Chinaunix首页 | 论坛 | 博客

牛蹄印章_VX_iibull

首页　| 　博文目录　| 　关于我

iibull

  • 博客访问： 9417441
  • 博文数量： 1748
  • 博客积分： 12961
  • 博客等级： 上将
  • 技术积分： 20070
  • 用 户 组： 普通用户
  • 注册时间： 2009-01-09 11:25
个人简介

偷得浮生半桶水(半日闲), 好记性不如抄下来(烂笔头). 信息爆炸的时代, 学习是一项持续的工作.

文章分类

全部博文（1748）

文章存档

2024年（24）

2023年（26）

2022年（112）

2021年（217）

2020年（157）

2019年（192）

2018年（81）

2017年（78）

2016年（70）

2015年（52）

2014年（40）

2013年（51）

2012年（85）

2011年（45）

2010年（231）

2009年（287）

我的朋友
最近访客
推荐博文
相关博文
视觉SLAM 的一些算法.

分类： 其他平台

2020-04-30 11:07:13



Visual Odometry / SLAM Evaluation 2012


The odometry benchmark consists of 22 stereo sequences, saved in loss less png format: We provide 11 sequences (00-10) with ground truth trajectories for training and 11 sequences (11-21) without ground truth for evaluation. For this benchmark you may provide results using monocular or stereo visual odometry, laser-based SLAM or algorithms that combine visual and LIDAR information. The only restriction we impose is that your method is fully automatic (e.g., no manual loop-closure tagging is allowed) and that the same parameter set is used for all sequences. A development kit provides details about the data format.


From all test sequences, our evaluation computes translational and rotational errors for all possible subsequences of length (100,...,800) meters. The evaluation table below ranks methods according to the average of those values, where errors are measured in percent (for translation) and in degrees per meter (for rotation). A more detailed comparison for different trajectory lengths and driving speeds can be found in the plots underneath. Note: On 03.10.2013 we have changed the evaluated sequence lengths from (5,10,50,100,...,400) to (100,200,...,800) due to the fact that the GPS/OXTS ground truth error for very small sub-sequences was large and hence biased the evaluation results. Now the averages below take into account longer sequences and provide a better indication of the true performance. Please consider reporting these number for all future submissions. The last leaderboard right before the changes can be found !

