Alessio Del Bue Homepage

We are PAVIS

admin — Mon, 09 Jan 2012 10:23:37 +0000

The new year brings some changes…we are now PAVIS – Pattern Analysis and Computer Vision at IIT.

Nature methods cover

admin — Sat, 17 Dec 2011 19:21:29 +0000

The work in collaboration with the Nanophysics facility at IIT has the cover on the December issue of Nature Methods!

F. Cella Zanacchi, Z. Lavagnino, M. Perrone, A. Del Bue, L. Furia, M. Faretta, and A. Diaspro, "Live-cell 3D super-resolution imaging in thick biological samples," Nature Methods, pp. 1047-1049, 2011.

@Article{Cella:etal:NM2011, author = {F. {Cella Zanacchi} and Z. Lavagnino and M. {Perrone} and A. {Del Bue} and L. Furia and M. Faretta and A. Diaspro}, title = {{Live-cell 3D super-resolution imaging in thick biological samples}}, journal = {Nature Methods}, pages = {1047-1049}, year = {2011}, month = {December}, url = {http://www.nature.com/nmeth/journal/v8/n12/full/nmeth.1744.html}, Doi = {10.1038/nmeth.1744} }

bibtex

ICCV tutorial slides online

admin — Wed, 16 Nov 2011 22:15:47 +0000

ICCV 2011 Tutorial on

Non-Rigid Registration and Reconstruction

Sunday, 6 November 2011, 13th International Conference on Computer Vision (ICCV 2011), Barcelona, Spain

Online Slides

Tutorial opening and introduction
Deformable Image Registration
Template-Based Deformable Shape Inference
Non-rigid Structure from Motion (NRSfM): Low-rank methods
NRSfM: Trajectory space and Piecewise methods

You can check the tutorial page for more information and links to the previous tutorial editions.

2nd PrintART Workshop

admin — Sun, 13 Nov 2011 17:28:21 +0000

2nd PrintART workshop

My colleagues in Lisbon have organised the second edition of the workshops of the PrintART project. Please check this link for the full program and for a description of the workshop theme. In the following you can find an excerpt from the workshop introduction:

The tile paintings is rightly characterized as one of the art forms that more easily identifies the Portuguese artistic heritage. It has enjoyed a growing attention in recent years and the number of links between the analysis of the tiles work and traditionally unrelated knowledge domains has been increasing. For example, the Robotics and, particularly, Computer Vision, is one of these domains and the PrintART Project, underway since 2010, can represent a milestone in this relationship between different fields of knowledge.

[...]

The main objective of the PrintART project is to provide a tool for retrieving the prints that inspired a given painting on a panel of tiles. This tool compares prints and images of tiles to find matching shapes and compositions. The question that the project seeks to answer is: by introducing an image of a panel of tiles in a vast database of images of prints, will it be able to identify one or more prints similar to that panel?

For the art history researcher, to look for prints that inspired the artists to paint a particular composition is still a job that requires thorough study and enormous capacity to exploit the researcher’s visual memory. Although the human eye is still infinitely more effective that any software tool, new technologies, and massive digitization of books and images, present great opportunities to the community.

[...]

The use of statistical techniques for pattern recognition has enormous potential to perform automatic analysis of databases of art images, thus reducing the time normally spent by the observer who does not use any software tool. The technology developed in this project shows promising results and the use of these new techniques can open new perspectives to art history.

Computer Vision on several orders of magnitude

admin — Mon, 24 Oct 2011 22:19:28 +0000

The exciting aspect of the Istituto Italiano di Tecnologia (IIT) is the possibility to work on several aspects of science. This year was indeed very intense, until now I did not quite realise on which extent Computer Vision was applied in the projects we have in our group. While I was trying to put some order in the scientific activities there was something quite surprising. The visual shapes we were dealing with spanned from the smallest biological molecules to astral bodies in space — almost reaching a total of 14 orders of magnitude in size. Here there is a summary from the smallest to the biggest:

• 3D localisation of fluorescent particles for Nanoscopy – 10^-8: Localising the tiniest molecules in thick biological samples is a challenge in modern Microscopy. In this project we helped the IIT Nanophysics group to raise the state of the art in terms of accuracy in 3D localisation. More details on the new Nature methods paper here.

