Piotr Biliński, Ph.D.


Post-Doctoral Researcher at University of Oxford


I am a Post-Doctoral Researcher at University of Oxford in England since September 2016.

I work on Semantic 3D Dense SLAM, which links:

  • dense geometry reconstruction using depth fusion, and

  • semantic labelling that uses deep learning.

INRIA

Research Interests: Computer Vision, Machine Learning, Artificial Intelligence, Algorithms.

In particular: semantic segmentation, gesture, action, activity and behavior detection (recognition and localization), object detection, tracking, image and video retrival, deep learning, artificial intelligence, machine learning, data mining, statistics, audio and signal processing, recommendation systems, optimization, algorithms and data structures.

My main Research Experience is in the topics of: semantic segmentation, human, group and crowd action and activity detection, object detection, object and feature point tracking, neural networks, deep learning, HRTF personalization, recommendation systems, and others.


From January 2015 till September 2016, I was a Post-Doctoral Researcher at INRIA in France, Sophia-Antipolis Research Center, STARS team.

I was working on automatic video understanding and behavior detection in videos, in particular on:

  • Detection of human actions in videos (recognition and localization),

  • Assessment of people cognitive and behavioral status from activities of daily living (detecting people with dementia and estimating their health status; home care environments),

  • Crowd behavior detection in surveillance videos, violence and fight recognition and localization.
INRIA


From April till August 2015, I was a Visiting Resercher at the Honeywell Laboratory, Research Institute, where I was working on video understanding in surveillance videos for security applications; in particular, on automatic recognition and localization of crowd violence in surveillance videos.
Honeywell


In December 2014, I obtained my Ph.D. degree in Computer Science (Computer Vision) from the INRIA Research Institute and from the University of Nice Sophia-Antipolis in France (Topic: Human Action Recognition in Videos |
Jury: Ramakant Nevatia, Frederic Jurie, Ivan Laptev, Frederic Precioso, Francois Bremond | 2010 - 2014 | more information). During my Ph.D. studies, I was conducting research at the INRIA under the supervision of François Brémond, Reseach Director.
INRIA
University of Nice Sophia-Antipolis

During my Ph.D.:
  • I was a Research Intern at the Microsoft Research in Redmond, United States, where I was working in audio and signal processing domain, on HRTF Personalization using Anthropometric Features. I created algorithms for automatic extraction of anthropometric features from 3D head scans, and I proposed several techniques for HRTF recommendation and HRTF synthesis. More information on the official website of the Microsoft Research.
    U.S. Patent.
Microsoft Research
  • I was working on trajectory management and analysis as a Student Developer at the Google Inc. / OpenCV as part of Google Summer of Code.
Google


Prior to that, I obtained my M.Sc. degree in Computer Science (2008 - 2009) and B.Sc. degree in Computer Science (2004 - 2008) from the Poznan University of Technology in Poland. During this time, I recevied award scholarship at every semester, and Minister of Science and Higher Education Award for outstanding academic achievements. I was also participating in various algorithmic and IT competitions, and I had some considerable achievements. Among various scientific and industrial experiences, I was a research intern in the PULSAR team at INRIA, where I was conducting research on visual tracking - on feature point and multiple object tracking in videos that is working with occlusions and is based on appearance descriptors.
Poznan University of Technology



piotr.bilinski [_ at _] eng.ox.ac.uk

If you have any questions or queries, or you would like to know more about me, please feel free to drop me a line.

Ph.D.


On December 5, 2014, I obtained my Ph.D. degree in Computer Science from the INRIA and from the University of Nice Sophia-Antipolis in France. The topic of my Ph.D. was "Human Action Recognition in Videos" (in French: "Reconnaissance d’action humaine dans des vidéos"). All my research was conducted at the INRIA Sophia-Antipolis Research Institute under the supervision of François Brémond, Reseach Director.