Important Policy Update: As more and more non-published work and re-implementations of existing work is submitted to KITTI, we have established a new policy: from now on, only submissions with significant novelty that are leading to a peer-reviewed paper in a conference or journal are allowed. Minor modifications of existing algorithms or student research projects are not allowed. Such work must be evaluated on a split of the training set. To ensure that our policy is adopted, new users must detail their status, describe their work and specify the targeted venue during registration. Furthermore, we will regularly delete all entries that are 6 months old but are still anonymous or do not have a paper associated with them. For conferences, 6 month is enough to determine if a paper has been accepted and to add the bibliography information. For longer review cycles, you need to resubmit your results.
Additional information used by the methods
  •  Stereo: Method uses left and right (stereo) images
  •  Laser Points: Method uses point clouds from Velodyne laser scanner
  •  Loop Closure Detection: This method is a SLAM method that detects loop closures
  •  Additional training data: Use of additional data sources for training (see details)
Method Setting Code Translation Rotation Runtime Environment
1
This method makes use of Velodyne laser scans.
0.54 % 0.0013 [deg/m] 0.1 s 2 cores @ 2.5 Ghz (C/C++)
J. Zhang and S. Singh: . IEEE International Conference on Robotics and Automation(ICRA) 2015.
2
This method makes use of Velodyne laser scans.
0.55 % 0.0013 [deg/m] 0.1 s 2 cores @ 2.5 Ghz (C/C++)
J. Zhang and S. Singh: . Robotics: Science and Systems Conference (RSS) 2014.
3
This method uses stereo information.
0.65 % 0.0014 [deg/m] 0.1 s 2 cores @ 2.5 Ghz (C/C++)
I. Cvi?i?, J. ?esi?, I. Markovi? and I. Petrovi?: . Journal of Field Robotics 2017.
4
This method makes use of Velodyne laser scans.
0.69 % 0.0018 [deg/m] 1.25 s 1 core @ >3.5 Ghz (C/C++)
J. Deschaud: . 2018 IEEE International Conference on Robotics and Automation (ICRA) 2018.
5
This method makes use of Velodyne laser scans.
0.69 % 0.0016 [deg/m] 0.1 s 4 cores @ 2.5 Ghz (C/C++)
F. Neuhaus, T. Koss, R. Kohnen and D. Paulus: . German Conference on Pattern Recognition 2018.
6
This method uses stereo information.
This method makes use of Velodyne laser scans.
0.70 % 0.0016 [deg/m] 0.1 s 2 cores @ 2.5 Ghz (C/C++)
7
This method uses stereo information.
0.80 % 0.0026 [deg/m] 0.08 s 4 cores @ 3.0 Ghz (C/C++)
8 0.81 % 0.0025 [deg/m] 0.1 s 1 core @ 2.5 Ghz (C/C++)
9
This method uses stereo information.
0.82 % 0.0020 [deg/m] 0.2 s 4 cores @ 2.5 Ghz (C/C++)
K. Lenac, J. ?esi?, I. Markovi? and I. Petrovi?: . The International Journal of Robotics Research 2018.
10
This method uses stereo information.
0.83 % 0.0026 [deg/m] 0.25 s 2 cores @ 2.0 Ghz (C/C++)
M. Buczko and V. Willert: . 19th IEEE Intelligent Transportation Systems Conference (ITSC) 2016.
M. Buczko, V. Willert, J. Schwehr and J. Adamy: . IEEE Intelligent Vehicles Symposium (IV) 2018.
