Download presentation
Presentation is loading. Please wait.
1
Computer Vision and the Web Images on the Web Movies, news, sports 916,271 titles 10 mil. videos, 65,000 added daily 350 mil. photos, 1 mil. added daily 1.6 bil. images indexed as of summer 2005 Satellite imageryCity streets
2
Exploring online photo collections: Flickr.com Flickr.com Flickrvision Flickr tag browser Flickr time graphs
3
Artistic Visualization Jason Salavon –100 Special Moments100 Special Moments –Homes for SaleHomes for Sale –Every Playboy centerfoldEvery Playboy centerfold
4
Problems with image access on the Web To find anything on the Web, you must search with words! But what if the question is a picture? Small sand-colored African mammal? Small furry animal with pointy snout and long tail? Cute gopher-like beastie that likes to stand up and hang out in groups? Beige-colored animal with dark eyespots, ears, and tip of tail?
5
Short-term solution: Human computation Identifying bugs and birdsbugsbirds Labeling images: ESP Game, PeekaboomESP GamePeekaboom
6
Where does computer vision come in?
7
Visual search Linking physical objects to the web: Microsoft Lincoln projectLincoln Product search: Like.comLike.com Searching personal photo albums: Riya.comRiya.com
8
Visual sentiment analysis Attractiveness: Hot or Not (research project)Hot or Notresearch project Cuteness: Cute overloadCute overload Interestingness: FlickrFlickr
9
Exploring photo collections in 3D Photo tourism Snavely, Seitz and Szeliski (SIGGRAPH 2006)Photo tourism Automatic photo pop-up Hoiem and Efros (SIGGRAPH 2005)Automatic photo pop-up
10
Practical details Tuesdays at 3:30, room SN 115 Basic format: weekly discussion group Paper list: http://www.cs.unc.edu/~lazebnik/fall07 http://www.cs.unc.edu/~lazebnik/fall07 Variable credit: talk to me about doing extra work!
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.