Learning a Deep Metric: A Lightweight Relation Network for Loop Closure in Complex Industrial Scenarios

JIN Sheng; CHEN Liang; GAO Yu; SHEN Changqing; SUN Rongchuan

doi:10.1049/cje.2020.11.005

Volume 30 Issue 1

Jan. 2021

Turn off MathJax

Article Contents

Article Navigation > Chinese Journal of Electronics > 2021 > 30(1): 45-54

JIN Sheng, CHEN Liang, GAO Yu, et al., “Learning a Deep Metric: A Lightweight Relation Network for Loop Closure in Complex Industrial Scenarios,” Chinese Journal of Electronics, vol. 30, no. 1, pp. 45-54, 2021, doi: 10.1049/cje.2020.11.005

Citation:

JIN Sheng, CHEN Liang, GAO Yu, et al., “Learning a Deep Metric: A Lightweight Relation Network for Loop Closure in Complex Industrial Scenarios,” Chinese Journal of Electronics, vol. 30, no. 1, pp. 45-54, 2021, doi: 10.1049/cje.2020.11.005

Citation:

PDF( 4319 KB)

Learning a Deep Metric: A Lightweight Relation Network for Loop Closure in Complex Industrial Scenarios

doi: 10.1049/cje.2020.11.005

JIN Sheng^1
,,
CHEN Liang^{1
,
,},
GAO Yu^1
,,
SHEN Changqing^2
,,
SUN Rongchuan^1
,

1.
School of Mechanical and Electric Engineering, Soochow University, Suzhou 215137, China
2.
School of Rail Transportation, Soochow University, Suzhou 215137, China

Funds:

the National Natural Science Foundation of China 51875375

the National Natural Science Foundation of China 61673288

More Information

Author Bio:
JIN Sheng  was born in 1995. He received the M.S. degree from Soochow University in 2020. His research interests include computer vision and visual loop closure detection. (Email: 773903267@qq.com)

GAO Yu   received the Ph.D. degree in electrical engineering from Chonbuk National University. He is now a lecturer at Soochow University. His research topic is intelligent robots. (E-mail: ygao@suda.edu.cn)

SHEN Changqing   received the Ph.D. degree in systems engineering and engineering management from the City University of Hong Kong in 2014. He is currently an Associate Professor at Soochow University. His research topic is intelligent fault diagnosis. (Email: cqshen@suda.edu.cn)

SUN Rongchuan   received the Ph.D. degree in mechanical and electronic engineering from Shenyang Institute of Automation Chinese Academy of Sciences. He is now an Associate Professor at Soochow University. His research topic is visual SLAM. (E-mail: sunrongchuan@suda.edu.cn)
Corresponding author: CHEN Liang (corresponding author) received the Ph.D. degree in control engineering from Zhejiang University in 2009. He is now an Associate Professor at Soochow University. His research topic is deep learning based intelligent control. (E-mail: chenl@suda.edu.cn)
Received Date: 2019-08-26
Accepted Date: 2020-07-10
Publish Date: 2021-01-01

Abstract

Abstract

The similarity metric in Loop closure detection (LCD) is still considered in an old fashioned way, i.e. to pre-define a fixed distance function, leading to a limited performance. This paper proposes a general framework named LRN-LCD, i.e. a Lightweight relation network for LCD, which combines the feature extraction module and similarity metric module into a simple and lightweight network. The LRN-LCD, an end-to-end framework, can learn a non-linear deep similarity metric to detect loop closures from different scenes. Moreover, the LRN-LCD supports image sequences as input to speed up the similarity metric in real-time applications. Extensive experiments on several open datasets illustrate that LRN-LCD is more robust to strong condition variations and viewpoint variations than the mainstream methods.
- Loop closure detection,
- Relation network,
- Image sequences,
- End to end

FullText(HTML)

References(22)

