176.2024.10661761 · Cited by 3
“And the right one shows the agent's (vehicle's) speed at each time step, Vis-only [23] 9.3454 3.3164 0.1392 0.1789 100 121.573481472 VIO based on Flownet [24] 14.3478 where darker color signifies lower speed.”
BackgroundInfluentialHu, L., Li, X., Bonardi, F., Oufroukh, N. A., & Ali, S. A. (2024) · WRC Symposium on Advanced Robotics and Automation (WRC SARA) · doi:10.1109/wrcsara64167.2024.10685706 · Cited by 3
“Additionally, the rise of deep learning has led to extensive research in deep learning-based visual odometry (VO) [5][6][7][8].”
BackgroundMethodMOVRO2: Loosely coupled monocular visual radar odometry using factor graph optimization
Štironja, V.-J., Peršić, J., Petrović, L., Marković, I., & Petrovic, I. K. (2024) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2024.104860 · Cited by 3
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
