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 201–300 of 969
LEAP-VO: Long-term Effective Any Point Tracking for Visual Odometry
Chen, W., Chen, L., Wang, R., & Pollefeys, M. (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/cvpr52733.2024.01876 · Cited by 32
“Advancements in deep learning have spawned end-to-end monocular VO methods in supervised [3, 27, 31, 35, 37, 38] and unsupervised [25, 39, 44] settings.”
BackgroundSelf-Supervised Scale Recovery for Monocular Depth and Egomotion Estimation
Wagstaff, B., & Kelly, J. W. (2021) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros51168.2021.9635938 · Cited by 33
“The most straightforward means of enforcing metricallyscaled depth and egomotion predictions is through supervised learning (see [6], [17], [18]).”
BackgroundAn Integrated Navigation Method for Small-Sized AUV in Shallow-Sea Applications
Zhang, X., He, B., & Gao, S. (2022) · IEEE Transactions on Vehicular Technology · doi:10.1109/tvt.2022.3216003 · Cited by 29
Fu, T., Su, S., Lu, Y., & Wang, C. (2024) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2024.3382533 · Cited by 30
“Most studies on the subject employed end-to-end structures, including both supervised [27, 28, 29, 5, 30, 9] and unsupervised methods [31, 32, 33, 34, 35, 36].”
BackgroundInfluentialAccurate Visual Localization for Automotive Applications
Brosh, E., Friedmann, M., Kadar, I., Lavy, L. Y., Levi, E., Rippa, S., Lempert, Y., Fernandez-Ruiz, B., Herzig, R., & Darrell, T. (2019) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw.2019.00170 · Cited by 28
“attempts of solving this problem using deep learning techniques still involved complex additional steps such as computing dense optical flow [9] or using SfM to label the data [17] to work. Wang at al [44] were the first to suggest an end-to-end approach using a recurrent neural network and show competitive performance…”
MethodPosition Correction Model Based on Gated Hybrid RNN for AUV Navigation
Lv, P., He, B., & Guo, J. Q. (2021) · IEEE Transactions on Vehicular Technology · doi:10.1109/tvt.2021.3080134 · Cited by 28
“DeepVO [11], a typical supervised learning model of visual odometry (VO), uses the combination of convolutional neural network (ConvNet) and recurrent neural network to automatically learn to pose transformation from the original image to realize the end-to-end learning of VO.”
BackgroundVisual-LiDAR SLAM Based on Unsupervised Multi-channel Deep Neural Networks
An, Y., Shi, J.-Z., Gu, D.-B., & Liu, Q. (2022) · Cognitive Computation · doi:10.1007/s12559-022-10010-w · Cited by 29
“Wang et al. [21] extended the CNN by adding a recurrent neural network (RNN) structure.”
BackgroundLoosely-Coupled Ultra-wideband-Aided Scale Correction for Monocular Visual Odometry
Nguyen, T. H., Nguyen, T., Cao, M., & Xie, L. (2020) · Unmanned Systems · doi:10.1142/s2301385020500119 · Cited by 27
DF-VO: What Should Be Learnt for Visual Odometry?
Zhan, H., Weerasekera, C. S., Bian, J.-W., Garg, R., & Reid, I. (2021) · arXiv · doi:10.48550/arxiv.2103.00933 · Cited by 31
“(Wang et al. 2017) propose a recurrent network for learning VO from videos.”
A multi-sensor fusion framework with tight coupling for precise positioning and optimization
Xia, Y.-J., Wu, H., Zhu, L.-C., Qi, W.-W., Zhang, S., & Zhu, J. (2023) · Signal Processing · doi:10.1016/j.sigpro.2023.109343 · Cited by 27
RobustStateNet: Robust ego vehicle state estimation for Autonomous Driving
Dahal, P., Mentasti, S., Paparusso, L., Arrigoni, S., & Braghin, F. (2023) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2023.104585 · Cited by 27
RGBD Objects in the Wild: Scaling Real-World 3D Object Learning from RGB-D Videos
Xia, H., Fu, Y., Liu, S., & Wang, X. (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/cvpr52733.2024.02112 · Cited by 48
“Additionally, center object segmentation masks can be attained by bounding-box detection using text prompt of object category in Grounding-DINO [41], segmentation using Segment-Anything [33] and mask tracking using XMem [11], which are largely integrated into [12, 78].”
