Michael Isard
Researcher
Microsoft Research Silicon Valley
 
misard@microsoft.com
650 693 1474

Background

I received my D.Phil in computer vision from the Oxford University Engineering Science Department in 1998. In 1999 I started work as a Researcher at the Compaq Systems Research Center in Palo Alto, and I have worked for Microsoft Research in Silicon Valley since 2002. The majority of my early research was in the field of visual tracking and sequential filtering, and I helped to introduce particle filters to the computer vision community with the Condensation algorithm. From the time I joined Compaq my interests have broadened to include distributed systems research, and this is where I am spending the majority of my time at Microsoft.

I spent much of mid-2003 to early 2005 working closely with the MSN Search product group on the design and implementation of their V1 search engine. I was involved from the start of the project and owned several production components in the live query pipeline. I was particularly involved in the design and implementation of the monitoring and failure recovery systems, the query distribution and result aggregation system, and the inverted file lookup.

Current research projects cover a range from inference methods for Bayesian networks with applications in visual tracking and dense stereo estimation; to programming models for large-scale distributed systems and many-core processor architectures. I led the Dryad project to build a large-scale distributed execution platform for datacenters that is now widely used internally by Microsoft product groups. I have recently started working with several colleagues on AME, a new concurrent programming model that makes use of transactional memory without explicit atomic blocks. I have also been collaborating with Andrew Zisserman's Visual Geometry Group in Oxford on various aspects of object retrieval from large databases.

Systems Publications

Refereed papers

Semantics of Transactional Memory and Automatic Mutual Exclusion
Martín Abadi, Andrew Birrell, Tim Harris and Michael Isard
Proc. Symp. Principles of Programming Languages, 63-74 (2008)
Automatic Mutual Exclusion
Michael Isard and Andrew Birrell
Proc. Workshop on Hot Topics in Operating Systems (2007)
Dryad: Distributed Data-Parallel Programs from Sequential Building Blocks
Michael Isard, Mihai Budiu, Yuan Yu, Andrew Birrell, and Dennis Fetterly
European Conference on Computer Systems (EuroSys), 59-72 (2007)
An Internet Protocol (IP) Sound System
Bob Atkinson, Tom Blank, Michael Isard, James D (JJ) Johnston and Kirk Olynyk
AES 177th Convention (2004)

Invited papers

Autopilot: Automatic Data Center Management
Michael Isard
Operating Systems Review 41(2): 60-67, April 2007.
A Design for High-Performance Flash Disks
Andrew Birrell, Michael Isard, Chuck Thacker and Ted Wobber
Operating Systems Review 41(2): 88-93, April 2007.

Computer Vision, Machine Learning and Graphics Publications

Recent publications

Total Recall: Automatic Query Expansion with a Generative Feature Model for Object Retrieval
Ondřej Chum, James Philbin, Josef Sivic, Michael Isard and Andrew Zisserman
Proc Int. Conf. Computer Vision (2007)
Scalable Near Identical Image and Shot Detection
Ondřej Chum, James Philbin, Michael Isard and Andrew Zisserman
Proceedings of the International Conference on Image and Video Retrieval (CIVR) 2007.
Object retrieval with large vocabularies and fast spatial matching
James Philbin, Ondřej Chum, Michael Isard, Josef Sivic and Andrew Zisserman
Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2007.

Book

Active Contours
Andrew Blake and Michael Isard
Springer-Verlag, 1998.

Thesis

Visual Motion Analysis by Probabilistic Propagation of Conditional Density
Michael Isard
D.Phil. Thesis, Oxford University, 1998

Journal papers

Distributed rendering of interactive soft shadows
Michael Isard, Mark Shand and Alan Heirich
Parallel Computing 29 (3): 311-323, March 2003
Bayesian Object Localisation in Images
Josephine Sullivan, Andrew Blake, Michael Isard and John MacCormick
International Journal of Computer Vision 44 (2): 111-135, September 2001
Statistical models of visual shape and motion
Andrew Blake, Benedicte Bascle, Michael Isard and John MacCormick
Proc. Roy. Soc. Lond. A, 356, 1283--1302, (1998)
CONDENSATION -- conditional density propagation for visual tracking
Michael Isard and Andrew Blake
Int. J. Computer Vision, 29, 1, 5--28, (1998)
Learning to track the visual motion of contours
Andrew Blake, Michael Isard and David Reynard
Artificial Intelligence, 78, 101--134, (1995)