M. Buczko: . 2018.
11
This method makes use of Velodyne laser scans.
0.84 % 0.0022 [deg/m] 0.2 s 2 cores @ 2.5 Ghz (C/C++)
J. Graeter, A. Wilczynski and M. Lauer: LIMO: Lidar-Monocular Visual Odometry. arXiv preprint arXiv:1807.07524 2018.
12
This method uses stereo information.
0.84 % 0.0023 [deg/m] 0.05 s 4 cores @ 3.0 Ghz (C/C++)
13
This method makes use of Velodyne laser scans.
0.86 % 0.0025 [deg/m] 2 s 8 cores @ 3.5 Ghz (Python)
D. Yin, Q. Zhang, J. Liu, X. Liang, Y. Wang, J. Maanp??, H. Ma, J. Hyypp? and R. Chen: . 2020.
14
This method uses stereo information.
0.86 % 0.0031 [deg/m] 0.09 s 1 core @ >3.5 Ghz (C/C++)
J. Zhu: . International Joint Conference on Artificial Intelligence, IJCAI 2017.
15
This method makes use of Velodyne laser scans.
0.86 % 0.0022 [deg/m] 0.2 s 2 cores @ 2.5 Ghz (C/C++)
J. Graeter, A. Wilczynski and M. Lauer: LIMO: Lidar-Monocular Visual Odometry. arXiv preprint arXiv:1807.07524 2018.
16
This method makes use of Velodyne laser scans.
0.87 % 0.0025 [deg/m] 0.03 s 4 cores @ 2.5 Ghz (C/C++)
K. Ji and T. Huiyan Chen: . IEEE Intelligent Vehicles Symposium (IV) 2018.
17 0.88 % 0.0029 [deg/m] 2 s >8 cores @ >3.5 Ghz (Python)
18
This method uses stereo information.
0.88 % 0.0022 [deg/m] 0.1 s 2 cores @ 2.5 Ghz (C/C++)
I. Cvi?i? and I. Petrovi?: . European Conference on Mobile Robots (ECMR) 2015.
19
This method uses stereo information.
0.88 % 0.0025 [deg/m] 0.3 s 2 cores @ 2.0 Ghz (C/C++)
M. Buczko and V. Willert: . 19th IEEE Intelligent Transportation Systems Conference (ITSC) 2016.
20 0.88 % 0.0021 [deg/m] 0.1 s 1 core @ 2.5 Ghz (C/C++)
N. Yang, L. Stumberg, R. Wang and D. Cremers: D3VO: Deep Depth, Deep Pose and Deep Uncertainty for Monocular Visual Odometry. The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2020.
21
This method makes use of Velodyne laser scans.
0.89 % 0.0030 [deg/m] 0.2 s 8 cores @ 3.5 Ghz (C/C++)
H. Hong and B. Lee: Probabilistic normal distributions transform representation for accurate 3d point cloud registration. IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2017.
22 0.90 % 0.0021 [deg/m] 0.1 s GPU @ 2.5 Ghz (C/C++)
N. Yang, R. Wang, J. Stueckler and D. Cremers: Deep Virtual Stereo Odometry: Leveraging Deep Depth Prediction for Monocular Direct Sparse Odometry. European Conference on Computer Vision (ECCV) 2018.
23
This method makes use of Velodyne laser scans.
0.93 % 0.0026 [deg/m] 0.2 s 2 cores @ 2.5 Ghz (C/C++)
J. Graeter, A. Wilczynski and M. Lauer: LIMO: Lidar-Monocular Visual Odometry. ArXiv e-prints 2018.
24
This method uses stereo information.
0.93 % 0.0020 [deg/m] 0.1 s 1 core @ 3.4 Ghz (C/C++)
R. Wang, M. Schw\"orer and D. Cremers: . International Conference on Computer Vision (ICCV), Venice, Italy 2017.
25
This method uses stereo information.
This method makes use of Velodyne laser scans.
0.96 % 0.0018 [deg/m] 0.5 s 1 core @ 3.0 Ghz (C/C++)
26
This method uses stereo information.
0.98 % 0.0028 [deg/m] 0.3 s 2 cores @ 2.0 Ghz (C/C++)