MethodLiu, J., Zhuo, D., Feng, Z., Zhu, S., Peng, C., Liu, Z., & Wang, H. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-72684-2_27 · Cited by 27
TransFusionOdom: Transformer-Based LiDAR-Inertial Fusion Odometry Estimation
Sun, L., Ding, G., Qiu, Y., Yoshiyasu, Y., & Kanehiro, F. (2023) · IEEE Sensors Journal · doi:10.1109/jsen.2023.3302401 · Cited by 27
“The baseline method proposed in PoseNet [55] and employed in DeepVO [21], ATVIO [6], VINet [7] and SelectFu-sion [12] is as follows, where L2 loss is used as the distance function to calculate the error between prediction and ground truth [ˆ x, ˆ q ] , the manually tuned hyperparameter β is used to weigh the error bet…”
Learning Kalman Network: A deep monocular visual odometry for on-road driving
Zhao, C., Sun, L., Yan, Z., Neumann, G. M., Duckett, T., & Stolkin, R. (2019) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2019.07.004 · Cited by 26
“In f contrast to conventional model-based VO, transforming optical flow directly to poses is a popular approach for deep-learning-based VO approaches such as DeepVO [30][10] based on learned optical flow features and LS-VO [18] based directly on learned optical flow. oo In addition, some researches try to combine global…”
MethodInfluentialSelf-Supervised Deep Pose Corrections for Robust Visual Odometry
Wagstaff, B., Peretroukhin, V., & Kelly, J. W. (2020) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra40945.2020.9197562 · Cited by 26
Xiong, M., Zhang, Z., Zhong, W., Ji, J., Liu, J., & Xiong, H. (2020) · International Joint Conference on Artificial Intelligence · doi:10.24963/ijcai.2020/134 · Cited by 26
“[Wang et al. , 2017] formulates VO as a sequence learning problem.”
BackgroundGolroudbari, A. A., & Sabour, M. H. (2023) · Measurement · doi:10.1016/j.measurement.2023.113105 · Cited by 34
Li, Z., & Zhou, A. (2023) · Applied Intelligence · doi:10.1007/s10489-023-04754-7 · Cited by 27
“The selected comparison method includes the handcrafted engineering method (VINS [45]) and the deep learning method (PoseNet [46] and DeepVO [47]).”
MethodMonocular Vision based Autonomous Landing of Quadrotor through Deep Reinforcement Learning
Xu, Y., Liu, Z., & Wang, X. (2018) · Chinese Control Conference (CCC) · doi:10.23919/chicc.2018.8482830 · Cited by 26
Adversarial Learning for Joint Optimization of Depth and Ego-Motion
Wang, A., Fang, Z., Gao, Y., Tan, S., Wang, S., Ma, S., & Hwang, J. (2020) · IEEE Transactions on Image Processing · doi:10.1109/tip.2020.2968751 · Cited by 25
Self-citationDeepLO: Geometry-Aware Deep LiDAR OdometryPreprint
Cho, Y., Kim, G., & Kim, A. (2019) · arXiv.org · doi:10.48550/arxiv.1902.10562 · Cited by 40
“Wang et al. [10] proposed a supervised visual odometry network using a recurrent network structure, while Zhou et al. [11] introduced a method of configuring a tracking and mapping process in a network structure.”
BackgroundMethodImproving autonomous detection in dynamic environments with robust monocular thermal SLAM system
Wu, Y., Wang, L., Zhang, L., Bai, Y., Cai, Y., Wang, S., & Li, Y. (2023) · ISPRS Journal of Photogrammetry and Remote Sensing · doi:10.1016/j.isprsjprs.2023.08.002 · Cited by 25
Self-citationDeep Visual Odometry With Adaptive Memory
Xue, F., Wang, X., Wang, J., & Zha, H. (2020) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2020.3014100 · Cited by 27
“The learning-based baselines include supervised approaches such as DeepVO [24], ESP-VO [25], GFS-VO [26], CL-VO [56] and unsupervised approaches such as SfmLearner [15], NeuralBundler [49], Vid2Depth [19], GeoNet [17], Depth-VO-Feat [16], CC [21], UnOS [22], and UndeepVO [18].”