F. Cella Zanacchi, Z. Lavagnino, M. Perrone, A. Del Bue, L. Furia, M. Faretta, and A. Diaspro, "Live-cell 3D super-resolution imaging in thick biological samples," Nature Methods, pp. 1047-1049, 2011.

@Article{Cella:etal:NM2011, author = {F. {Cella Zanacchi} and Z. Lavagnino and M. {Perrone} and A. {Del Bue} and L. Furia and M. Faretta and A. Diaspro}, title = {{Live-cell 3D super-resolution imaging in thick biological samples}}, journal = {Nature Methods}, pages = {1047-1049}, year = {2011}, month = {December}, url = {http://www.nature.com/nmeth/journal/v8/n12/full/nmeth.1744.html}, Doi = {10.1038/nmeth.1744} }

bibtex

• Analysis of the morphology of neuronal networks – 10^-5:
This ongoing work is based on the in vitro analysis of the morphology of neuronal networks using images from a high-density microelectrode array (MEA). The aim is to extract the positions and links among different cluster of neurons. This information will be coupled together with electro-physiological signals in order to understand the functionality of the network. In collaboration with Alessandro Maccione and Luca Berdondini from the Neuroscience and Brain Technology (NBT) department @ IIT.

• Automatic microphones localisation – 10^-2:
The set of microphones and the audio arrays in the Cerberus lab need a precise 3D localisation in order to work. Recently we have found a new set of computational tools able to localise the microphone position solely from sound events. The overall accuracy of the system is less then half a centimeter. More news soon!

• 3D deformable modelling – 10⁰:
With Multiview 3D warps – M3DW we presented together with A. Bartoli a framework that can model dense deformations of an image from an image sequence. The ideal applications are surface retexturing and deformation cloning.

A. Del Bue and A. Bartoli, "Multiview 3D Warps," in 13th Internationl Conference on Computer Vision (ICCV 2011), Barcelona, Spain, 2011.

@inproceedings{DelBue:Bartoli:2011, author = {A. {Del Bue} and A. Bartoli}, title = {Multiview 3D Warps}, booktitle = {13th Internationl Conference on Computer Vision (ICCV 2011), Barcelona, Spain}, year = {2011}, month = {November}, url = {http://users.isr.ist.utl.pt/~adb/publications/2011_DelBue_Bartoli_ICCV11.pdf} }

bibtex

• Large-scale 3D reconstruction – 10³:
We have recently extended the BALM algorithm to deal with large-scale reconstruction of a size such as Piazza San Marco in Venice. A new update of the code and the data will be published soon.

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

• Asteroid 3D reconstruction – 10⁵:
We have created in collaboration with the University of Trento a new pipeline for the 3D reconstruction of Asteroids using close approach images. The Lutetia asteroid size is of 100 Km at the diameter!

L. Baglivo, A. Del Bue, M. Lunardelli, F. Setti, V. Murino, and M. De Cecco, "A Method for Asteroids 3D Surface Reconstruction from Close Approach Distances," in Computer Vision Systems, 2011, pp. 21-30.

@inproceedings{Baglivo:etal:2011, author={L. Baglivo and A. {Del Bue} and M. Lunardelli and F. Setti and V. Murino and M. {De Cecco}}, title = {A Method for Asteroids 3D Surface Reconstruction from Close Approach Distances}, booktitle = {Computer Vision Systems}, series = {Lecture Notes in Computer Science}, editor = {Crowley, James and Draper, Bruce and Thonnat, Monique}, publisher = {Springer Berlin / Heidelberg}, isbn = {978-3-642-23967-0}, keyword = {Computer Science}, pages = {21-30}, volume = {6962}, note = {10.1007/978-3-642-23968-7_3}, year = {2011} }

bibtex

Cerberus – Third Head

admin — Sun, 14 Aug 2011 19:59:57 +0000

The last set of sensors is ready and this conclude the instruments available at the PLUS lab.
Now the lab can listen using a set of microphones that we will use to test novel self-calibration algorithms for microphone arrays. Also, our postdoc Marco Crocco has built a custom 32 microphone array for acoustic imaging applications. In the future, some projects in our lab will exploit the multi-modality of all these sensors also in the context of video-surveillance applications.