The jury of my Ph.D. defense was:
  • Ramakant Nevatia, Professor, University of Southern California, reviewer,
  • Frederic Jurie, Professor, University of Caen, reviewer,
  • Frederic Precioso, Professor, Polytech Nice Sophia, president,
  • Ivan Laptev, Research Director, INRIA Paris, examinator,
  • Francois Bremond, Research Director, INRIA Sophia-Antipolis, advisor.

During my Ph.D., I was also:
  • Research Intern at the Microsoft Research, Redmond, USA,
  • Student Developer at the Google Inc. / OpenCV as part of Google Summer of Code.
but this work is not included in my manuscript. Information about this work can be found here.

Activities & Memberships


Recent Professional Activities & Memberships
Member of the Program Committee of the IEEE International Conference on Automatic Face and Gesture Recognition (IEEE FG), Xi'an, China, 2018.
Member of the Technical Program Committee of the IEEE International Conference on Automatic Face and Gesture Recognition (IEEE FG), Washington, D.C., USA, 2017.
Member of the Technical Program Committee of the IEEE International Conference on Advanced Video and Signal based Surveillance (IEEE AVSS), Lecce, Italy, 2017.
Member of the Program Committee of the International Conference of the Biometrics Special Interest Group (BIOSIG), Darmstadt, Germany, 2017.
Member of the Program Committee of the IEEE International Symposium on Multimedia (IEEE ISM), Taichung, Taiwan, 2017.
Member of the Technical Program Committee of the International Conference ACM Multimedia (ACM MM), Amsterdam, Netherlands, 2016.
Member of the Program Committee of the International Workshop on Crowd Understanding, organized in conjunction with the ECCV 2016, Amsterdam, Netherlands, 2016.
Member of the Program Committee of the IEEE International Symposium on Multimedia (IEEE ISM), San Jose, California, USA, 2016.
Computer Vision and Image Understanding (CVIU) Journal Reviewer.
IEEE Transactions on Circuits and Systems for Video Technology Reviewer.
IEEE Signal Processing Letters Reviewer.
Pattern Recognition Letters Reviewer.
Signal Processing Reviewer.
KSII Transactions on Internet and Information Systems Reviewer.
ICCV 2013 Reviewer.
IEEE Student Member.
ACM Student Member.

Invited Talks in 2015 & 2016

Invited Talks and Presentations in 2015 & 2016.
DateEvent & Place
08.2017Presentation of our paper at the 25th European Signal Processing Conference (EUSIPCO) in Kos island.
09.2016Oral presentation of our paper at the 15th International Conference of the Biometrics Special Interest Group in Darmstadt.
07.2016Invited talk at the University of Southern California in Los Angeles.
07.2016Oral presentation at the 13th IEEE International Conference on Advanced Video and Signal based Surveillance (IEEE AVSS) in Colorado Springs.
12.2015Invited talk at the CENTAUR 2nd Annual Webinar.
07.2015Oral and poster presentations at the 24th International Joint Conference on Artificial Intelligence (IJCAI) in Buenos Aires.
06.2015Invited talk at the Center for Machine Perception (CMP), Department of Cybernetics, Czech Technical University (CTU) in Prague.
05.2015Invited talk at the Institute of Information Theory and Automation (UTIA, Ústav teorie informace a automatizace) in Prague.
05.2015Invited talk at the Honeywell, for Honeywell Prague, Honeywell Brno and Honeywell USA.
01.2015Invited talk at the Agency for Science, Technology and Research (A*STAR)'s Institute for Infocomm Research (I2R) in Singapore.
01.2015Invited talk at the Nanyang Technological University (NTU) in Singapore.

Publications



New paper accepted. More information soon.

