ACM Multimedia - GeoMM Workshop - Keynote
Download as KEY, PDF0 likes2,050 views
The document discusses the advancements in accessing and sharing location information for media, highlighting the shift from manual to automatic annotation methods. It outlines challenges related to location accuracy and control over geo data while emphasizing the need for innovative methods of encoding place and better user control. The focus for the future includes leveraging existing location data and addressing issues of personal geography.
ACM Multimedia - GeoMM Workshop - Keynote
- 1. STATE Adam Rae Yahoo! Research OF THE GEOTAG @raemond Where are we?
- 3. WHY ARE WE HERE? • Over the last few years we have gained better/easier access to location information for our media • We've also been generating and sharing more media • Manual annotation has been augmented with automatic methods
- 4. Who? What? When? Why? Where?
- 5. Who? What? When? Why? Where?
- 11. • Satellite systems • GPS, GLONASS, Galileo, QZSS (Japan), Beidou (PRC), IRNSS (India) • Assisted GPS, cell triangulation
- 12. Flickr: Steve -
- 13. ADAM'S 8 LUCKY POINTS FOR GEO PROSPERITY Warning! Includes cats
- 14. 1: LOCATION IS A SOLVED PROBLEM
- 19. Flickr: herby_fr penguinbush ♥ Rainbowcatz ♥
- 20. Flickr: herby_fr penguinbush ♥ Rainbowcatz ♥
- 21. Flickr: herby_fr penguinbush ♥ Rainbowcatz ♥
- 22. Flickr: herby_fr penguinbush ♥ Rainbowcatz ♥
- 23. 3: ONE POINT LOCATION IS NOT ENOUGH Flickr: Glamhag
- 24. 4: IMAGE ≠ VIDEO ≠ AUDIO (≠ HOLOGRAM?) Flickr: ROCSDT
- 25. 5: A FEW METRES IS NOT ACCURATE ENOUGH Flickr: Daisuke Kawamura
- 26. Flickr: kejadlen
- 27. 6: THERE'S MORE THAN JUST INFORMATION RETRIEVAL Flickr: Eric Fischer
- 34. 7: PERSONAL/SUBJECTIVE GEOGRAPHY
- 35. 7: PERSONAL/SUBJECTIVE GEOGRAPHY or
- 36. 7: PERSONAL/SUBJECTIVE GEOGRAPHY or “Show me photos taken at Grandma’s House”
- 39. 8: WE MUST TACKLE GEO DATA CONTROL ISSUES Flickr: jonathan mcintosh
- 40. FOCUS FOR THE FUTURE • We need to develop newer, better ways of encoding place • We need to exploit the location data we already have • We need to ensure users are in control
- 41. Thank you Flickr: Walt StoneBurner
Editor's Notes
- #2: My name is Adam Rae, have been working Yahoo Research in Barcelona, in the Geographic Context and Experience Group\n\nToday's talk is going to be an overview of where the multimedia geo community is, at the moment and how its work in our field has lead to fascinating new areas of information retrieval, multimedia processing, social analysis and more, through the advent of greater geographic data associated with our media\nI'm also going to do a little future gazing, to see where I think the field is going, but also where I think the field *should* be going\n\nFuturism is a dangerous activity, and is rarely accurate, otherwise we would all be wearing silver jump suits, and going to work via jet pack\n
- #3: \n
- #4: Why is location important for media?\n- What does it allow us to do?\n(Support browsing, search etc, just as adding tags, time, text etc has already)\nWhen Flickr started allowing encoding, was through machine tags, obscure, geeky, now very mainstream, partly due to invisible location data availability\nWe've been able to add the Where to the What, Who and Why\n\n
- #5: \n
- #6: \n
- #7: Triangulation by heavenly bodies and accurate timing\n
- #8: Brought that update by creating our heavenly bodies and atomic clocks whizzing around the earth\n
- #9: Specifically for media, GPS loggers\n
- #10: Of course, increasing number of cameras have gps built in.\nBiggest growth area - camera phones\n\n
- #11: \n
- #12: Past, location derived via...\nNow GPS, GLONASS, Galileo, JP, ZH\nAssisted GPS, triangulation\nAlso, still manual, not always possible to get automatic location\n\nGive us a single point\n\n
- #13: Also, still manual, not always possible to get automatic location\n\nAll give a single point\n
- #14: Combination of weaknesses I’ve identified in field, plus areas I think we should spend more time thinking about\n
- #15: \n
- #16: \n
- #17: \n
- #18: \n
- #19: \n
- #20: \n
- #21: Currently, the best we can hope for is a single location, expressed as a single point, to describe the location context of an image\n\nWhich location? That of capture point? ('Easy' to define, single point, unambiguous), that of content? (More difficult, could be multiple, may not be single points, subjective, but, more valuable for the user?\n\nSingle point doesn't describe 'place'\nThink of:\n- extent\n- multiple points\n\nLocation of media content vs capture\n\n
- #22: What does location mean for video? Multiple shots, movement!\n\n
- #23: Notion of accuracy\n\nCurrently around a few metres, usually far less\nWhat happens when we want to assign location within the home? (Cat with camera)\n\n\n(Find indoor piping papers from WhereCamp?)\n\nMore use cases where local location information may be needed...\n\n\n\n
- #24: \n
- #25: Support media exploration/browsing through geo aspect\n
- #26: \n
- #27: Red dots are locations of Flickr pictures. Blue dots are locations of Twitter tweets. White dots are locations that have been posted to both.\n
- #28: Microsoft PhotoSynth\n
- #29: Microsoft PhotoSynth\n
- #30: Microsoft PhotoSynth\n
- #31: Depending on where you are in the in the world, you see it differently\nDepending on who you are, you see it differently\n
- #32: Depending on where you are in the in the world, you see it differently\nDepending on who you are, you see it differently\n
- #33: \n
- #34: \n
- #35: \n
- #36: Subjective meaning of place vs objective location point\nHow can we best support users in both aspects, how do we marry them together?\n\nPersonal geography, as described by a user's media\n\n Representation of place - coords, or 'home'\n\nPersonal interpretation of place, how can we support this subjective representation of place along side an objective one\n\n
- #37: Location data can be sensitive, only beginning to learn what can be inferred\nUsers must have control\nGeo-fencing\nSelective access to geo context\nDecrease accuracy in sensitive areas? How do we handle that?\n
- #38: By addressing these key areas, we’ll be able to not only support users in more effective media handling, but also learn more about our users and their places in the world.\n\n
- #39: \n
- #40: \n
- #41: \n
- #42: \n
- #43: \n
- #44: \n