BackgroundMethodInfluentialSuperPointVO: A Lightweight Visual Odometry based on CNN Feature Extraction
Xiao, H., Tao, Y., Li, Z., Cen, R., & Xue, F. (2020) · International Conference on Automation, Control and Robotics Engineering (CACRE) · doi:10.1109/cacre50138.2020.9230348 · Cited by 25
“[5–7] define the state-of-the-art technology in this category.”
BackgroundMethodOvercoming Obstructions via Bandwidth-Limited Multi-Agent Spatial Handshaking
Glaser, N., Liu, Y., Tian, J., & Kira, Z. (2021) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros51168.2021.9636761 · Cited by 25
“This procedure resembles that of Deep Homography and Visual Odometry methods [4, 19].”
MethodZhao, Y., Chen, L., Zhang, X., Xu, S., Bu, S., Jiang, H., Han, P., Li, K., & Wan, G. (2021) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2021.3090203 · Cited by 26
“Recently, supervised [33]–[35] and unsupervised [36], [37] deep learning-based visual odometer (VO) present novel results compared with traditional geometry-based methods.”
MethodResult3D Structure from 2D Dimensional Images Using Structure from Motion Algorithms
Elkhrachy, I. (2022) · Sustainability · doi:10.3390/su14095399 · Cited by 25
Survey on 3D Reconstruction Techniques: Large-Scale Urban City Reconstruction and Requirements
Christodoulides, A., Tam, G. K. L., Clarke, J., Smith, R., Horgan, J., Micallef, N., Morley, J. G., Villamizar, N., & Walton, S. P. (2025) · IEEE Transactions on Visualization and Computer Graphics · doi:10.1109/tvcg.2025.3540669 · Cited by 24
“DeepVO 7 (cid:2) [157], focusing on SLAM’s localization component, shows how feature extraction, matching, and motion estimation can be done end-to-end with Recurrent Convolutional Neural Networks (RCNNs).”
Deep Unsupervised Learning for Simultaneous Visual Odometry and Depth Estimation
Lu, Y., & Lu, G. (2019) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2019.8803247 · Cited by 24
“[12] proposed a AlexNet-based recurrent convolutional neural network architecture called DeepVO to learn monocular odometry from video sequences.”
Xu, Y., Lin, J., Shi, J., Zhang, G., Wang, X., & Li, H. (2022) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2022.3140794 · Cited by 24
“In recent years, following the learning-based visual odometry [12], [13], [14], [15], [16], learning-based LiDAR odometry methods [17], [6], [18], [19] have shown their competitiveness in both accuracy and real-time efficiency.”
Bellocchio, E., Crocetti, F., Costante, G., Fravolini, M. L., & Valigi, P. (2021) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2021.104615 · Cited by 24
Zhang, Y., Gao, J., Chen, Y., Bian, C., Zhang, F., & Liang, Q. (2022) · International Journal of Robust and Nonlinear Control · doi:10.1002/rnc.6051 · Cited by 24
“With the equations of a underactuated underwater vehicle (1) and (11), and visual measurements (23) and (24), the objective of the PBVS-based docking control can be formulated as two parts:”
BackgroundUsing Unsupervised Deep Learning Technique for Monocular Visual Odometry
Liu, Q., Li, R., Hu, H., & Gu, D. (2019) · IEEE Access · doi:10.1109/access.2019.2896988 · Cited by 23
Integrating Egocentric Localization for More Realistic Point-Goal Navigation AgentsPreprint
Datta, S., Maksymets, O., Hoffman, J., Lee, S., Batra, D., & Parikh, D. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2009.03231 · Cited by 24
“Most similar to our approach for training our agent’s odometer is [18] that estimates relative camera pose from image sequences.”
BackgroundChiNet: Deep recurrent convolutional learning for multimodal spacecraft pose estimation
Rondao, D., Aouf, N., & Richardson, M. A. (2022) · IEEE Transactions on Aerospace and Electronic Systems · doi:10.1109/taes.2022.3193085 · Cited by 24
“The CNN features are fed to two stacked LSTM layers with 1000 hidden states each; stacked LSTM layers have been previously adopted for architectures such as DeepVO [41] and DeepLO [43] and shown empirically to help in modelling complex motion dynamics.”