M. Buczko and V. Willert: . IEEE Intelligent Vehicles Symposium (IV) 2016.
27
This method makes use of Velodyne laser scans.
0.98 % 0.0044 [deg/m] 0.2 s 2 core @ 3.0 Ghz (C/C++)
28
This method makes use of Velodyne laser scans.
0.98 % 0.0026 [deg/m] 0.1 s 2 cores @ 2.5 Ghz (C/C++)
29
This method makes use of Velodyne laser scans.
1.00 % 0.0041 [deg/m] 0.1 s 2 cores @ >3.5 Ghz (C/C++)
30 1.01 % 0.0025 [deg/m] 1 s >8 cores @ 2.5 Ghz (C/C++)
31
This method uses stereo information.
1.02 % 0.0023 [deg/m] 0.0095 s AGX Jetson Xavier (0.03s Jetson Nano)
32 1.02 % 0.0040 [deg/m] 0.1s 1 core @ 2.8 Ghz (C/C++)
33
This method makes use of Velodyne laser scans.
1.03 % 0.0053 [deg/m] 0.15 s 1 core @ 2.5 Ghz (C/C++)
34
This method makes use of Velodyne laser scans.
1.06 % 0.0034 [deg/m] 0.1 s 1 core @ 3.5 Ghz (C/C++)
X. Chen, A. Milioto, E. Palazzolo, P. Gigu\`ere, J. Behley and C. Stachniss: . IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2019.
35
This method uses stereo information.
This method makes use of Velodyne laser scans.
1.06 % 0.0033 [deg/m] 0.15 s 2 cores @ 2.0 Ghz (C/C++)
36
This method uses stereo information.
1.09 % 0.0029 [deg/m] 0.145 s GPU @ 3.5 Ghz (C/C++)
M. Persson, T. Piccini, R. Mester and M. Felsberg: Robust Stereo Visual Odometry from Monocular Techniques. IEEE Intelligent Vehicles Symposium 2015.
37
This method uses stereo information.
1.09 % 0.0033 [deg/m] 0.1s 1 core @ 3.0 Ghz (C/C++)
T. Qin, J. Pan, S. Cao and S. Shen: . 2019.
38
This method uses stereo information.
1.10 % 0.0023 [deg/m] 0.08 s 4 cores @ 2.3 Ghz (C/C++)
39
This method uses stereo information.
1.11 % 0.0028 [deg/m] 1 s 2 cores @ 2.0 Ghz (C/C++)
M. Buczko and V. Willert: Monocular Outlier Detection for Visual Odometry. IEEE Intelligent Vehicles Symposium (IV) 2017.
40 1.12 % 0.0029 [deg/m] 0.1 s 1 core @ 2.5 Ghz (C/C++)
41
This method makes use of Velodyne laser scans.
1.14 % 0.0049 [deg/m] 0.1 s 2 cores @ 2.5 Ghz (C/C++)
J. Zhang, M. Kaess and S. Singh: Real-time Depth Enhanced Monocular Odometry. IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2014.
42
This method uses stereo information.
1.15 % 0.0027 [deg/m] 0.06 s 2 cores @ >3.5 Ghz (C/C++)
R. Mur-Artal and J. Tard\'os: ORB-SLAM2: an Open-Source SLAM System for Monocular, Stereo and RGB-D Cameras. IEEE Transactions on Robotics 2017.
43
This method uses stereo information.
1.15 % 0.0032 [deg/m] 0.018 s 1.5 cores @ 3.3 Ghz (C/C++)
D. Slepichev, M. Smirnov, E. Vendrovsky and S. Volodarskiy: . .
44
This method uses stereo information.
1.17 % 0.0035 [deg/m] 0.45 s 1 core @ 3.0 Ghz (C/C++)
J. Deigmoeller and J. Eggert: . German Conference on Pattern Recognition (GCPR) 2016.
45 1.18 % 0.0022 [deg/m] 0.1 s 1 core @ 2.5 Ghz (C/C++)
46
This method uses stereo information.
1.19 % 0.0025 [deg/m] 0.03 s 4 cores @ 3.0 Ghz (C/C++)
T. Pire, T. Fischer, G. Castro, P. De Crist\'oforis, J. Civera and J. Jacobo Berlles: . Robotics and Autonomous Systems (RAS) 2017.
T. Pire, T. Fischer, J. Civera, P. Crist\'{o}foris and J. Jacobo-Berlles: . IROS 2015.
