Garage Cinema Research
UNIVERSITY OF CALIFORNIA AT BERKELEY SCHOOL OF INFORMATION
WHAT'S NEW
About Us
Garage Cinema Research and UC Berkeley partner with Yahoo! Inc. to create Yahoo! Research Berkeley.
Research
Garage Cinema Research is doing cutting edge research in media metadata, context-aware mobile media applications, automated media capture, automatic media editing, and the social uses of personal media.
Events
Publications
People
Studio
The SIMS Digital Media Studio receives new high-performance web servers as part of our UC Discovery Grant for Digital Media.
Resources
Learn about related researchers, conferences, and journals.
Sponsors
Garage Cinema Research works with leading sponsors in industry and academia.
 


Mobile Media Metadata 2
 Introduction

This site is a brief illustrated introduction to the Mobile Media Metadata 2 (MMM2) prototype. Below are screenshots showing both the MMM2 cameraphone and web applications, and sample Nokia 7610 cameraphone photos. The Mobile Media Metadata (MMM) project leverages the spatio-temporal context and social community of media capture to infer the content and the sharing recipients of media captured on cameraphones. Over the past two years, we have deployed and tested MMM1 (context-to-content inferencing on cameraphones to infer media content) and MMM2 (context-to-community inferencing on cameraphones to infer sharing recipients) with 60 users in the fall of 2003 and another 60 users in the fall of 2004.

MMM2 consists of an application that resides on the cameraphone (currently the Nokia 7610) and a web-based application. The MMM2 cameraphone client application uses a modified version of the ContextPhone application developed by Mika Raento from the Helsinki Institute for Information Technology (HIIT) Context project to upload captured media and metadata and to automatically launch the Series 60 Opera for Mobile browser for additional phone-based user interaction. The MMM2 cameraphone client application automatically uploads cameraphone photos along with automatically gathered contextual metadata collected at the time of capture and any manually created metadata, such as a photo caption, to the MMM2 web server. Photos are also easily shared from the MMM2 phone client because it suggests to the phone user a short list of likely share receipients based on the user's and the community's prior sharing history and the contextual metadata from the point of capture such as the time of capture, CellID and GPS location, and Bluetooth-sensed human co-presence. Additional captioning, messaging, albuming, and photoblogging may be done from the MMM2 phone user interface in the Opera for Mobile browser.

Each MMM2 user also has a personal MMM2 website where their photos can be viewed, rotated, captioned, annotated, organized into albums, and most importantly, shared with system-supplied suggestions about likely sharing receipients. Photos may be shared with other MMM2 users, or with anyone with an email or SMS address, either directly from the cameraphone at the time of capture, or from the MMM2 website (accessible from the phone or from a computer). Recipients receive a URL and, in their email, a phhoto thumbnail.

We have seen a 2189% increase in the number of photos uploaded per user per day in MMM2 (1.31) compared to MMM1 (0.06). Our upper third most active MMM2 users upload on average more than 3 photos per day. Our MMM2 "context-to-community" sharing recipients guesser has enabled MMM2 users to maintain a high rate of photos shared to photos uploaded (26%) and on average, for every 100 photos uploaded, 18 instances of sharing occur (some of which have multiple photos per sharing instance). Sharing instances are distributed equally between the MMM2 phone and web applications, but on average twice as many photos are shared per sharing instance on the web as from the phone application. Our qualitative and quantitative studies have shown that MMM2 users are pleased with the share guesser's ability to suggest sharing recipients based on prior sharing history and contextual metadata. The current share guesser guesses the correct intended sharing recipient 60% of the time in the first 5 guesses and 70% of the time in the first 10 guesses. We are working with Prof. John Canny of the UC Berkeley Computer Science department on a next generation share guesser which uses contextual metadata in conjunction with collaborative filtering to provide even better guesses of likely sharing recipients. We believe our MMM research will help solve a fundamental problem in personal media production, sharing, and reuse: the need to have simple and effective applications for users to be able to upload, share, and reuse media captured on mobile devices.

The MMM project is closely associated with the Social Uses of Personal Media project which is co-lead by Prof. Nancy Van House. The Social Uses of Personal Media project studies the higher order purposes or "social uses" of personal photos, and how networked digital images and cameraphones are being used for pre-existing and emerging social uses. The MMM2 project and the Social Uses of Personal Media project are working together to study and refine the MMM2 prototype as well as to develop new methods for integrating social science and design research in the construction of sociotechnological systems for mobile media and metadata creation, sharing, and reuse.

Click Here to Watch The Video

MMM2 System Screenshots


Carrie takes a trip to New York City with some friends, leaving her housemate Matthew in California. She takes a photo in Times Square to send to Matthew. She has been snapping and sending photos of her trip with MMM2 since she got on the plane at the Oakland airport.