MethodInfluential3D plant root system reconstruction based on fusion of deep structure-from-motion and IMU
Lu, Y., Wang, Y., Chen, Z., Khan, A., Salvaggio, C., & Lu, G. (2021) · Multimedia Tools and Applications · doi:10.1007/s11042-020-10069-3 · Cited by 24
“Unsupervised learning networks are also proposed to estimate depth map and/or camera odometry based on intensity consistency between input image pairs [3, 16, 19, 21, 41, 43, 59, 67, 70].”
BackgroundDeep Online Correction for Monocular Visual Odometry
Zhang, J., Sui, W., Wang, X., Meng, W., Zhu, H., & Zhang, Q. (2021) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48506.2021.9561642 · Cited by 23
“One of the first DL-based VO estimation methods is proposed by Konda et al. [8] and Wang et al. [10].”
BASL-AD SLAM: A Robust Deep-Learning Feature-Based Visual SLAM System With Adaptive Motion Model
Han, J., Dong, R., & Kan, J. (2024) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2024.3367906 · Cited by 24
“Some researchers propose training end-to-end deep neural networks to replace the entire VSLAM system, like PoseNet [21], Deep VO [22], and Undeep VO [23].”
BackgroundSemantic Scene Mapping with Spatio-temporal Deep Neural Network for Robotic Applications
Li, R., Gu, D., Liu, Q., Long, Z., & Hu, H. (2017) · Cognitive Computation · doi:10.1007/s12559-017-9526-9 · Cited by 23
“[25] proposed DeepVO which used Recurrent Convolutional Neural Network (RCNN) to perform visual odometry (VO).”
MethodViPR: Visual-Odometry-aided Pose Regression for 6DoF Camera Localization
Ott, F., Feigl, T., Löffler, C., & Mutschler, C. (2020) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw50498.2020.00029 · Cited by 23
“In addition to PoseNet+LSTM [57], there are also similar approaches that exploit time-context that is inherently given by consecutive images (i.e., DeepVO [58], ContextualNet [45], and VidLoc [10]).”
BackgroundMethodInfluentialEsfahani, M. A., Wu, K., Yuan, S., & Wang, H. (2019) · Image and Vision Computing · doi:10.1016/j.imavis.2019.06.014 · Cited by 21
Fast Semantic-Aware Motion State Detection for Visual SLAM in Dynamic Environment
Singh, G., Wu, M.-Q., Đỗ, M. V., & Lam, S.-K. (2022) · IEEE transactions on intelligent transportation systems (Print) · doi:10.1109/tits.2022.3213694 · Cited by 21
“Modern vSLAMs estimate the pose either by exploiting the geometric relationship among the static feature correspondences observed in the images [5], [21], [22] or using end-to-end deep learning methods [23], [24].”
BackgroundTerrainFusion: Real-time Digital Surface Model Reconstruction based on Monocular SLAM
Wang, W., Zhao, Y., Han, P., Zhao, P., & Bu, S. (2019) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros40897.2019.8967663 · Cited by 23
Golroudbari, A. A., & Sabour, M. H. (2023) · doi:10.22541/au.168664884.43899660/v1 · Cited by 22
Tong, W., Cai, Y., Jie, Y.-W., Duan, Y., Hou, Y., & Wu, E. Q. (2025) · IEEE Transactions on Automation Science and Engineering · doi:10.1109/tase.2025.3546713 · Cited by 22
“Wang et al. [33]. proposed a learning-based monocular visual odometry for image sequences, which directly extracts instance representations from RGB images to estimate pose.”
Scale Correct Monocular Visual Odometry Using a LiDAR Altimeter
Giubilato, R., Chiodini, S., Pertile, M., & Debei, S. (2018) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2018.8594096 · Cited by 21
“Stereo Visual Odometry has been implemented since the first Mars Exploration Rovers [1] to provide accurate localization and up to now is one of the most favorable choices for autonomous driving applications [2].”
MethodDeep-learning-based vision for earth-moving automation
Borngrund, C., Sandin, F., & Bodin, U. (2021) · Automation in Construction · doi:10.1016/j.autcon.2021.104013 · Cited by 24
XVO: Generalized Visual Odometry via Cross-Modal Self-Training
Lai, L., Shangguan, Z., Zhang, J., & Ohn-Bar, E. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.00926 · Cited by 21
“Monocular Visual Odometry: Despite recent advances, both geometrical and learning-based VO approaches are still mostly evaluated over limited datasets under similar training and testing conditions [9, 25, 40, 64, 72, 90, 92, 94].”