47
This method uses stereo information.
1.20 % 0.0033 [deg/m] 0.07 s 1 core @ 3.5 Ghz (C/C++)
J. Engel, J. St\"uckler and D. Cremers: . Int.~Conf.~on Intelligent Robot Systems (IROS) 2015.
48
This method uses stereo information.
1.22 % 0.0029 [deg/m] 0.1 s 1 core @ 2.0 Ghz (C/C++)
J. Tardif, M. George, M. Laverne, A. Kelly and A. Stentz: . 2010 IEEE/RSJ International Conference on Intelligent Robots and Systems, October 18-22, 2010, Taipei, Taiwan 2010.
49
This method makes use of Velodyne laser scans.
1.22 % 0.0058 [deg/m] 0.2 s 1 core @ 2.5 Ghz (C/C++)
T. Tang, D. Yoon and T. Barfoot: . arXiv preprint arXiv:1809.06518 2018.
50
This method makes use of Velodyne laser scans.
1.22 % 0.0042 [deg/m] 0.5 s 1 core @ 2.5 Ghz (C/C++)
J. Graeter, A. Wilczynski and M. Lauer: LIMO: Lidar-Monocular Visual Odometry. ArXiv e-prints 2018.
51
This method uses stereo information.
1.25 % 0.0041 [deg/m] 0.17 s 4 cores @ 3.3 Ghz (C/C++)
X. Qu, B. Soheilian and N. Paparoditis: . ISPRS Journal of Photogrammetry and Remote Sensing 2017.
52
This method makes use of Velodyne laser scans.
1.26 % 0.0061 [deg/m] 0.2 s 1 core @ 2.5 Ghz (C/C++)
T. Tang, D. Yoon, F. Pomerleau and T. Barfoot: . 15th Conference on Computer and Robot Vision (CRV) 2018.
53
This method uses stereo information.
1.26 % 0.0038 [deg/m] 0.03 s 1 core @ 3.5 Ghz (C/C++)
W. Meiqing, L. Siew-Kei and S. Thambipillai: . IEEE Transaction on Intelligent Transportation Systems 2017.
54
This method makes use of Velodyne laser scans.
1.26 % 0.0034 [deg/m] 0.1 s 1 core @ 2.5 Ghz (C/C++)
55
This method uses stereo information.
1.28 % 0.0010 [deg/m] 0.1 s 1 core @ 3.4 Ghz (C/C++)
56 1.28 % 0.0022 [deg/m] 0.2 s 2 cores @ 2.5 Ghz (C/C++)
57 1.28 % 0.0040 [deg/m] 0.07 s 1 core @ 2.5 Ghz (C/C++)
58
This method uses stereo information.
1.30 % 0.0030 [deg/m] 0.1 s 1 core @ 2.2 Ghz (C/C++)
H. Badino, A. Yamamoto and T. Kanade: . First International Workshop on Computer Vision for Autonomous Driving at ICCV 2013.
59 1.31 % 0.0023 [deg/m] 0.03 s 4 cores @ 2.5 Ghz (C/C++)
60 1.36 % 0.0041 [deg/m] 0.1 s 1 core @ 2.5 Ghz (C/C++)
61
This method uses stereo information.
1.36 % 0.0038 [deg/m] 0.1 s 1 Core @2.8GHz (C/C++)
W. Lu, Z. Xiang and J. Liu: High-performance visual odometry with two- stage local binocular BA and GPU. Intelligent Vehicles Symposium (IV), 2013 IEEE 2013.
62
This method uses stereo information.
1.37 % 0.0035 [deg/m] 0.1 s 1 core @ 3.0 Ghz (C/C++)
I. Kre?o and S. ?egvi?: . VISAPP 2015.
63
This method makes use of Velodyne laser scans.
1.37 % 0.0051 [deg/m] 0.6 s 1 core @ 2.5 Ghz (C/C++)
D. Kovalenko, M. Korobkin and A. Minin: . 2019 European Conference on Mobile Robots (ECMR) 2019.
64
This method makes use of Velodyne laser scans.
1.39 % 0.0034 [deg/m] 0.1 s 1 core @ 3.5 Ghz (C/C++)
J. Behley and C. Stachniss: . Robotics: Science and Systems (RSS) 2018.
65
This method uses stereo information.
1.39 % 0.0035 [deg/m] 0.02 s 1 core @ 3.0 Ghz (C/C++)
D. Schlegel, M. Colosi and G. Grisetti: . ArXiv e-prints 2017.
