|
|
My name
is Gabriel Brostow, and I am a professor
in Computer
Science here at UCL. My group
explores research problems relating to Computer
Vision and Computer Graphics. The students and
colleagues here have diverse interests, but my focus
is on "Human in the Loop Computer Vision," for
analysis and synthesis applications. To me, this
means having or finding satisfying answers to these
questions about a system, whether interactive or
fully automated:
I) Does the system
know the intended purpose of the data being
captured?
II) Can the system
assess its own accuracy?
III) Does the
system ask questions of users, when it's
uncertain?
I love this field because it allows us to apply our
expertise to a variety of tough problems, including
film and photo special effects (computational
photography), action analysis (of people, animals,
and cells), and authoring systems (for architecture,
animation, presentations) that make the most of user
effort. "Motion reveals everything" used to be my
main research mantra, but that has now taken hold
sufficiently (obviously NOT just through my
efforts!) that it no longer needs championing.
Previously, I had a visiting researcher appointment
at ETH Zurich's CVG Group,
and a Marshall
Sherfield
Fellowship in the Computer
Vision & Robotics Group at Cambridge
University. I did my PhD with Irfan
Essa at Georgia
Tech, and my undergraduate degree at UT Austin. |
|
News:
See my group's YouTube
Channel.
(June 2017) Our work on detecting bats in the Olympic Park
hit the press today!
See the BBC
article on the front of their Science section, or
the front-page
of UCL's site.
Some behind-the-scenes context:
The real machine learning R&D was done by Daniyar
Turmukhambetov, Oisin Mac Aodha, and Michael Firman. All
are current/recent postdocs from our group. Most started
as MSc students here!
Sarah Gallacher is the linchpin of the project. Not
surprising, as she started with a background in software
engineering and machine learning, and then got into HCI.
There are many other people involved, and you should read
about them and see the project landing page: https://naturesmartcities.com/.
I'll just add that Libby Kinsey did her MSc a few years
ago in our group, working with Oisin and Kate Jones, to
get the initial bat-call detector working on full-size
computers. So what you're seeing running in the Olympic
Park is the result of these people's hard work to build ML
models, herd volunteers, gather data, build hardware,
shrink ML models, waterproof prototypes, etc. That's
several people's very hard work over a sustained period of
time! Thanks to EPSRC EP/K015664/1 and EP/K503745/1 for
co-funding this work.
Collecting some of the press coverage here:
http://www.ucl.ac.uk/news/headlines/0617/290617-shazam-for-bats
To see the bats ping'ing, check out the live map (best at
night): http://batslondon.com.
(Mar-Sep 2017) I am Program Co-Chair for BMVC 2017.
Come to the conference in September!
We had a record number of submissions, as this is a top
venue for publishing new work in Computer Vision and
Machine Learning.
(May 2017) Our handwriting-synthesis
project has passed 50k views on YouTube.
(Oct 2016) I'll be an Area Chair for CVPR 2017 - the top
ranked Computer Science publication.
(July 2016) I have joined IEEE TPAMI as an Associate Editor.
Submit your great papers, so they can appear in
the top-ranked journal in all of Computer Science!
(Sep 2016) We have filed for a patent on our MonoDepth
work! (Now a CVPR 2017 paper)
Students &
Postdocs:
Alumni:
Research:
C Ilisescu, H. A.
Kanaci, M. Romagnoli, N. Campbell, G.J. Brostow, "Responsive Action-Based Video Synthesis,"
2017, conditionally accepted to SIGCHI
- Project Page
- BibTex
- PDF
|
|
T. Intharah, D.
Turmukhambetov, G.J. Brostow, (HILC) "Help, It Looks Confusing:
GUI Task Automation Through Demonstration and
Follow-up Questions," in ACM Intelligent
User Interfaces (IUI) 2017
- Project Page
- BibTex
- PDF
|
|
C. Godard, O. Mac
Aodha, G.J. Brostow, "Unsupervised
Monocular Depth Estimation with Left-Right
Consistency," 2016, on arXiv for
now.
- Project
Page
- PDF
|
|
P. Hedman, T.
Ritschel, G. Drettakis, G.J. Brostow, "Scalable Inside-Out
Image-Based Rendering,"
conditionally accepted to SIGGRAPH ASIA 2016
- Project
Page
- BibTex
- PDF
|
|
D. E. Worrall, C.
Wilson, G. Brostow, "Automated
Retinopathy of Prematurity Case Detection with
Convolutional Neural Networks," in
Deep
Learning
in Medical Image Analysis 2016
- Project
Page
- BibTex
- PDF
|
|
W. Li, F. Viola, J.
Starck, G.J. Brostow, N.D.F. Campbell, "Roto++:
