About

Computer vision started with the goal of building machines that can see like humans and perform perception for robots, but it has become much broader than that. Applications such as image database search in the world wide web, computational photography, biological imaging, vision for graphics, GIS, biometrics, vision for nanotechnology, were unanticipated and other applications keep arising as computer vision technology develops. Areas such as document analysis and medical image analysis have developed rapidly and have their own conferences. As our computers achieve even a crude understanding of video imagery, computer vision will profoundly change our lives as visual sensors becomes increasingly ubiquitous and enable us to transcend current human limitations. Rapid developments in supportive technologies -- such as digital cameras and computers -- ensure that computer vision systems will become increasingly more capable and affordable. Moreover, the field of robotics itself has enormous potential to revolutionize manufacturing, to provide service by assistive robots, to perform medical surgery -- applications which all require perceptual input from computer vision systems. In addition, there are many applications to defense, homeland security, and the intelligence community.

We propose a workshop to address these issues and to explore the frontiers of computer vision. The goals of the workshop is to (1) to identify the future impact of computer vision on the economic, social, and security needs of the nation; (2) to outline the scientific and technological challenges to address; and (3) to draft a roadmap to address those challenges and realize the benefits.

More details are given in the current white paper.

The final report from the 1991 NSF workshop on computer vision -- "Challenges in Computer Vision Research; Future Directions of Research" -- is also available: 1991 final report, 1991 appendices.

A forum is now open for discussing 10 proposed key objectives and for proposing others.

Frontiers in Computer Vision Workshop

August, 21-24, 2011

Computer Science and Artificial Intelligence Laboratory
Stata Center
Patil/Kiva Room, 32-G449
Massachusetts Institute of Technology
Cambridge, MA 02139

This meeting brings together experts in computer vision and related disciplines from academia and industry. The goal of the workshop is for the community to develop and promote a unified agenda for computer vision research and development between US agencies, universities, and industries (while recognizing that research thrives in a flexible environment). We seek to address issues such as what are the open computer vision tasks, what are the technical and scientific barriers we must overcome in order to solve these tasks, and what strategies – scientific, organizational, funding – are most likely to lead to the greatest progress in addressing these challenges.


The schedule below in PDF.


Click on the blue names to download slides.