References

[1]	A. Angeli, S. Doncieux, J. A. Meyer, et al., "Real–time visual loop–closure detection", Proc. of IEEE International Conference on Robotics and Automation, Pasadena, USA, pp. 1842–1847, 2008.
[2]	M. Zaffar, A. Khaliq, S. Ehsan, et al., "Are state–of–the–art visual place recognition techniques any good for aerial robotics?", arXiv preprint arXiv: 1904.07967v1, 2019.
[3]	H. Bay, A. Ess, T. Tuytelaars, et. al., "Speeded–up robust features (SURF)", Computer Vision and Image Understanding, Vol. 110, No. 3, pp. 346–359, 2008. doi: 10.1016/j.cviu.2007.09.014
[4]	D. G. Lowe, "Distinctive image features from scale–invariant keypoints", International Journal of Computer Vision, Vol. 60, No. 2, pp. 91–110, 2004. doi: 10.1023/B:VISI.0000029664.99615.94
[5]	D. Gálvez–López and J. D. Tardós, "Realtime loop detection with bags of binary words", Proc. of IEEE International Conference on Intelligent Robots and Systems, San Francisco, California, USA, pp. 51–58, 2011.
[6]	R. Mur–Artal, J. M. M. Montiel and J. D. Tardos, "ORB–SLAM: A versatile and accurate monocular SLAM system", IEEE Transactions on Robotics, Vol. 31, No. 5, pp. 1147–1163, 2015. doi: 10.1109/TRO.2015.2463671
[7]	M. Cummins and P. Newman, "FAB–MAP: Probabilistic localization and mapping in the space of appearance", The International Journal of Robotics Research, Vol. 27, No. 6, pp. 647–665, 2008. doi: 10.1177/0278364908090961
[8]	Y. Chen, W. Gan, L. Zhang, et al., "A survey on visual place recognition for mobile robots localization", Proc. of Web Information Systems and Applications Conference, Liuzhou, China, pp. 187–192, 2017.
[9]	S. Lowry, N. Sünderhauf, P. Newman, et al., "Visual place recognition: A survey", IEEE Transactions on Robotics, Vol. 32, No. 1, pp. 1–19, 2016. doi: 10.1109/TRO.2015.2496823
[10]	Z. Chen, O. Lam, A. Jacobson, et al., "Convolutional neural network–based place recognition", Proc. of Australasian Conference on Robotics and Automation, Victoria, Australia, pp. 1–8, 2014.
[11]	N. Sünderhauf, S. Shirazi, F. Dayoub, et al., "On the performance of ConvNet features for place recognition", Proc. of IEEE International Conference on Intelligent Robots and Systems, Germany, pp. 4297–4304, 2015.
[12]	Y. Hou, H. Zhang and S. Zhou, "Convolutional neural network–based image representation for visual loop closure detection", Proc. of IEEE International Conference on Information and Automation, Lijiang, China, pp. 2238–2245, 2015.
[13]	B. Zhou, A. Lapedriza, J. Xiao, et al., "Learning deep features for scene recognition using places database", Proc. of International Conference on Neural Information Processing Systems, pp. 487–495, 2014.
[14]	D. Olid, J. M. Fácil and J. Civera, "Singleview place recognition under seasonal changes", arXiv preprint arXiv: 1808.06516v1, 2018.
[15]	Z. Chen, A. Jacobson, N. Sünderhauf, et al., "Deep learning features at scale for visual place recognition", Proc. of IEEE International Conference on Robotics & Automation, Singapore, Singapore, pp. 3223–3230, 2017.
[16]	B. Chen, D. Yuan, C. Liu, et al., "Loop closure detection based on multi–scale deep feature fusion", Applied Sciences, Vol. 9, No. 6, Article No. 1120, 2019. doi: 10.3390/app9061120
[17]	Q. Li, K. Li, X. You, et al., "Place recognition based on deep feature and adaptive weighting of similarity matrix", Neurocomputing, Vol. 199, pp. 114–127, 2016. doi: 10.1016/j.neucom.2016.03.029
[18]	O. Vinyals, C. Blundell, T. Lillicrap, et al., "Matching networks for one shot learning", Proc. of Neural Information Processing Systems, Barcelona, Spain, pp. 3637–3645, 2016.
[19]	C. Finn, P. Abbeel and S. Levine, "Modelagnostic meta–learning for fast adaptation of deep networks", Proc. of International Conference on Machine Learning, Sydney, NSW, Australia, pp. 1126–1135, 2017.
[20]	S. Ravi and H. Larochelle, "Optimization as a model for few–shot learning", Proc. of International Conference on Learning Representations, Toulon, France, 2017.
[21]	J. Snell, K. Swersky and R. S. Zemel, "Prototypical networks for few–shot learning", Proc. of International Conference on Neural Information Processing Systems, Long Beach, USA, pp. 4080–4090, 2017.
[22]	P. Wohlhart and V. Lepetit, "Learning descriptors for object recognition and 3D pose estimation", Proc. of IEEE Conference on Computer Vision and Pattern Recognition, Boston, MA, USA, pp. 3109–3118, 2015.