She takes the photo, fills in the caption, "Times Square NYC," and clicks "Share." The "Share-a-photo" screen appears on Carrie's phone. Since she often shares the cameraphone photos she takes when she is away on her New York trips with Matthew, Matthew's name appears high on the list. She presses "Share."


Carrie gets a confirmation that the photo was sent to Matthew. She takes a few more photos without sharing them; she figures that she will show them to Matthew when she gets home.


When Carrie returns home she looks through the photos she took on her MMM2 web page, where they have been automatically uploaded from her cameraphone.


Carrie scrolls down to find the rest of her Times Square photos. She could also filter her photos by looking for all photos taken in the same CellID as her Times Square photo.


Carrie cliks on the photo she shared with Matthew. She particularly likes this photo when she sees it in a larger format, and decides to email it to Jeff, one of the friends who was with her in New York.


She finds Jeff's name in the "Suggested Recipients" since Jeff's co-presence with Carrie on their New York trip was detected via Bluetooth sensing.


Carrie writes a short note to her friend and presses "Share". After sharing Carrie sees a confirmation of the share and continues to look through her photos.


Later that afternoon Jeff checks his email and finds that he has a message from Carrie. He reads her message and clicks the thumbnail to see the full size image of the Times Square photo.

MMM Related Publications
Marc Davis, Nancy Van House, Jeff Towle, Simon King, Shane Ahern, Carrie Burgener, Dan Perkel, Megan Finn, Vijay Viswanathan, and Matthew Rothenberg. "MMM2: Mobile Media Metadata for Media Sharing." In: Extended Abstracts of the Conference on Human Factors in Computing Systems (CHI 2005) in Portland, Oregon, ACM Press, 1335-1338, 2005.
PDF

Nancy Van House, Marc Davis, Morgan Ames, Megan Finn, Vijay Viswanathan. "The Uses of Personal Networked Digital Imaging: An Empirical Study of Cameraphone Photos and Sharing." In: Extended Abstracts of the Conference on Human Factors in Computing Systems (CHI 2005) in Portland, Oregon, ACM Press, 1853-1856, 2005.
PDF

Marc Davis, Simon King, Nathan Good, and Risto Sarvas. "From Context to Content: Leveraging Context to Infer Media Metadata." In: Proceedings of the 12th Annual ACM International Conference on Multimedia (MM 2004) Brave New Topics Session on "From Context to Content: Leveraging Contextual Metadata to Infer Multimedia Content" in New York, New York, ACM Press, 188-195, 2004.
PDF

Marc Davis. "Mobile Media Metadata: Metadata Creation System for Mobile Images (Video)." In: Video Proceedings of 12th Annual ACM International Conference on Multimedia (MM 2004) in New York, New York, ACM Press, 2004.
MPEG

Marc Davis. "Mobile Media Metadata: Metadata Creation System for Mobile Images (Video Description)." In: Proceedings of 12th Annual ACM International Conference on Multimedia (MM 2004) in New York, New York, ACM Press, 936-937, 2004.
PDF

Marc Davis and Risto Sarvas. “Mobile Media Metadata for Mobile Imaging.” In: Proceedings of the IEEE International Conference on Multimedia and Expo (ICME 2004) Special Session on "Mobile Imaging" in Taipei, Taiwan, IEEE Computer Society Press, 2004.
PDF

Risto Sarvas, Erick Herrarte, Anita Wilhelm, and Marc Davis. “Metadata Creation System for Mobile Images.” In: Proceedings of the Second International Conference on Mobile Systems, Applications, and Services (MobiSys 2004) in Boston, Massachusetts, ACM Press, 36-48, 2004.
PDF

Anita Wilhelm, Yuri Takhteyev, Risto Sarvas, Nancy Van House, and Marc Davis. "Photo Annotation on a Camera Phone." In: Extended Abstracts of the Conference on Human Factors in Computing Systems (CHI 2004) in Vienna, Austria, ACM Press, 1403-1406, 2004.
PDF

MMM2 Team


Prof. Marc Davis

Prof. Nancy Van House
Shane Ahern
Allison Billings
Carrie Burgener
Katherine Chan
Joshua Chao
Scott Fisher
Nathan Good
Benjamin Hill
Andrew Iskandar
Simon King
Vam Makam
Rahul Nair
Dan Perkel
Madhu Prabaker
Nick Reid
Bruce Rinehart
Matthew Rothenberg
Jeff Towle
Eric Tse
Michael Wooldridge

MMM2 System Requirements

1. Series 60 phone
2. Opera for Mobile Browser installed
3. Phone service must include a data plan
4. Minimum of 4MB of internal memory
5. Bluetooth

Example Nokia 7610 Photos




MMM2 Sponsors
Home | About Us | Research | People | Publications | Events | SIMS Studio | Resources | Sponsors | Contact Us