Hongdong Li

Hongdong Li

RSISE, Building-115, CECS
Australian National University (ANU)
ACT 0200, Australia

Email: firstname.lastname@anu.edu.au
Tel: +61 2 6125 7708 (RSISE office)

About Me

         I am a faculty member of the College of Engineering and Computer Science,  at ANU (Australian National University).    I do research in Computer Vision, Pattern Recognition, Image Processing, Machine Learning, Artificial Intelligence  @ANU.   My main interests include 3D Vision, Robot Vision,  Visual Odometry,  UAV and robot navigation, Virtual and Augmented Reality,  Image Processing,  Optimisation, Pattern Recognition and Deep Learning. 

     Currently I am a Chief Investigator and AA Theme Leader for Australia ARC Centre of Excellence for Robotic Vision.     I serve in the Program Committees and Area Chair for CVPR, IEEE  ICCV,  ECCV and BMVC, and an Associate Editor for IEEE- TPAMI, IET-Computer Vision, IVC-Journal of Image and Vision Computing, ISPJ Journal of CVA (Computer Vision and Application), etc.    I serve the Area Chair for IEEE ICCV, CVPR, and ECCV in recent years, as well as an Area Chair for BMVC and 3DV, etc.   I am a Program Co-Chair for ACRA 2015 (Australiasian conference on robotics and automation),   Program Co-Chair for the 1st International workshop on Big Data in 3D Computer Vision at ICCV’13,   Application of Computer Vision in Biological Science workshop at ICCV’13, and  Area Chair for ECCV 2016 and ICCV 2017.

       Previously, I was a contributing researcher to the Bionic Vision Australia and as an AI-Associate Investigator on the ARC-funded Australia Bionic Eye Project (special initiative).  Prior to 2010 I was a Fellow with the RSISE @ ANU and a Senior Research Scientist with NICTA (National ICT Australia), Canberra Labs, where I am greatly privileged to be working under the mentoring of Professor Richard Hartley who is a pioneer researcher in 3D computer vision and a co-author of MVG (the bible in Multiview geometry, (amazon link) , Professor Hartley and Professor A. Zisserman).  My research career in 3D computer vision started from my post-doc period under Richard’s  supervision.     My researches have been funded by the ARC (Australia Research Council), ARC Centre of Excellence, National ICT Australia (NICTA), Bionic Vision Australia (via the Australia Bionic Eye project), Microsoft Research (via an ARC Linakge grant), NSFC (National Natural Science Foundation of China, as International Partner Investigator), as well as global companies and manufacturers (Toshiba, JVC, Microsoft Australia, General Motors, etc.).   Potential practical application areas include 3D reconstruction, 3D modelling, virtual reality and augmented reality, robot vision navigation and SLAM, machine learning, multimedia information retrieval, autonomous driving, human robot interaction, image understanding, medical imaging, visual data analysis, and UAV drone autonomous flying, visual guidance.

In the past, jointly with my PhD students I have won a number of prestigious awards in Computer Vision research, including the CVPR Best Paper Award,   IEEE ICPR Best Student Paper Award, IEEE ICIP Best Student Paper Award, DSTO Best Paper Prize, CiSRA Best Paper Award at DICTA.   I have supervised and co-supervised over 30 PhD students, most of them in the area of Computer Vision and Pattern recognition, Image processing, Geometry, Robotic Vision, and Machine Learning.    ANU has various scholarship schemes available for PhD applicants.   We also welcome students with other source of funding (e.g. Chinese CSC scholarship, Brazilian, and Chilean Government PhD Scholarship) to apply the ANU Computer Vision PhD Program.  I obtained my PhD and MSc degree from Zhejiang University China, where I conducted research in robotics, and pattern classification.   I was a member of the ALV research group –China’s first autonomous driving vehicle. 

ANU is the finest research university in Australia, located in the beautiful “Bush Capital”- city of Canberra.   

Current Research Projects

1.       Novel Camera Sensors for robust robotic vision (funded by ARC Centre of Excellence for Robotic Vision).

2.       Unorganised image based City-scale 3D modelling (funded by an ARC Discovery project).

3.       Visual Navigation for autonomous driving cars (self-driving cars, funded by a global automobile manufacturer). 

4.       High-performance computing and human-computer interaction for medical images (funded by Microsoft Research, via an ARC Linkage grant).