BackgroundMethodDigital Enhancement of Cultural Experience and Accessibility for the Visually Impaired
Iakovidis, D. K., Diamantis, D. E., Dimas, G., Ntakolia, C., & Evaggelos Spyrou, A. (2019) · EAI/Springer Innovations in Communication and Computing · doi:10.1007/978-3-030-16450-8_10 · Cited by 21
Towards High-Speed Localisation for Autonomous Drone Racing
Cocoma-Ortega, J. A., & Martínez-Carranza, J. (2019) · Lecture notes in computer science · doi:10.1007/978-3-030-33749-0_59 · Cited by 20
Lim, H., Park, C., & Myung, H. (2019) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros40897.2019.8968551 · Cited by 20
“Therefore, many researchers exploited LSTM for sequential modeling after feature extraction by Convolutional Neural Networks (CNN) [23]–[25].”
BackgroundLiang, Z., Wang, Q., & Yu, Y. (2021) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros51168.2021.9636131 · Cited by 20
“Earlier studies [4] and [5] design a VO system by using CNN and RNN to directly (cid:2)t the pose.”
MethodEgo-Motion Estimation Using Recurrent Convolutional Neural Networks through Optical Flow Learning
Zhao, B., Huang, Y., Wei, H., & Hu, X. (2021) · Electronics · doi:10.3390/electronics10030222 · Cited by 20
“trel rrel trel rrel trel rrel trel rrel trel rrel trel rrel DeepVO [12] 8.”
BackgroundMethodInfluentialQiu, J., Wang, X., Maybank, S. J., & Tao, D. (2019) · 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.2019.00869 · Cited by 20
“[80, 81] implement a recurrent ConvNet architecture for visual odometry estimation.”
LS-Net: Learning to Solve Nonlinear Least Squares for Monocular StereoPreprint
Clark, R., Bloesch, M., Czarnowski, J., Leutenegger, S., & Davison, A. J. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1809.02966 · Cited by 20
“Learning for Monocular Stereo There has been much interest recently in using end-to-end learning to estimate the motion of a camera [6, 5, 22] and reconstruct scenes from monocular images [8].”
BackgroundSelf-citationmilliEgo: Single-chip mmWave Radar Aided Egomotion Estimation via Deep Sensor FusionPreprint
Lu, C. X., Saputra, M. R. U., Zhao, P., Almalıoğlu, Y., de Gusmão, P. P. B., Chen, C., Sun, K., Trigoni, N., & Markham, A. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2006.02266 · Cited by 20
“To evaluate the proposed model, we adopt the mean square of Relative Pose Error (RPE) and Absolute Trajectory Error (ATE), as they are widely used for measuring the accuracy of odometry trajectories [10, 57, 60].”
BackgroundMethodSelf-citationLearning to Calibrate - Estimating the Hand-eye Transformation Without Calibration Objects
Pachtrachai, K., Vasconcelos, F., Edwards, P., & Stoyanov, D. V. (2021) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2021.3098942 · Cited by 19
“…in solving computer vision problems [20] using deep neural networks such as classification [21], [22], localisation [23]–[25], sensors calibration [26], [27], and segmentation [28], [29], there are only a few papers proposing automatic hand-eye calibration method utilising deep neural networks [30].”
MethodThe YTU dataset and recurrent neural network based visual-inertial odometry
Gürtürk, M., Yusefi, A., Aslan, M. F., Soycan, M., Durdu, A., & Masiero, A. (2021) · Measurement · doi:10.1016/j.measurement.2021.109878 · Cited by 19
MotionHint: Self-Supervised Monocular Visual Odometry with Motion Constraints
Wang, C., Wang, Y.-P., & Manocha, D. (2022) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra46639.2022.9812288 · Cited by 21
HFNet-SLAM: An Accurate and Real-Time Monocular SLAM System with Deep Features
Liu, L., & Aitken, J. M. (2023) · Sensors · doi:10.3390/s23042113 · Cited by 21
“DeepVO [21] introduced an end-to-end recurrent CNN (RCNN) for pose estimation.”