Accelerating Professional Rotoscoping Using
Shape Manifolds," SIGGRAPH 2016
- Project
Page
- BibTex
- PDF
|
|
M. Firman, O. Mac
Aodha, S. Julier, G.J. Brostow, "Structured
Prediction of Unobserved Voxels From a Single
Depth Image," CVPR 2016
- Project
Page
- BibTex
- PDF
|
|
D. Worrall, G.J.
Brostow, C. Wilson, "Automated Optic
Disk Localization in The Neonatal Fundus Image,"
ARVO
2016
- Project
Page
- BibTex
- PDF
|
|
T. Haines, O. Mac
Aodha, G.J. Brostow, "My Text in Your
Handwriting," Transactions on
Graphics (ToG), to be presented at SIGGRAPH 2016
- Project
Page
- BibTex
- PDF
|
|
C. Godard, P.
Hedman, W. Li, G.J. Brostow, "Multi-view
Reconstruction
of Highly Specular Surfaces in Uncontrolled
Environments," 3DV 2015
- Project
Page
- Code
- BibTex
- PDF
|
|
A. Monszpart, N.
Mellado, G.J. Brostow, N.J. Mitra, "Rebuilding Scans of
Man-made Scenes with Regular Arrangements of
Planes," SIGGRAPH 2015 /
Transactions on Graphics (ToG).
- Project
Page
- BibTex
- PDF
|
|
E. Johns, O. Mac
Aodha, G.J. Brostow, "Becoming
the
Expert - Interactive Multi-Class Machine
Teaching," CVPR 2015
- Project
Page
- Video
Presentation
- BibTex
- PDF
|
|
M. Gryka, M. Terry,
G.J. Brostow, "Learning
to Remove Soft Shadows," Transactions on
Graphics (ToG), presented at SIGGRAPH 2015
- Project
Page
- PDF
- BibTex
|
|
D. Waithe, P.
Rennert, G.J. Brostow, M.D.W. Piper, "QuantiFly:
Robust
Trainable Software for Automated Drosophila Egg
Counting," PLoS ONE, 2015
- Project
Page
- Code
- BibTex
|
|
O. Mac
Aodha, N.D.F. Campbell, J. Kautz,
G.J. Brostow, "Hierarchical
Subquery
Evaluation
for Active Learning on a Graph," CVPR 2014
- Project
Page
- Talk
Video
- BibTex
- PDF
|
|
Hugo
Lopez-Tovar, G.J. Brostow, "Learning
Sketch-based 3D Modelling From A User's
Sketching Gestures," Sketch: Pen and
Touch Recognition, IUI Workshop 2014
- PDF
- BibTex
|
|
O. Mac Aodha, V.
Stathopoulos, M. Terry, K.E. Jones, G. J. Brostow,
M. Girolami, "Putting the
Scientist in the Loop - Accelerating Scientific
Progress with Interactive Machine Learning,"
Invited paper in ICPR 2014
- Project
Page
- PDF
- BibTex
|
|
O. Mac Aodha, G.J.
Brostow, "Revisiting
Example
Dependent Cost-Sensitive Learning with Decision
Trees," ICCV 2013
- Project
Page
- BibTex
- PDF
|
|
O. Mac Aodha, A.
Humayun, M. Pollefeys, G.J. Brostow, "Learning
a Confidence Measure for Optical Flow," in
IEEE PAMI 2012
- Project
Page
- BibTex
- PDF
|
|
C. Garc?
Cifuentes, M. Sturzel, F. Jurie G.J.
Brostow, "Motion
Models
That Only Work Sometimes," BMVC 2012
- Project
Page
- BibTex
- PDF
|
(Ksummit17's
input
video http://youtu.be/ygEFhh1MPOE)
|
O. Mac Aodha,
N.D.F. Campbell, A. Nair, G.J. Brostow, "Patch
Based Synthesis for Single Depth Image
Super-Resolution,"
ECCV 2012
- Project
Page
- BibTex
- PDF
|
|
A. Humayun, O. Mac
Aodha, G.J. Brostow, "Learning to Find
Occlusion Regions,"
CVPR 2011
- Project
Page
- BibTex
- PDF
|
|
M. Reynolds, J.
Dobo?, L. Peel, T. Weyrich, G.J. Brostow, "Capturing
Time-of-Flight Data with Confidence,"
CVPR 2011
- Project
Page
- BibTex
- PDF
|
|
G.J. Brostow, C.
Hern?dez, G. Vogiatzis, B. Stenger, R. Cipolla, "Video Normals from
Colored Lights," IEEE PAMI
2011
- PDF
- Project
Page
(does not include PDF though)
- Bibtex
|
|
A. Taneja, L.
Ballan, J. Puwein, G.J. Brostow, M.Pollefeys, "3D
Reconstruction and Video-Based Rendering of
Casually Captured Videos," chapter in Video
Processing and Computational Video, Springer 2011
- Book
- BibTex
|
|
L. Ballan, G. J.
Brostow, J. Puwein, M. Pollefeys, "Unstructured
Video-Based Rendering: Interactive Exploration
of Casually Captured Videos,"
ACM Transactions on Graphics (Proc.