5.       Non-rigid Shape and Motion Capture and Motion Segmentation via subspace or sub-manifold clustering.

6.       Global optimisation and inference for MRF/CRF.

7.       Artificial neural networks and deep learning for visual object tracking and object detection.

8.      Mathematics Optimisation in low level computer vision problems.

9.       Autonomous driving/flying of autonomous vehicle  in all-weather conditions.

Research Team:

Dr.  Yuchao Dai,  ARC DECRA Fellow;Dr.  Laurent Kneip,  ARC DECRA Fellow;Dr.  Viorela Ila,  ACRV Research Fellow; Dr.  Anoop Cherian,  ACRV Research Fellow;Dr. Dingfu Zhou,   ANU Postdoctoral Fellow ;PhD students: ;Jiaolong Yang,  Jie Liang, Cristian  Rodrigiuze,  Joey Zhou,  Mehdani M., Liu Liu, Yiran Zhong, Dylan Campbell, Pan Ji, Sadeep Jayasumana,; Ajanthan Thalaiyasinga,Gao Zhu,Yansheng Ming, Mohammad Sadegh,Zhipeng Xiao, Jae-hak Kim, Jun Sun, Gucan Long.

Selected Recent Publications

The CVPR, ICCV and ECCV are premier conferences in Computer Vision research community. The IEEE TPAMI,TIP, IJCV are top journals in the field of computer vision, image and pattern recognition.


Kernel Subspace Clustering for Motion Segmentation,  ICCV 2017 submission 4


Globally Optimal PnP for Perspective Camera Pose,  ICCV 2017 submission 3 


Non-Rigid Shape Reconstruction, ICCV 2017  submission 2


Efficient Image Based Camera Localization, ICCV 2017 submission 1

Neural Aggregation Networks for Video face Recognition,  CVPR 2017

Learning image matching by watching a video,  ECCV 2016

Rolling Shutter Camera Relative Pose, IEEE CVPR 2016

Robust Optical Flow Estimation of Double-Layer Images under Transparency or Reflection,  IEEE CVPR 2016

Beyond Local Search: Tracking Objects Everywhere with Instance-Specific Proposals, IEEE CVPR 2016

Robust Multi-body Feature Tracker: A Segmentation-free Approach,  IEEE CVPR 2016

Pan Ji, Mathieu Salzmann, Hongdong Li, 

Shape Interaction Matrix Revisited and Robustified: Efficient Subspace Clustering with Corrupted and Incomplete Data,  IEEE ICCV 2015, Oral (3% acceptance rate).

Training an artificial neural networks for image edge detection,   JZJU-Science, vol-2, 2000.

3D computer vision for self-driving car:  Visual odometry and vehicle ego-motion estimation for autonomous driving using side-viewing multiple-camera rig,  TR-ANU-CVRG, Jan.  2016.

Alvaro Parra Bustos, Tat-Jun Chin, et al,  Fast Rotation Search with Stereographic Projections for 3D Registration, accepted by IEEE T-PAMI, 2016.

A. Thalaiyasingam, R. Hartley, M. Salzmann and H. Li
Iteratively Reweighted Ishikawa Graph Cut for Multi-label MRFs with Non-convex Priors,

IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Boston, MA, June 2015.

Jiaolong Yang, Hongdong Li, 

Dense, Highly-Accurate Optical Flow Computation: the Piecewise Parametric Model Revisited.

IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Boston, MA, June 2015. ( Code will be available soon)

Lie-Struck: Affine Tracking on Lie Groups Using Structured SVM.

WACV : 63-70

Rank Minimization or Nuclear-Norm Minimization: Are We Solving the Right Problem?

DICTA : 1-8 (DSTO Best Paper Award)

IEEE ICIP : 283-287 (Best Student Paper Award)

Pan Ji, et. al.

Subspace method for multibody motion segmentation with unknown correspondences,  in ECCV 2014. 