BackgroundMonocular depth estimation using self-supervised learning with more effective geometric constraints
Xiong, M., Zhang, Z., Liu, J., Zhang, T., & Xiong, H. (2023) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2023.107489 · Cited by 19
Etxeberria-Garcia, M., Labayen, M., Zamalloa, M., & Arana-Arexolaleiba, N. (2020) · IEEE/SICE International Symposium on System Integration (SII) · doi:10.1109/sii46433.2020.9026246 · Cited by 20
“Wang et al. [27] presented DeepVO, an approach that mixes CNN and RNN called Recurrent Convolutional Neural Network (RCNN).”
MethodA Multi-spectral Dataset for Evaluating Motion Estimation Systems
Dai, W.-C., Zhang, Y., Chen, S., Sun, D., & Kong, D. (2020) · IEEE International Conference on Robotics and Automation · doi:10.1109/icra48506.2021.9561906 · Cited by 23
“Besides, methods [13], [14] with deep learning methods have attracted the interest”
MethodJPerceiver: Joint Perception Network for Depth, Pose and Layout Estimation in Driving Scenes
Zhao, H., Zhang, J., Zhang, S., & Tao, D. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-19839-7_41 · Cited by 19
Application of Event Cameras and Neuromorphic Computing to VSLAM: A Survey
Tenzin, S., Rassau, A., & Chai, D. (2024) · Biomimetics · doi:10.3390/biomimetics9070444 · Cited by 20
DenseFusion: Large-Scale Online Dense Pointcloud and DSM Mapping for UAVs
Chen, L., Zhao, Y., Xu, S., Bu, S., Han, P., & Wan, G. (2020) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros45743.2020.9341413 · Cited by 18
“In recent years, with the rising of deep learning, there are some SLAM frameworks based on it: supervised [9]–[11], and unsupervised [12], [13] methods.”
BackgroundRobust self-supervised monocular visual odometry based on prediction-update pose estimation network
Xiu, H., Liang, Y., Zeng, H., Li, Q., Liu, H., Fan, B., & Li, C. (2022) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2022.105481 · Cited by 18
Sparse2Dense: From Direct Sparse Odometry to Dense 3-D Reconstruction
Tang, J., Folkesson, J. B., & Jensfelt, P. (2019) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2019.2891433 · Cited by 18
Self-Supervised Ego-Motion Estimation Based on Multi-Layer Fusion of RGB and Inferred Depth
Jiang, Z., Taira, H., Miyashita, N., & Okutomi, M. (2022) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra46639.2022.9811842 · Cited by 19
Deep Unsupervised Visual Odometry Via Bundle Adjusted Pose Graph Optimization
Lu, G. (2023) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48891.2023.10160703 · Cited by 19
“Deep neural networks have been applied in visual odometry to recover both trajectory and scene depth simultaneously [38] [34] [31] [24] [40] [37] [16] [29] [2] [33].”
BackgroundMethodInfluentialAutomated Process for Incorporating Drivable Path into Real-Time Semantic Segmentation
Zhou, W., Worrall, S., Zyner, A., & Nebot, E. (2018) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2018.8460486 · Cited by 16
“Stereo visual odometry requires precise sensor calibration and high-quality feature detection and matching [10].”
BackgroundEMPNet: Neural Localisation and Mapping Using Embedded Memory Points
Avraham, G., Zuo, Y., Dharmasiri, T., & Drummond, T. W. (2019) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2019.00821 · Cited by 16
A Hybrid Sparse-Dense Monocular SLAM System for Autonomous Driving
Gallagher, L., Kumar, V. R., Yogamani, S. K., & McDonald, J. (2021) · European Conference on Mobile Robots · doi:10.1109/ecmr50962.2021.9568797 · Cited by 17
Cao, Y., Zhang, X.-S., Luo, F., Peng, P., Lin, C., Yang, K.-F., & Li, Y. (2022) · Pattern Recognition · doi:10.1016/j.patcog.2022.109262 · Cited by 17
“The supervised methods minimize the distance between predicted values (depth and ego-motion) and corresponding ground truth by using such strategies as a recurrent neural network (RNN) [27], or feature-metric BA [15].”
MethodTwo Stream Networks for Self-Supervised Ego-Motion EstimationPreprint
Ambruş, R., Guizilini, V., Li, J., Pillai, S., & Gaidon, A. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1910.01764 · Cited by 16
“DeepVO [50] is trained on Seq.”