Expression Recognition for Severely Demented Patients in Music Reminiscence - Therapy
A. Dantcheva,  P. Bilinski,  T. Nguyen, J. C. Broutart, and F. Bremond
25th European Signal Processing Conference (EUSIPCO)
Kos island, Greece, August-September 2017


Emotion facial recognition by the means of automatic video analysis
A. Dantcheva,  P. Bilinski,  J. C. Broutart, P. Robert, and F. Bremond
Gerontechnology, vol. 15, pp. 12s
September 2016

AVSS 2016

Real-Time Crowd Violence Recognition and Detection in Surveillance Videos
P. Bilinski and F. Bremond
13th International Conference on Advanced Video and Signal-Based Surveillance (AVSS)
Colorado Springs, Colorado, United States, August 2016

BIOSIG 2016
Can a smile reveal your gender?
P. Bilinski, A. Dantcheva, and F. Bremond
15th Internatiional Conference of the Biometrics Special Interest Group (BIOSIG)
Oral presentation
Darmstadt, Germany, September 2016

U.S. Patent: HRTF Personalization based on Anthropometric Features

U.S. Patent:

HRTF Personalization based on Anthropometric Features

P. Bilinski, J. Ahrens, M. Thomas, I. Tashev, J. Platt, and D. Johnston
United States, Pub. No.: US 2015/0312694 A1, Pub. Date: Oct. 29, 2015.
IJCAI 2015
Video Covariance Matrix Logarithm for Human Action Recognition in Videos
P. Bilinski and F. Bremond
24th International Joint Conference on Artificial Intelligence (IJCAI)
Oral and poster presentations
Buenos Aires, Argentina, July 2015

Acceptance rate: 28.8% (1996 submission)
Article

PDF

BibTex

Piotr Bilinski's Ph.D. Thesis
Human Action Recognition in Videos
P. Bilinski
Ph.D. Thesis
University of Nice Sophia-Antipolis
Sophia-Antipolis, France, December 2014

The jury of the Ph.D. defense:
  • Ramakant Nevatia, Professor, University of Southern California, reviewer,
  • Frederic Jurie, Professor, University of Caen, reviewer,
  • Frederic Precioso, Professor, Polytech Nice Sophia, president,
  • Ivan Laptev, Research Director, INRIA Paris, examinator,
  • Francois Bremond, Research Director, INRIA Sophia-Antipolis, advisor.
Article

PDF

BibTex

ICIP 2014

3D Trajectories for Action Recognition
M. Koperski, P. Bilinski, F. Bremond
21st IEEE International Conference on Image Processing (ICIP)
Paris, France, October 2014
Article

PDF

BibTex

AVSS 2014

Representing Visual Appearance by Video Brownian Covariance Descriptor
P. Bilinski, M. Koperski, S. Bak, F. Bremond
11th IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS)
Oral presentation
Seoul, Korea, August 2014
Article

PDF

BibTex

ICASSP 2014

HRTF Magnitude Synthesis via Sparse Representation of Anthropometric Features
P. Bilinski, J. Ahrens, M. R. P. Thomas, I. J. Tashev, J. C. Platt
39th International Conference on Acoustics, Speech and Signal Processing (ICASSP)
Florence, Italy, May 2014

More details about this approach is avaiable in my presentation at the Microsoft Research.
Article

PDF

BibTex

VISAPP 2014
Towards Unsupervised Sudden Group Movement Discovery for Video Surveillance
S. Zaidenberg, P. Bilinski and F. Bremond
9th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISAPP)
Oral presentation
Lisbon, Portugal, January 2014
Article

PDF

BibTex

FG 2013
Relative Dense Tracklets for Human Action Recognition
P. Bilinski, E. Corvee, S. Bak and F. Bremond
10th IEEE International Conference on Automatic Face and Gesture Recognition (FG)
Oral presentation
Shanghai, China, April 2013

Video demos are available here.
Article

PDF

BibTex

ECCV 2012 WS
Statistics of Pairwise Co-occurring Local Spatio-Temporal Features for Human Action Recognition
P. Bilinski and F. Bremond
4th VECTaR2012 workshop, 12th European Conference on Computer Vision (ECCV)
Oral presentation
Firenze, Italy, October 2012
Article

PDF

BibTex

AVSS 2012

Contextual Statistics of Space-Time Ordered Features for Human Action Recognition
P. Bilinski and F. Bremond
9th IEEE International Conference on Advanced Video and Signal-Based Surveillance (AVSS)
Oral presentation
Beijing, China, September 2012
Article