SIGGRAPH) 2010
- Project
Page
- BibTex
- further
efforts
|
|
O. Mac Aodha, G.J. Brostow,
M. Pollefeys, "Segmenting
Video Into Classes of
Algorithm-Suitability," CVPR
2010
- Project
Page
- BibTex
|
|
C. Amati, G.J.
Brostow, "Modeling 2.5D
Plants from Ink Paintings," in
Proc. ACM SIGGRAPH/Eurographics Symposium on SBIM
(Sketch-Based Interfaces and Modeling), p41-48,
2010
- PDF
- BibTex
|
|
G.J. Brostow, J.
Fauqueur, R. Cipolla, "Semantic object
classes in video: A high-definition ground
truth database," Pattern
Recognition Letters 30(2): 88-97, 2009
- PDF
- CamVid
Project Page
- Data
(The
raw
video files are mirrored here)
- BibTex
|
|
G.J. Brostow, J.
Shotton, J. Fauqueur, R. Cipolla, "Segmentation and
Recognition using Structure from Motion Point
Clouds", ECCV (1):
44-57, 2008
- New
project
page with original data (CamVid)
- Original
Project Page (with some broken links)
-
Raw
video
files are mirrored here
- BibTex
|
|
C. Hern?dez, G.
Vogiatzis, G.J. Brostow, B. Stenger, R. Cipolla, "Non-Rigid
Photometric Stereo with Colored Lights,"
ICCV 2007
- Project
Page
- BibTex
|
|
J. Fauqueur, G.
Brostow, R. Cipolla, "Assisted Video
Object Labeling By Joint Tracking of Regions
and Keypoints," In Proc. IEEE ICCV
2007 Interactive Computer Vision Workshop, p1-7,
2007
- PDF
- BibTex
|
|
G. J. Brostow, R.
Cipolla, "Unsupervised
Bayesian Detection of Independent Motion in
Crowds," In Proc. IEEE CVPR, Vol I:
594-601, June 2006.
- PDF
- BibTex
- Project
Page
|
|
M. Johnson, G.J.
Brostow, J. Shotton, O. Arandjelovic, V. Kwatra,
R. Cipolla, "Semantic Photo
Synthesis," Computer
Graphics Forum (Proc. Eurographics) Volume 25, #3,
p407-413, 2006.
- PDF
- BibTex
- Project
Page
|
|
A.R. Shen, G.J.
Brostow, R. Cipolla, "Toward Automatic
Bloodspatter Analysis in Crime Scenes,"
In Proceedings IET Crime and Security Conference (Imaging for Crime
Detection and Prevention), p378-383, 2006.
- PDF
- BibTex
- Project
Page
|
|
M. Wilczkowiak,
G.J. Brostow, B. Tordoff, R. Cipolla, "Hole Filling
Through Photomontage," p492-501 in
BMVC 2005.
- PDF
(1.8Mb)
- BibTex
- Project
Page
|
|
M. Terry, G.J.
Brostow, G. Ou, J. Tyman, D. Gromala, "Making
Space
for Time in Time-Lapse Photography,"
In SIGGRAPH Technical Sketches, 2004.
- PDF
(2.5Mb)
- BibTex
- Project
Page
|
|
G.J.
Brostow,
I. Essa, D. Steedly, V. Kwatra, "Novel
Skeletal Representation For Articulated
Creatures," In Proceedings of ECCV,
May 2004, Vol III: 66-78.
Official
version ?
Springer-Verlag
-
ECCV 2004.pdf
(5Mb)
-
BibTex
G.J.
Brostow,
"Novel Skeletal Representation For
Articulated Creatures," Ph.D. Thesis,
Georgia Institute of Technology, 2004.
-
PDF
(21Mb)
-
BibTex
- Project
Page
|
|
K.E. Sukel, R.
Catrambone, I.A. Essa, G.J. Brostow, "Presenting
Movement
in a Computer-Based Dance Tutor," In
International Journal of Human-Computer
Interaction, 15(3), 2003
- PDF
- BibTex |
|
G.J. Brostow and
I.A. Essa, "Image-Based
Motion
Blur for Stop Motion Animation," In
Computer Graphics, Proceedings of SIGGRAPH 2001.
- Project
Page
- BibTex |
|
I.A. Essa and G.J.
Brostow, "A
Course on Digital Video Special Effects,"
in Proceedings of the IEEE CS Workshop on
Undergraduate Education & Image Computation,
Hilton Head, South Carolina, June, 2000.
- PDF
version
- DVFX
Home Page
- BibTex |
|
J. O' Brien, B.
Bodenheimer, G.J. Brostow, J. Hodgins, "Automatic Joint
Parameter Estimation From Magnetic Motion
Capture Data," Graphics Interface
2000, Montreal, Canada.
- Abstract
- PDF version
- BibTex
|
|
G.J. Brostow and
I.A. Essa, "Motion Based Decompositing of
Video," in Proceedings of ICCV,
International Conference on Computer Vision
1999, Corfu, Greece, September 1999.
- PDF
version
-
Project
Home Page
- BibTex
|
|
Teaching:
Vacancies:
Projects for MSc's:
Internal-only link to view
the slides is here
(note which academic year!). (You must be browsing from
within the UCL domain).
Simple LaTeX template
files for building a UCL Computer Science MSc
Thesis.
|
|