MethodAn Accurate, Robust Visual Odometry and Detail-Preserving Reconstruction System
Gong, X., Liu, Y., Wu, Q., Huang, J., Zong, H., & Wang, J. (2020) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2020.3017886 · Cited by 16
“Though our method has relatively weak performance on sequences TUM /fr1/xyz , TUM /fr1/rpy and TUM /fr3/structure−texture−far , when compared to ORB-SLAM and DeepVO, its performance is very close to 2https://github.com/tinghuiz/SfMLearner 3github.com/raulmur/ORB SLAM2 4https://github.com/magican/OpenDTAM.git 5https://…”
InfluentialYang, B., Xu, X., Ren, J., Cheng, L., Guo, L., & Zhang, Z. (2021) · Pattern Recognition Letters · doi:10.1016/j.patrec.2021.11.028 · Cited by 16
Real-Time 3D Mapping in Isolated Industrial Terrain with Use of Mobile Robotic Vehicle
Buratowski, T., Garus, J., Giergiel, M., & Kudriashov, A. (2022) · Electronics · doi:10.3390/electronics11132086 · Cited by 16
Unsupervised Learning of Depth Estimation and Camera Pose With Multi-Scale GANs
Xu, Y., Wang, Y., Huang, R., Lei, Z., Yang, J., & Li, Z. (2022) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2021.3093592 · Cited by 16
“Early learning based methods for predicting depth maps [1]–[4] and camera pose [5], [6] operate in a supervised learning strategy.”
BackgroundSelf-Supervised Deep Visual Odometry Based on Geometric Attention Model
Dai, J., Gong, X., Li, Y., Wang, J., & Wei, M. (2022) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2022.3227917 · Cited by 16
“DL methods can learn the most distinct motion information from pure frames or depth images, thereby enhancing the performance of VO in complex environments; for instance, supervised VO methods [10], [11], [12] have been designed to solve the problem of camera ego-motion estimation.”
MethodSelfD: Self-Learning Large-Scale Driving Policies From the Web
Zhang, J., Zhu, R., & Ohn-Bar, E. (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/cvpr52688.2022.01680 · Cited by 18
“While the speed and command inputs to fθ can be recovered through visual odometry techniques [73], these result in highly noisy trajectories in our online video settings (discussed in Sec.”
BackgroundMethodInfluentialDynamic Object-Aware Visual Odometry (VO) Estimation Based on Optical Flow Matching
Cho, H. M., & Kim, E. (2023) · IEEE Access · doi:10.1109/access.2023.3241961 · Cited by 17
“Withthedevelopment of deep learning, studies have been conducted to predict the pose between two images by training deep neural networks [14], [15], [16].”
BackgroundMobile robot monocular vision-based obstacle avoidance algorithm using a deep neural network
Rezaei, N., & Darabi, S. (2023) · Evolutionary Intelligence · doi:10.1007/s12065-023-00829-z · Cited by 16
“Nowadays, deep learning has demonstrated its excellent performance in robotics and computer vision [18–20].”
BackgroundWang, J., Yu, H., Lin, X., Li, Z., Sun, W. J., & Akhtar, N. (2023) · Expert Systems with Applications · doi:10.1016/j.eswa.2023.122755 · Cited by 16
Salient Sparse Visual Odometry With Pose-Only Supervision
Chen, S., Liu, K., Wang, C., Yuan, S., Yang, J., & Xie, L. (2024) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2024.3384757 · Cited by 16
“In response, end-to-end deep learning-based techniques [8]–[13] are proposed to address these challenges but suffer from reduced performance when tested under conditions different from training data [14].”
BackgroundMethodInfluentialDeep Visual Odometry with Events and Frames
Pellerito, R., Cannici, M., Gehrig, D., Belhadj, J., Dubois-Matra, O., Casasco, M., & Scaramuzza, D. (2024) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros58592.2024.10801307 · Cited by 17
“Su-pervised methods, on the other hand, either rely on end-to-end camera-motion regressors [15], [16], [17], [18] or exploit hybrid solutions that combine geometric models with deep neural networks [19], [20], [21].”