PDF

BibTex

ICVS 2011

Evaluation of local descriptors for action recognition in videos
P. Bilinski and F. Bremond
8th International Conference on Computer Vision Systems (ICVS)
Oral presentation
Sophia Antipolis, France, September 2011
Article

PDF

BibTex

ICDP 2009

Multiple Object Tracking with Occlusions using HOG Descriptors and Multi Resolution Images
P. Bilinski, F. Bremond and M.-B. Kaaniche
3rd International Conference on Imaging for Crime Detection and Prevention (ICDP)
London, United Kingdom, December 2009
Article

PDF

BibTex

PETS 2009
Crowd Event Recognition Using HOG Tracker
C. Garate, P. Bilinski and F. Bremond
12th IEEE International Workshop on Performance Evaluation of Tracking and Surveillance (Winter-PETS), IEEE Winter Vision Meetings 2009 (WVM)
Oral presentation
Snowbird, USA, December 2009
Article

PDF

Article

Videos & Demos



Violent Crowd Behavior Detection in Surveillance Videos.

The following videos present samples of a violent crowd behavior detection in surveillance videos. Red box indicates a detection of a violence (2 seconds before it occurs). Videos with violence localization and more information will be added soon.




Action Recognition in Videos.

The following videos present head estimation algorithm and extracted dense trajectories. The head estimation is based on people, head and face detectors and an object tracking algorithm. More information is presented in our paper: P. Bilinski, E. Corvee, S. Bak, F. Bremond, Relative Dense Tracklets for Human Action Recognition, 10th IEEE International Conference on Automatic Face and Gesture Recognition (FG), 2013.
Videos are blurred to preserve the privacy of individuals.




HRTF Personalization using Anthropometric Features.

My presentation at the Microsoft Research in Redmond, United States given on September 27, 2013.

Abstract: Head-related transfer functions (HRTFs) represent the acoustical transfer function between a sound source and the entrance of the blocked ear canal. Imposing HRTFs onto a non-spatial audio signal and playing back the result over headphones evokes the perception of a virtual 3D auditory space. The commercial breakthrough of HRTF-based audio has been hindered by the fact that HRTFs are highly individual and using HRTFs other than the user’s own can significantly impair the result. Since the measurement of HRTFs is an expensive process, automatic selection or synthesis of HRTFs is desirable. We therefore propose methods for HRTF selection and synthesis based on Decision Trees, Sparse Representation and Neural Networks. These methods are based on the mapping of people’s anthropometric features and other personal information to the acoustic HRTFs. The results are analyzed both objectively using measures such as spectral distortion as well as in a listening experiment.

Video with captions, slides, transcript and more are available on the official website of the Microsoft Research.




Face, Eye and Pupil Detection in a Web Camera.

By curiosity, I spent a few hours during a weekend in 2013 to write a small toy software detecting face, eyes and pupils in a web camera (in this case camera built in my laptop). Two simple videos presenting the result (person looking straight ahead, up, down, left and right) are included below. Of course, there are plenty of opportunities to improve this toy method.


Working with Internet Explorer, Firefox, Google Chrome and http protocol. The video player may not work with the https protocol.

Source Code

The links below are intended for people I am collaborating with.
Source Code
Covariance for IDT (C++).
Distance covariance for IDT (C++).
Improved Fisher Vectors with spatial information encoding (C++ & Matlab).
SVM with linear and χ2 kernels for a single machine and a PC cluster (C++).
MKL with linear and χ2 kernels for a single machine and a PC cluster (C++).
Algorithms Library (C++), including: Dijkstra, Bellman Ford, Max-Flow (Ford-Fulkerson, Dinic), Min-Cost Max-Flow (Edmonds and Karp), Minimum Spanning Tree (Prim, Kruskal), Knuth-Morris-Pratt, Manacher's algorithm, QuickSort, MergeSort, HeapSort, Interval Tree, BST Tree, Convex Hull, etc.