66
This method uses stereo information.
1.43 % 0.0038 [deg/m] 0.05 s 1 core @ 3.4 Ghz (C/C++)
R. Sardana, R. Kottath, V. Karar and S. Poddar: . Proceedings of the Advances in Robotics 2019 2019.
67
This method uses stereo information.
1.51 % 0.0042 [deg/m] 0.02 s 2 cores @ 2.5 Ghz (C/C++)
H. Badino and T. Kanade: . IAPR Conference on Machine Vision Application 2011.
68 1.51 % 0.0070 [deg/m] 0.1 s 1 core @ 2.5 Ghz (C/C++)
69 1.57 % 0.0050 [deg/m] 0.05 s 1 core @ 2.5 Ghz (C/C++)
70
This method uses stereo information.
1.57 % 0.0044 [deg/m] 0.5 s 8 cores @ 3.5 Ghz (C/C++)
F. Bellavia, M. Fanfani, F. Pazzaglia and C. Colombo: . ICIAP 2013 2013.
F. Bellavia, M. Fanfani and C. Colombo: Selective visual odometry for accurate AUV localization. Autonomous Robots 2015.
M. Fanfani, F. Bellavia and C. Colombo: Accurate Keyframe Selection and Keypoint Tracking for Robust Visual Odometry. Machine Vision and Applications 2016.
71
This method makes use of Velodyne laser scans.
1.59 % 0.0053 [deg/m] 0.08 s 8 core @ 3.4 Ghz (C/C++)
72 1.65 % 0.0050 [deg/m] 0.1 s GPU
73 1.70 % 0.0028 [deg/m] 0.01 s 1 core @ 2.5 Ghz (C/C++)
74 1.72 % 0.0054 [deg/m] 0.1 s test
75 1.74 % 0.0030 [deg/m] 0.01 s 1 core @ 2.5 Ghz (C/C++)
76 1.74 % 0.0030 [deg/m] 0.01 s 1 core @ 2.5 Ghz (C/C++)
77 1.75 % 0.0056 [deg/m] 1 s 1 core @ 2.5 Ghz (C/C++)
78
This method uses stereo information.
1.76 % 0.0036 [deg/m] 0.05 s 2 cores @ 2.0 Ghz (C/C++)
M. Sanfourche, V. Vittori and G. Besnerais: . IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2013.
79
This method uses stereo information.
1.76 % 0.0026 [deg/m] 0.1 s 4 cores @ 2.5 Ghz (C/C++)
J. Huai, C. Toth and D. Grejner-Brzezinska: . Proceedings of the 27th International Technical Meeting of The Satellite Division of the Institute of Navigation (ION GNSS+ 2015) 2015.
80 1.76 % 0.0036 [deg/m] 0.1 s 1 core @ 2.5GHz (Python)
F. Pereira, J. Luft, G. Ilha, A. Sofiatti and A. Susin: . 2017 Workshop of Computer Vision (WVC) 2017.
81 1.89 % 0.0083 [deg/m] 0.02 s 1 core @ 2.5 Ghz (C/C++)
M. Dimitrievski., D. Hamme., P. Veelaert. and W. Philips.: Robust Matching of Occupancy Maps for Odometry in Autonomous Vehicles. Proceedings of the 11th Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications - Volume 3: VISAPP, (VISIGRAPP 2016) 2016.
82
This method uses stereo information.
2.04 % 0.0051 [deg/m] 0.03 s 1 core @ 2.5 Ghz (C/C++)
A. Comport, E. Malis and P. Rives: Accurate Quadrifocal Tracking for Robust 3D Visual Odometry. ICRA 2007.
M. Meilland, A. Comport and P. Rives: . ICRA 2011.
83 2.05 % 0.0051 [deg/m] 1 s 1 core @ 2.5 Ghz (Python + C/C++)
N. Fanani, A. Stuerck, M. Ochs, H. Bradler and R. Mester: Predictive monocular odometry (PMO): What is possible without RANSAC and multiframe bundle adjustment?. Image and Vision Computing 2017.
84 2.12 % 0.0056 [deg/m] 0.03 s 2 cores @ 1.5 Ghz (C/C++)
Y. Zhao and P. Vela: Good Feature Matching: Towards Accurate, Robust VO/VSLAM with Low Latency. submitted to IEEE Transactions on Robotics 2019.
85