Multiview 3D Warps @ ICCV 2011

admin — Tue, 05 Jul 2011 17:09:17 +0000

NEWS: there is a page with more details about the M3DW, please check this link.

Adrien Bartoli and I are going to present a new way of modelling shape deformations that tries to unify the classical image warp modelling and Structure from Motion approaches. We call such framework Multiview 3D Warps (M3DW). More news about the approach will appear soon on these pages. For the moment, a preview of a retexturing application using M3DW:

Original video sequence:

Retextured Logo:

Tutorial @ ICCV 2011

admin — Tue, 24 May 2011 21:29:50 +0000

Lourdes, Adrien and I are going to present the next edition of the tutorial on Computer Vision in a Non-rigid World at ICCV 2011 this year. Please check the tutorial page for information and links to the previous tutorial editions. Hope to see you in Barcelona!!!

2D and 3D Visual Recognition School

admin — Thu, 17 Mar 2011 20:17:36 +0000

Next week the 2nd PLUS school will take place at the Italian Institute of Technology (IIT). The school will cover the object recognition and scene understanding themes. For 4 days Fei-Fei Li and Silvio Savarese will present the recent advancements in this exciting field of Computer Vision. More details at the school web page.

NORDIA 2011 – Fourth Workshop on Non-Rigid Shape Analysis and Deformable Image Alignment

admin — Wed, 09 Feb 2011 11:12:07 +0000

I am co-organising the Fourth Workshop on Non-Rigid Shape Analysis and Deformable Image Alignment (NORDIA’11) to be held in conjuction with CVPR 2011 conference. Here you can find the call for papers:

Fourth Workshop on Non-Rigid Shape Analysis and Deformable Image Alignment (NORDIA’11)
24 June 2011, Colorado Springs, USA

Chairs:
Lourdes Agapito – Queen Mary, University of London
Adrien Bartoli – Université d’Auvergne
Alex Bronstein – Tel Aviv University
Michael Bronstein – University of Lugano
Alessio Del Bue – Italian Institute of Technology

Nonrigid and deformable shapes are ubiquitous in the world surrounding us, on all levels from micro to macro. The need to study such shapes and model their behavior arises in a wide spectrum of applications, ranging from medical imaging and augmented reality to design engineering and homeland security. In recent years, problems of nonrigid shape analysis and deformable image alignment have attracted a growing interest in computer vision and pattern recognition communities, which has led to rapid development of the field, where state-of-the-art results from very different sciences – theoretical and numerical geometry, optimization, linear algebra, graph theory, machine learning and computer graphics, to mention a few are applied to find solutions.

NORDIA 2011 comes following the success of the previous issues of the workshop in 2008, 2009 and 2010. The purpose is to bring together leading researchers dealing with different aspects of nonrigid shape analysis and deformable image alignment in order to promote new interdisciplinary collaborations and expose each side to the most recent results and problems in each field. The purpose is to identify new problems as well as potential solutions. The unique value of the workshop is bringing together people from communities traditionally considered to be working in different areas and rarely meeting in the same conferences.

Topics
• Deformable models
• Shape similarity and recognition
• Large-scale non-rigid shape retrieval
• Partial shape similarity
• Invariant shape descriptors
• Self-similarity and symmetry of nonrigid shapes
• Geometric and topological noise modeling
• Structural shape similarity and correspondence
• 2D/2D, 2D/3D, and 3D/3D alignment and nonrigid correspondence problems
• Inverse problems involving nonrigid shapes
• Synthesis of nonrigid shapes
• Learning-based methods Important dates
• Efficient optimization algorithms
• Merging of detection and alignment
• Multimodal alignment and sensor fusion
• Applications

Paper submission: 5 March
Notification of acceptance: 5 April
Workshop: 24 June