Selected conference papers

Total Recall: Automatic Query Expansion with a Generative Feature Model for Object Retrieval
Ondřej Chum, James Philbin, Josef Sivic, Michael Isard and Andrew Zisserman
Proc Int. Conf. Computer Vision (2007)
Scalable Near Identical Image and Shot Detection
Ondřej Chum, James Philbin, Michael Isard and Andrew Zisserman
Proceedings of the International Conference on Image and Video Retrieval (CIVR) 2007.
Object retrieval with large vocabularies and fast spatial matching
James Philbin, Ondřej Chum, Michael Isard, Josef Sivic and Andrew Zisserman
Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2007.
Estimating Disparity and Occlusions in Stereo Video Sequences
Oliver Williams, Michael Isard and John MacCormick
Proc Computer Vision and Pattern Recognition, (II) 250-257 (2005)
Tracking Loose-limbed People
Leonid Sigal, Sidharth Bhatia, Stefan Roth, Michael J Black and Michael Isard
Proc Computer Vision and Pattern Recognition, (I) 421-428 (2004)
Attractive People: Assembling Loose-Limbed Models using Non-parametric Belief Propagation
Leonid Sigal, Michael Isard, Benjamin H. Sigelman and Michael J. Black
Advances in Neural Information Processing Systems 16, 1539-1546 (2003)
Pampas: Real-Valued Graphical Models for Computer Vision
Michael Isard
Proc Computer Vision and Pattern Recognition, (I) 613-620 (2003)
Automatic Camera Calibration from a Single Manhattan Image
Jon Deutscher, Michael Isard and John MacCormick
Proc European Conf. Computer Vision, LNCS 2352, 175-188 (2002)
BraMBLe: A Bayesian Multiple-Blob Tracker
Michael Isard and John MacCormick
Proc Int. Conf. Computer Vision, vol. 2, 34-41 (2001)
Partitioned sampling, articulated objects, and interface-quality hand tracking
John MacCormick and Michael Isard
Proc European Conf. Computer Vision, vol. 2, 3-19 (2000)
Object Localization by Bayesian Correlation
Josephine Sullivan, Andrew Blake, Michael Isard and John MacCormick
Proc Int. Conf. Computer Vision, 1068-1075, (1999)
Learning multi-class dynamics
Andrew Blake, Ben North and Michael Isard
Advances in Neural Information Processing Systems 11, 389-395, MIT Press, (1999)
ICONDENSATION: Unifying low-level and high-level tracking in a stochastic framework
Michael Isard and Andrew Blake
Proc 5th European Conf. Computer Vision, vol. 1 893-908, (1998)
A smoothing filter for Condensation
Michael Isard and Andrew Blake
Proc 5th European Conf. Computer Vision, vol. 1 767-781, (1998)
A mixed-state Condensation tracker with automatic model-switching
Michael Isard and Andrew Blake
Proc 6th Int. Conf. Computer Vision, 107-112, (1998)
The Condensation Algorithm --- conditional density propagation and applications to visual tracking
Andrew Blake and Michael Isard
Advances in Neural Information Processing Systems 9, 361--368, ed. T. Petsche, MIT Press (1997)
Contour tracking by stochastic propagation of conditional density
Michael Isard and Andrew Blake
Proc. European Conf. on Computer Vision, vol. 1, pp. 343--356, Cambridge UK, (1996)
3D position, attitude and shape input using video tracking of hands and lips
Andrew Blake and Michael Isard
Proc. ACM Siggraph, 185--192, (1994)
Learning to track curves in motion
Andrew Blake, Michael Isard and David Reynard
Proc. IEEE Int. Conf. Decision Theory and Control, 3788--3793, (1994)