Alessio Del Bue Homepage

The second set of sensor is now ready in the PLUS lab. We have a number of sensors to detect the depth based on TOF (Swissranger SR4000), Stereo (TXYZ) and a brand new Panasonic D-IMager. Ah, almost forgetting, a Kinect as well. Depth sensing is not the only invisible pattern we are interested in. We have equipped the lab with a brand new A315 thermocamera from Flir.

The tutorial is approaching soon, here you can find the official flyer.

We are building a multi-sensors room called Cerberus in the PLUS lab at the underground level in IIT with the aim to test different sensor modalities. These sensors are grouped in three different modalities (heads):

• Video
• Audio
• Unseen

The first head has just been born. We have equipped our lab space with 5 Basler high resolution color cameras. The stream of the 5 cameras is syncrhonised up to frame precision in order to perform accurate (human) motion analysis. This system will be soon joined by the two other heads. We are currently working on a set of audio arrays and thermal/depth sensing devices. More news soon!

A two day tutorial (6-7 December) on “Deformable Image Registration and
Non-Rigid 3D Reconstruction from Images” will be organised by Lourdes
Agapito, Adrien Bartoli, Alessio Del Bue and hosted by the PLUS lab at
the Italian Institute of Technolgy, Genova, Italy.

ABSTRACT:
The tutorial presents modern computer vision techniques to deal with the
registration and 3D reconstruction of deformable shapes using images.
The aim of this 2-day course is to present a principled procedure for
dealing with images where each object can arbitrarily change its shape
– a common occurrence in human motion analysis, medical imaging and
video-surveillance scenarios. The tutorial will discuss a set of
techniques general in their formulation but that can be customized given
the specific imaging problem of the user. In particular, an emphasis is
put over the use of physical or statistical priors which can aid the
solution of such an ill-posed problem. Real examples on the human motion
analysis and medical imaging domains will show the effectiveness of the
approaches in dealing with different deforming shapes.

SYLLABUS:
Non-rigid image registration:
– image deformation models
– feature-based registration
– pixel-based (direct) registration
Non-rigid 3D reconstruction:
– template-based 3D reconstruction
– Non-Rigid Structure-from-Motion

VENUE and DATES:
This 2 day tutorial (6-7 December) will take place in the Italian
Institute of Technology (www.iit.it), Genova, Italy. The tutorial is
free of charge but please send an email to alessio.delbue@iit.it to
confirm your attendance to the tutorial or to ask for further
information. Please check: www.isr.ist.utl.pt/~adb/tutorial/ for further
details and updates on the tutorial schedule.

You can find in the code page a MATLAB implementation of the BALM algorithm presented this year at ECCV 2010. The code is useful to optimise cost functions in bilinear form and it is customised to deal with arbitrary manifold constraints and missing data. In particular, in this release we provide a solution for the Non-rigid Structure from Motion (NRSFM) and Photometric Stereo problems along with their manifold projectors. Please refer to this page for a more detailed description of the BALM algorithm.

• A. Del Bue, J. Xavier, L. Agapito, and M. Paladini, "Bilinear Factorization via Augmented Lagrange Multipliers," in 11th European Conference on Computer Vision (ECCV 2010), Crete, Greece, 2010, pp. 283-296.
  @inproceedings{DelBue:etal:2010,
  author = {A. {Del Bue} and J. Xavier and L. Agapito and M. Paladini},
  title = {Bilinear Factorization via Augmented Lagrange Multipliers},
  editor = {Kostas Daniilidis and Petros Maragos and Nikos Paragios},
  booktitle = {11th European Conference on Computer Vision (ECCV 2010), Crete, Greece},
  publisher = {Springer},
  location = {Heidelberg},
  series = {Lecture Notes in Computer Science},
  volume = {6314},
  year = {2010},
  isbn = {978-3-642-15560-4},
  pages = {283--296}
}
  bibtex

There is more material and a new web page for the Adaptive 3D reconstruction and registration papers. The ECCV 2010 presentation can be found here (PDF – 1.85MB) along with the paper “Adaptive Metric Registration of 3D Models to Non-rigid Image Trajectories” (PDF – 590KB).

• A. Del Bue, "Adaptive Metric Registration of 3D Models to Non-rigid Image Trajectories," in 11th European Conference on Computer Vision (ECCV 2010), Crete, Greece, 2010, pp. 87-100.
  @inproceedings{DelBue:2010,
  author = {A. {Del Bue}},
  title = {Adaptive Metric Registration of 3D Models to Non-rigid Image Trajectories},
  editor = {Kostas Daniilidis and Petros Maragos and Nikos Paragios},
  booktitle = {11th European Conference on Computer Vision (ECCV 2010), Crete, Greece},
  publisher = {Springer},
  location = {Heidelberg},
  series = {Lecture Notes in Computer Science},
  volume = {6313},
  year = {2010},
  isbn = {978-3-642-15557-4},
  pages = {87--100}
}
  bibtex

At ECCV 2010 I will present a work on registering rigid shapes to non-rigid image trajectories. The algorithm simultaneously solves for the transformation parameters (a rotation plus an image projection) and find an adapted 3D shape to the image trajectories. In this video you can see some results of the algorithm:

If you need more details, there is a page in my research space dedicated to this class of approaches.

• A. Del Bue, "Adaptive Metric Registration of 3D Models to Non-rigid Image Trajectories," in 11th European Conference on Computer Vision (ECCV 2010), Crete, Greece, 2010, pp. 87-100.
  @inproceedings{DelBue:2010,
  author = {A. {Del Bue}},
  title = {Adaptive Metric Registration of 3D Models to Non-rigid Image Trajectories},
  editor = {Kostas Daniilidis and Petros Maragos and Nikos Paragios},
  booktitle = {11th European Conference on Computer Vision (ECCV 2010), Crete, Greece},
  publisher = {Springer},
  location = {Heidelberg},
  series = {Lecture Notes in Computer Science},
  volume = {6313},
  year = {2010},
  isbn = {978-3-642-15557-4},
  pages = {87--100}
}
  bibtex

This year Lourdes Agapito, Adrien Bartoli and I are going to present the 2nd edition of the tutorial on Computer Vision in a Non-Rigid World. It will take place at the end of November at the Istituto Italiano di Tecnologia (IIT) in Genova. The tutorial will cover image registration problems for deformable shapes, 3D reconstruction of deforming and articulated bodies, head and body pose estimation and many other “nonrigid” problems.

If you are interested in the first edition slide you can check last year slides. I will try to post soon more details about the exact date and program. Hope to see you in Genova!

After the PrintArt workshop, now is the time for the first PLUS-VIPS school on Social Signal Processing (July 18-22). The themes treated by the school are aimed to give an introduction to Social Signal Processing. This new research field tries to bridge the social intelligence gap between people and machines. At its core, social intelligence consists of sensing nonverbal behavioral cues displayed by people around us (facial expressions, gestures, vocalizations, postures), interpreting these cues in terms of social signals (relational attitudes like interest, hostility, empathy, agreement and disagreement, or dominance), and displaying as a response natural, consistent behaviors (interest for those we are interested in, or hostility for those we are hostile to).

The school is organized and sponsored by our PLUS (Pattern analysis, Learning, and image Understanding Systems) laboratory of the Istituto Italiano di Tecnologia, Genova (Italy) jointly with the VIPS (Vision, Image Processing, and Sound) lab of the University of Verona.

The location of the school is in the wonderful Baia del Silenzio in Sestri Levante:

View Larger Map

Certainly a perfect match for the warm Ligurian weather!