Convolutional Neural Network Based Sensors for Mobile Robot Relocalization
Sinha, H., Patrikar, J., Dhekane, E. G., Pandey, G., & Kothari, M. (2018) · International Conference on Methods and Models in Automation and Robotics · doi:10.1109/mmar.2018.8485921 · Cited by 16
“Recent work by [8], named DeepVO, performs end-to-end visual odometry on monocular RGB images.”
BackgroundShih, M.-L., Chen, Y., Tung, C.-Y., Sun, C. J., Cheng, C.-J., Chan, L., Varadarajan, S., & Sun, M. (2018) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2018.8593711 · Cited by 16
“D object location P t in the current camera coordinate at time t as follows, Implementation Similar to DeepVO [12], we fine tune the pre-trained FlowNet [13] on our dataset to predict the camera motion.”
BackgroundMethodCustomizing Object Detectors for Indoor Robots
Alabachi, S., Sukthankar, G., & Sukthankar, R. (2019) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra.2019.8793551 · Cited by 16
“This is related to the challenge of using deep learning to perform visual SLAM [2] but with the objective of tasking the robot to use the objects rather than learning landmarks for visual navigation.”
BackgroundDeep Sensor Fusion for Real-Time Odometry Estimation
Valente, M., Joly, C., & Fortelle, A. D. L. (2019) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros40897.2019.8967803 · Cited by 15
“On the other hand, for the angles comparison, since most of the time the 3 angles are around 0 degree (the only high rotation rates are on z axis during turns, which represent a very small part of a complete trajectory), all their drifts are in the same order of magnitude and should be relevant for the comparison; For…”
MethodInfluentialScale-Adaptive Neural Dense Features: Learning via Hierarchical Context Aggregation
Spencer, J., Bowden, R., & Hadfield, S. (2019) · 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.2019.00636 · Cited by 16
“Such is the case with end-to-end VO methods [42, 22], camera pose regression [19, 4] or stereo disparity estimation [47].”
A synthetic dataset for Visual SLAM evaluation
Wang, S., Yue, J., Dong, Y., He, S., Wang, H., & Ning, S. (2019) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2019.103336 · Cited by 16
Self-citationA Sparse Visual Odometry Technique Based on Pose Adjustment With Keyframe Matching
Lin, H., & Hsu, J.-L. (2020) · IEEE Sensors Journal · doi:10.1109/jsen.2020.3015922 · Cited by 16
“The other approach is based on the optical flow [34], [35].”
BackgroundSu, Y., & Yu, L. (2021) · Measurement Science and Technology · doi:10.1088/1361-6501/ac38f1 · Cited by 16
“DeepVO [15] is based on a deep recursive convolutional neural network.”
MethodTowards Scale Consistent Monocular Visual Odometry by Learning from the Virtual World
Zhang, S., Zhang, J., & Tao, D. (2022) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra46639.2022.9812347 · Cited by 17
Optical flow for autonomous driving: Applications, challenges and improvements
Shen, S., Kerofsky, L. J., & Yogamani, S. K. (2023) · Electronic Imaging · doi:10.2352/ei.2023.35.16.avm-128 · Cited by 16
PoseConvGRU: A Monocular Approach for Visual Ego-motion Estimation by Learning
Zhai, G., Liu, L., Zhang, L., Liu, Y., & Jiang, Y. (2020) · Pattern Recognition · doi:10.1016/j.patcog.2019.107187 · Cited by 15
SelfVoxeLO: Self-supervised LiDAR Odometry with Voxel-based Deep Neural NetworksPreprint
Xu, Y., Huang, Z., Lin, K.-Y., Zhu, X., Shi, J.-P., Bao, H., Zhang, G., & Li, H. (2020) · Conference on Robot Learning · doi:10.48550/arxiv.2010.09343 · Cited by 26
“Initially, several seminal methods were proposed for visual odometry [6, 7, 8].”
DEEPLIO: DEEP LIDAR INERTIAL SENSOR FUSION FOR ODOMETRY ESTIMATION
Javanmard-Ghareshiran, A., Iwaszczuk, D., & Roth, S. (2021) · ISPRS annals of the photogrammetry, remote sensing and spatial information sciences · doi:10.5194/isprs-annals-v-1-2021-47-2021 · Cited by 16
“In (Kend-all et al., 2015, Wang et al., 2019, Wang et al., 2017) the authors suggest a linearly weighted sum loss, with the weighting as a hyperparameter.”
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