This method uses stereo information.
2.14 % 0.0059 [deg/m] 0.5 s 8 cores @ 3.5 Ghz (C/C++)
F. Bellavia, M. Fanfani, F. Pazzaglia and C. Colombo: . ICIAP 2013 2013.
F. Bellavia, M. Fanfani and C. Colombo: Selective visual odometry for accurate AUV localization. Autonomous Robots 2015.
M. Fanfani, F. Bellavia and C. Colombo: Accurate Keyframe Selection and Keypoint Tracking for Robust Visual Odometry. Machine Vision and Applications 2016.
86 2.24 % 0.0049 [deg/m] 0.11 s 1 core @ 2.5 Ghz (C/C++)
H. Mirabdollah and B. Mertsching: Fast Techniques for Monocular Visual Odometry . Proceeding of 37th German Conference on Pattern Recognition (GCPR) 2015 .
87 2.38 % 0.0053 [deg/m] 1 s 1 core @ 2.5 Ghz (Python + C/C++)
N. Fanani, M. Ochs, H. Bradler and R. Mester: . Intelligent Vehicles Symposium (IV) 2016.
N. Fanani, A. Stuerck, M. Barnada and R. Mester: Multimodal scale estimation for monocular visual odometry. Intelligent Vehicles Symposium (IV) 2017.
88
This method uses stereo information.
2.44 % 0.0114 [deg/m] 0.05 s 1 core @ 2.5 Ghz (C/C++)
A. Geiger, J. Ziegler and C. Stiller: . IV 2011.
89 2.54 % 0.0057 [deg/m] 0.03 s 5 cores @ 2.5 Ghz (C/C++)
S. Song and M. Chandraker: Robust Scale Estimation in Real-Time Monocular SFM for Autonomous Driving. CVPR 2014.
S. Song, M. Chandraker and C. Guest: Parallel, Real-time Monocular Visual Odometry. ICRA 2013.
90
This method uses stereo information.
2.54 % 0.0078 [deg/m] 1.26 s 1 core @ 2.5 Ghz (C/C++)
C. Beall, B. Lawrence, V. Ila and F. Dellaert: . IROS 2010.
91 2.54 % 0.0089 [deg/m] 0.1 s GPU @ 2.0 Ghz (C/C++)
92 2.55 % 0.0086 [deg/m] 0.39 s 1 core @ 2.5 Ghz (C/C++)
M. Mirabdollah and B. Mertsching: . Proceeding of 36th German Conference on Pattern Recognition 2014.
93
This method uses stereo information.
2.69 % 0.0068 [deg/m] 0.56 s 1 core @ 2.0 Ghz (C/C++)
P. Alcantarilla: . 2011.
P. Alcantarilla, J. Yebes, J. Almazán and L. Bergasa: . ICRA 2012.
94
This method uses stereo information.
2.86 % 0.0044 [deg/m] 0.1 s 1 core @ 3.0 Ghz (C/C++)
N. Dias and G. Laureano: Accurate Stereo Visual Odometry Based on Keypoint Selection. 2019 Latin American Robotics Symposium (LARS), 2019 Brazilian Symposium on Robotics (SBR) and 2019 Workshop on Robotics in Education (WRE) 2019.
95
This method uses stereo information.
2.94 % 0.0077 [deg/m] 0.06 s 1 core @ 2.5 Ghz (C/C++)
B. Kitt, A. Geiger and H. Lategahn: . IV 2010.
96 3.02 % 0.0132 [deg/m] 0.2 s 1 core @ 2.5 Ghz (C/C++)
97
This method uses stereo information.
3.13 % 0.0104 [deg/m] 0.57 s 1 core @ 2.0 Ghz (C/C++)
P. Alcantarilla: . 2011.
P. Alcantarilla, J. Yebes, J. Almazán and L. Bergasa: . ICRA 2012.
98
This method uses stereo information.
3.26 % 0.0095 [deg/m] 0.20 s 2 cores @ 2.5 Ghz (C/C++)
R. Gomez-Ojeda and J. Gonzalez- Jimenez: . Robotics and Automation (ICRA), 2016 IEEE International Conference on 2016.
99 3.45 % 0.0144 [deg/m] 0.1 s 1 core @ 2.5 Ghz (C/C++)
100 3.49 % 0.0128 [deg/m] 0.7 s 1 core @ 2.5 Ghz (C/C++)
M. Velas, M. Spanel, M. Hradis and A. Herout: . ArXiv e-prints 2017.