Hongdong Li

A Parctical Algorithm for Quasi-convex Triangulation with Outliers, <full paper in PDF>, in IEEE CVPR 2007. (revised version: Thank Carl Olsson and Fredrik Kahl).


Jiaolong Yang, , Jia:

Optimal Essential Matrix Estimation via Inlier-Set Maximization.

ECCV (1) : 111-126  


Expanding the Family of Grassmannian Kernels: An Embedding Perspective.

ECCV (7) :






L. Kneip,  H. Li

Efficient Computation of Relative Pose for Multi-Camera Systems,  in IEEE CVPR 2014. 




Optimizing over Radial Kernels on Compact Manifolds. 

IEEE CVPR : 3802-3809(ORAL  presentation ) .



Efficient dense subspace clustering.

IEEE WACV : 461-468



Jiaolong Yang,Yuchao Dai,Hongdong Li,Henry Gardner,Yunde Jia

Single-Shot Extrinsic Calibration of a Generically Configured RGB-D Camera Rig from Scene Constraints, IEEE ISMAR 2013.  (acceptance rate <= 25%)


Yansheng Ming,   Hongdong Li,  Xuming He

Symmetry Detection by Contour Grouping,  IEEE ICIP 2013.


Jiaolong Yang,   Hongdong Li,   Yunde Jia

Go-ICP: a new method for registration of 3D surfaces optimally and efficiently,  International Conference on Computer Vision, ICCV 2013. ( Globally Optimal ICP algorithm:  free matlab code is available)


Jae-hak Kim, Yuchao Dai, Hongdong Li, Xin Du, Jon Kim

Multi-View 3D Reconstruction from Uncalibrated Radially-Symmetric Cameras,   International Conference on Computer Vision, ICCV 2013.


Sadeep  Jayasumana , Mathieu  Salzmann, Hongdong  Li, Mehrtash  Harandi

A Framework for Shape Analysis via Hilbert Space Embedding,  International Conference on Computer Vision, ICCV 2013.


Hongdong Li, Richard Hartley,

Five Point Motion Estimation Made Easy, <full paper in PDF>, Pattern Recognition, 18th International Conference, ICPR 2006.


Richard Hartley, Jochen Trumpf, Yuchao Dai, Hongdong Li

Rotation Averaging, <full paper in PDF>, International Journal of Computer Vision - IJCV,  2013.


Jean-Charles Bazin, Hongdong Li, In So Kweon, Cedric Demonceaux, P. Vasseur, Katsushi Ikeuchi

A Branch and Bound Approach to Correspondence and Grouping Problems, IEEE Transactions on Pattern Analysis and Machine Intelligence, T-PAMI, vol. 99, 2012.


Yansheng Ming, Hongdong Li, Xuming He

Connected Contours: a New Contour Completion Model that Respects the Closure Effect, in IEEE CVPR 2012 (ORAL). (acceptance rate is 2.5%, 48 papers accepted out of 1933 submissions) , June 2012.


Yuchao Dai , Hongdong Li, Mingyi He

A Simple Prior-free Method for Non Rigid Structure from Motion Factorization, in CVPR 2012 (ORAL). IEEE CVPR Best Paper Award-2012   (prestigious award in international Computer Vision community. The winner out of 1,933 submissions in 2012).  Free Matlab source code is available


Hongdong Li

Globally Optimal Consensus Maximization for Robust Multiple view Geometry Estimation, <full paper in PDF>, In ICCV 2009, Kyoto, 2009.


Hongdong Li

Multi-view Structure Computation without Explicitly Estimating Motion, in CVPR 2010 (ORAL), (acceptance rate is 4.8%, out of 1500 submissions)



Sadeep Jayasumana, Richard Hartley, Mathieu Salzmann, Hongdong Li and Mehrtash Harandi.

Combining Multiple Manifold-valued Descriptors for Improved Object Recognition, IEEE DICTA 2013,  CiSRA"Best Recognition Paper Award".



W. Kusakunniran, Q. Wu, J. Zhang, H. Li

Multiple view transformation based approach to view-invariant human gait recognition, International Conference on Pattern Recognition, ICPR 2010. (ICPR Best Student paper Award (Biometric Track)).