DAY	TIME	TYPE	TOPIC	ORGANIZER	PARTICIPANTS
SUNDAY 8/21	12:30 - 12:50		Coffee Prologue
	12:50 - 1:00	Opening Remarks	Introduction to the Workshop	Yuille / Oliva
	1:00 - 3:15	Talks	Relationship of Computer Vision to Studies of Biological Vision	Rosenholtz	Cottrell, Dicarlo, Gallant, Poggio, Simoncelli, Weiss
	3:15 - 3:45		Coffee Break
	3:45 - 5:30	Talks	Historical Perspective on Computer Vision	Chellappa	Adelson, Ahuja, Lowe, Malik, Seitz, Zhu
MONDAY 8/22	Beginning at 8:30am		POSTER SET-UP
	8:15 - 8:45		Breakfast - In Poster area
	8:45 - 10:15	Talks	Taxonomy of Computer Vision: Hilbert Problems for Vision	Malik	Adelson, Chellappa, Perona, Soatto, Zisserman
	10:15 - 10:45		Coffee Break
	10:45 - 12:00	Panel/Talk	Humans and Machines Collaborating on Vision Tasks	Belongie	T Berg, Geman, Grauman, Perona
	12:00 - 1:15		Lunch
	1:15 - 2:30	Panel/Talk	Applications of Machine Vision to Science	Perona	Belongie, Fergus, Golland, Poggio
	2:30 - 3:45	Talks	Cross-fertilization with other Disciplines	Freeman	Durand, Hebert, Hertzmann, Metaxas
	3:45 - 4:15		Coffee Break
	4:15 - 6:00	Panel	Foundations and Core	Yuille	Szeliski, Sawhney, Forsyth, Liu, Soatto, Sapiro, Zisserman
	6pm onwards		Group Beer at Cambridge Brewing Company
TUESDAY 8/23	8:00 - 8:30		Breakfast - In Poster area
	8:30 - 10:30	Talks	Visual Representation	Zhu	A Berg, Corso, Darrell, Felzenswalb, Hoiem, Kimia, Learned-Miller, Tenenbaum, Todorovic Posters: Corso, Learned-Miller
	10:30 - 11:00		Coffee Break
	11:00 - 12:30	Talks	Image, Video, and Scene Understanding	Oliva	Efros, Hebert, Lazebnik, Savarese, Schmid, Yuille Posters: Hays, Grauman, Lazebnik, Savarese
	12:30 - 1:45		Lunch
	1:45 - 3:00	Talk/Panel	The Promise and Perils of Benchmark Datasets	Zisserman (1, 2)	Efros, Fei-Fei, Forsyth, Torralba
	3:00 - 3:30		Coffee Break
	3:30 - 5:30	Talks	The Role of Learning in Vision	Fergus	Learned-Miller, LeCun, Ng, Ramanan, Sudderth, Yu, Yuille Posters: Q. Ji, M-H Yang
	5:30 - 6:30	Reception	BCS - Building 46 - Atrium
	6:30 - 7:30	Panel	Reviewing and Evaluating Computer Vision	Rangarajan	Belongie, Ji
WEDNESDAY 8/24	8:15 - 8:45		Breakfast
	8:45 - 10:30	Panel	Relations between Academia and Industry	Lowe	Adelson, Martin, Ramesh, Sharon, Szeliski, Yan
	10:30 - 11:00		Coffee Break
	11:00 - 12:00	Panel/Talks	Program Officers Session	Yuille	Dai, Donlon, Kamgar-Parsi, Geertsen, Yang, TBD
	12:00 - 12:30	Panel	Wrap-Up Session	Yuille	Hebert, Perona, Szeliski
	12:00 - 1:00		Lunch (on site)

Poster Presentations

Board Size: 4' tall x 8' wide

#	PRESENTER	PROJECT AUTHORS	TOPIC/TITLE	SESSION
1	Jason Corso	J. J. Corso, J. A. Delmerico, P. David, R. Alomari, V. Chaudhary	Layered Models for Bridging from Low to High Level Vision	Visual Representation
2	Deva Ramanan	Y. Yang & D. Ramanan	Articulated pose estimation using flexible mixtures of parts	TBD
3	Kristin Grauman	K. Grauman & D. Parikh	Relative Attributes	Image/Scene understanding
4	James Hays	J. Hays & G. Patterson	SUN Attributes: A Large-Scale Database of Scene Attributes	Image/Scene understanding
5	Eitan Sharon	E. Borenstein, A. Brandt, P. Srinivasan, S. Tran, M. Tek, A. Moshe, E. Sharon	VideoSurf's video recognition technology in the connected devices race	TBD
6	Qiang Ji	Q. Ji	Knowledge augmented visual learning	The Role of Learning in Vision
7	Ming-Hsuan Yang	Q. Wang & M.-H. Yang	Learning to track objects	The Role of Learning in Vision
8	Aaron Hertzmann	M. de Lasa & I. Mordatch	Full-Body Locomotion Control By Low-Dimensional Planning	TBD
9	Sinisa Todorovic	S. Todorovic	Bridging the gap between pixels and compositional activities for video parsing	TBD
10	Svetlana Lazebnik	J. Tighe & S. Lazebnik	Understanding Scenes on Many Levels	Image/Scene understanding
11	Aude Oliva	P. Isola, J. Xiao, D. Parikh, A. Torralba, A. Oliva	High-level attributes of images: how memorable is an image?	Image/Scene understanding
12	Edward Adelson	K. Johnson & E. Adelson	Gelsight	Academia and Industry
13	Silvio Savarese	S. Y. Bao, S. Savarese	Semantic Structure from Motion	Image/Scene understanding
14	Erik Learned-Miller	E. Learned-Miller, L. Sevilla Lara, M. Narayana, E. Shelhamer, B. Mears	Distribution Fields: A Unifying Representation for Low Level Vision	Visual Representation
15	Barbara Hidalgo-Sotelo	B. Hidalgo-Sotelo, T. Judd, K. Ehinger, F. Durand, A. Torralba, A. Oliva	What are you looking at? Predictability and Patterns in Human Eye Movements	Image/Scene understanding
16	Alexei A. Efros	T. Malisiewicz, A. Gupta, A. A. Efros	Ensemble of Exemplar-SVMs for Object Detection and Beyond	Image/Scene understanding
17	Anand Rangarajan	A. Rangarajan	Distance transforms, wave functions and shape analysis	TBD