101
This method uses stereo information.
3.73 % 0.0107 [deg/m] 0.9 s 8 cores @ 3.0 Ghz (C/C++)
D. Mankowitz and E. Rivlin: CFORB: Circular FREAK-ORB Visual Odometry. arXiv preprint arXiv:1506.05257 2015.
102
This method uses stereo information.
3.94 % 0.0099 [deg/m] 0.51 s 1 core @ 2.0 Ghz (C/C++)
M. Kaess, K. Ni and F. Dellaert: . ICRA 2009.
P. Alcantarilla, L. Bergasa and F. Dellaert: . ICRA 2010.
103 4.03 % 0.0096 [deg/m] 1 s 1 core @ 2.5 Ghz (Python)
104
This method uses stereo information.
4.17 % 0.0112 [deg/m] 0.52 s 1 core @ 2.0 Ghz (C/C++)
M. Kaess, K. Ni and F. Dellaert: . ICRA 2009.
P. Alcantarilla, L. Bergasa and F. Dellaert: . ICRA 2010.
105
This method makes use of Velodyne laser scans.
4.19 % 0.0087 [deg/m] 0.05 s GPU @ 1.0 Ghz (Python + C/C++)
106 4.36 % 0.0052 [deg/m] .001 s GPU @ 2.5 Ghz (Matlab)
A. Aguilar-González, M. Arias- Estrada, F. Berry and J. Osuna-Couti?o: . Microprocessors and Microsystems 2019.
107 4.59 % 0.0175 [deg/m] 1 s 1 core @ 2.5 Ghz (C/C++)
M. Velas, M. Spanel, M. Hradis and A. Herout: . ArXiv e-prints 2017.
108 5.45 % 0.0274 [deg/m] 1 s 1 core @ 2.5 Ghz (Matlab)
Z. Boukhers, K. Shirahama and M. Grzegorzek: . Circuits and Systems for Videos Technology (TCSVT), IEEE Transaction on 2017.
Z. Boukhers, K. Shirahama and M. Grzegorzek: Less restrictive camera odometry estimation from monocular camera. Multimedia Tools and Applications 2017.
109 6.24 % 0.0097 [deg/m] 0.1 s 1 core @ 2.5 Ghz (C/C++)
110 6.42 % 0.0109 [deg/m] 0.1 s 1 core @ 2.5 Ghz (Python + C/C++)
111 7.10 % 0.0215 [deg/m] 0.15 s GPU @ 2.5 Ghz (Python)
112 7.40 % 0.0142 [deg/m] 0.1 s 1 core @ 2.5 Ghz (C/C++)
113 7.46 % 0.0245 [deg/m] 0.15 s 1 core @ 2.5 Ghz (C/C++)
A. Geiger, J. Ziegler and C. Stiller: . IV 2011.
S. Song and M. Chandraker: Robust Scale Estimation in Real-Time Monocular SFM for Autonomous Driving. CVPR 2014.
114 9.21 % 0.0163 [deg/m] 0.1 s 1 core @ 2.5 Ghz (C/C++)
M. Velas, M. Spanel, M. Hradis and A. Herout: . ArXiv e-prints 2017.
115 11.79 % 0.0069 [deg/m] 0.1 s 1 core @ 2.5 Ghz (C/C++)
116 11.94 % 0.0234 [deg/m] 0.1 s 1 core @ 2.5 Ghz (C/C++)
A. Geiger, J. Ziegler and C. Stiller: . IV 2011.
117 12.59 % 0.0312 [deg/m] 1 s 1 core @ 2.5 Ghz (C/C++)
118 13.25 % 0.0097 [deg/m] 0.03 s GPU @ 2.5 Ghz (Python)
I. Slinko, A. Vorontsova, F. Konokhov, O. Barinova and A. Konushin: . 2019.
119 13.69 % 0.0355 [deg/m] 0.01 s 1 core @ 2.5 Ghz (C/C++)
120 16.06 % 0.0320 [deg/m] 0.1 s 1 core @ 2.5 Ghz (C/C++)
121 20.95 % 0.0135 [deg/m] 0.5 s 1 core @ 3.5 Ghz (C/C++)
D. Frost, O. K?hler and D. Murray: Object-Aware Bundle Adjustment for Correcting Monocular Scale Drift. Proceedings of the International Conference on Robotics and Automation (ICRA) 2012.
122 21.47 % 0.0425 [deg/m] 0.01 s 1 core @ 2.5 Ghz (C/C++)
 | 
阅读(3045) | 评论(0) | 转发(0) |
0

上一篇：小觅景深摄像头校准 校准

下一篇：D435i 标定

给主人留下些什么吧！~~
关于我们 | 关于IT168 | 联系方式 | 广告合作 | 法律声明 | 免费注册

Copyright 2001-2010 ChinaUnix.net All Rights Reserved 北京皓辰网域网络信息技术有限公司. 版权所有

感谢所有关心和支持过ChinaUnix的朋友们

16024965号-6