Yansheng Ming, Hongdong Li, Xuming He

Joint Contour Cue and Region cue Salient Image Segmentation  (PDF coming soon..). IEEE  CVPR 2013 (poster).


Hongdong Li

Consensus Set Maximization with Guaranteed Global Optimality for Robust Geometry Estimation, <full paper in PDF>, In ICCV 2009, Kyoto, 2009.


Hongdong Li,

Efficient Reduction for solving L_infinity Problems in Multiview Geometry, <preprint in PDF>. In proc. IEEE CVPR 2009, Miami, 2009.


Hongdong Li, Richard Hartley

The 3D-3D Registration Problem Revisited, in Proc. ICCV 2007. (This paper gives a branch and bound method for searching the rotation space and permutation space.)


Richard Hartley, Hongdong Li

An Efficient Hidden Variable Approach to Minimal-Case Camera Motion Estimation, IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI), IEEE PAMI 2012.


Hongdong Li

Two-view Motion Segmentation from Linear Programming Relaxation, <full paper in PDF>, in IEEE CVPR 2007 (ORAL).


Hongdong Li

A Practical Algorithm for L-infinity Triangulation with Outliers, <full paper in PDF>, in IEEE CVPR 2007. (revised version: Thank Carl Olsson and Fredrik Kahl).


Chunhua Shen, Hongdong Li, and Michael Brooks

Supervised Dimensionality Reduction via Sequential Semi-definite Programming,  Pattern Recognition, Elsevier, 2008.


Yuchao Dai, Hongdong Li, and MinyiHe

Element-wise Factorization for N-view Projective Reconstruction, <full paper in PDF>. in ECCV 2010. Yuchao DAI was funded by CSC (Chinese Scholarship Council).


Hongdong Li

A Simple Solution to the Six-point Two-view Focal-length Algorithm, <Source code can be found in the paper PDF>, in ECCV 2006.


Y.Dai , J.Trump, H. Li, Richard Hartley

Rotation Average with Application to Multi-camera Rig Calibration, in Proc. ACCV 2009.


H. Li, R. Hartley, J. Kim

A Linear Approach to Motion Estimation Using Generalized Camera Models, <preprint pdf, CVPR08> in IEEE CVPR 2008. [Two-view motion estimation for Multi-camera rig system, or more general generalized camera models]


Jae-hak Kim, H. Li, R.Hartley

Motion Estimation for a Multi-camera System Using Global Optimization, in Proc. IEEE CVPR 2008.


John Lim, Nick Barnes, Hongdong Li

Estimating Relative Camera Motion from the Antipodal Epipolar Constraint,   IEEE Transactions on Pattern Analysis and Machine Intelligence, PAMI 2011.


Hongdong Li, Richard Hartley

A New Conformal Spherical Representations for 3D Shapes, Pattern Recognition, an International Journal, Elsevier Publisher, 2007. DOI Bookmark: 10.1016/j.patcog.2007.01.021 (5-Year Impact Factor: 3.725).


Hongdong Li and Chunhua Shen

Interactive color image segmentation with linear programming, Journal of Machine Vision and Applications, (MVA), 2010.


Wen, Z.Y.; Fraser, D.; Lambert, A.; H. Li

Reconstruction of underwater Image by Bispectrum, IEEE International Conference on Image Processing, ICIP 2007.
(Youtube Link: http://www.youtube.com/watch?v=hMsXBWZdFjc )



Hongdong Li,  Richard Hartley

A new and compact algorithm for simultaneously matching and estimation,

in Proc IEEE- ICASSP 2004.


Hongdong Li,  Richard Hartley

A Non-iterative Method for Correcting Lens Distortion from Nine Point
  in Omni-Vision'05,  in conjunction with IEEE - ICCV 2005,  Beijing.  (Free Matlab source code is available on request).



< For more papers, please see Publications page >


Back to Top  >

Disclaimer: The information provided on this page, and on others beginning with http://rsise.anu.edu.au/~hongdong, is the personal responsibility of the author, and not to be taken as those of the ANU.
© 2013