The McGovern Institute for Brain Research at MIT is gratefully acknowledged for their support.

Workshop participants are lodging primarily at these hotels:

The Kendall Hotel
350 Main St
Cambridge MA 02139
For google map directions To/From the Kendall to the workshop click here.

Boston Marriott Cambridge Hotel
2 Cambridge Center
50 Broadway
Cambridge MA 02142
For google map directions To/From the Marriott to the workshop click here.




Directions to Patil/Kiva Seminar Room, 32-G449

The room is on the 4th floor of the Gates Tower in the Stata Center (Building 32).

If you enter Building 32 from the entrance on Vassar St, proceed straight ahead and elevators are on the right.
If you enter Building 32 from the entrance near the cafeteria, walk down the hallway (away from the cafeteria) toward Vassar street, and elevators are on your left.
Take the elevators to the 4th floor; exit to the left and then turn right at the end of the elevator bank. At the end of the short corridor bear to the left and continue around the R&D Dining Room. Patil/Kiva Seminar Room will be straight ahead.

Please allow extra time for the initial trip to the seminar room, as the Stata Center is a unique building and can be challenging to navigate for even the most adept individuals. Look out for fliers for Frontiers in Computer Vision.

Street address of the building:
32 Vassar St
Cambridge MA 02139

For navigating around MIT campus in general, Whereis.mit.edu is a handy resource.


Directions to MIT from Logan Airport:

by subway - From any terminal at Logan Airport, take the Silver Line bus to South Station. At South Station, change to the Red Line subway to Kendall/MIT (inbound toward Alewife). Under normal conditions the ride will take about one-half hour and the fare is $1.70-$2.00. Purchase a “Charlie Card” from the outside kiosks at the airport terminal before getting on the bus.

by taxi - Taxi fare from the airport is about $35–$40. During non-rush hour, the taxi ride will take about 15 minutes. During rush hour, the ride could take 30 minutes or more.

Taxi cabs stations are conveniently located outside of the Kendall T station (Main St) and outside of the Marriott hotel (Broadway). Additionally, here are a few local taxi companies:

Cambridge Taxi Company 617-686-9690
Cab of Cambridge 617-621-2500
Ambassador Brattle Cab 617-492-1100


Food Hints

Want coffee, quick lunch places, or a dinner idea? Here are some nearby options for coffee, food, drink.


Internet access

Visitors need to make sure the wireless card is on and enabled. The machine needs to be configured for DHCP (obtaining an IP address automatically). If the machine is running firewall software, it will need to be disabled until the registration process is complete.

Once their equipment is ready, visitors should open a web browser and point it to any web page.

After selecting Visitor registration, the returned page will display the MITnet Rules of Use, followed by a registration screen, requesting the visitor's contact information, number of days of connectivity, and the event for which they are on campus. Visitors can register between one and five (consecutive) days at a time, up to fourteen days per year. The network connection takes about ten minutes to activate, and remains active for the number of days selected.


For any questions, please contact bhs@mit.edu.

To set the stage for the workshop and start the discussion, ten key objectives for computer vision are proposed. To discuss these objectives, sign in. You can also add a new objective in the "Other discussions" section.

To submit a white paper, sign in and add a discussion to the "White papers submissions" section below.