Pai ain n th the e City ty Col olorfull orfully: y: Lo Locatio ation n Visualizat ualization ion fr from om Mu Multi tiple ple Th Them emes es Quan Fang 1,2 , Jitao Sang 1,2 , Changsheng Xu 1,2 , Ke Lu 1 1 Chinese Academy of Sciences 2 Chinese-Singapore Institute of Digital Media January 07, 2013
Moti tivat vation ion Singapore Where should I go?, and what can I enjoy …
Proble oblem m Exploit user-generated content to organize photos geographically and semantically Location visualization from multiple themes
Approach proach Geo-Tagged Flickr Data User metadata Photos POI Identification POI Themes Discovery Location Theme Aggregation • Singapore as the running example. • 110,846 photos, 26,623 geo-tagged Location Visualization photos, from 9,044 users in • Photo and associated text metadata.
POI OI Detection tection User factor 𝑂 𝑣𝑡𝑓𝑠 𝐺 𝑚 𝑙 = log(𝑂𝑞ℎ𝑝𝑢𝑝 ( 郆 𝑣𝑗) + 1) 6 Jalan Kayu, Central Singapore 𝑗=1 Photo number 5 Singapore Changi Airport 4 Jurong Town, South West 2 Central Catchment Reserve Central Catchment Reserve 3 1 Sentosa, South West Downtwon, Central Singapore Sentosa, South West Downtown, Central Singapore
POI OI Th Theme me Discovery scovery Challenges: visual variance & tag noise; Solution: incremental learning-based method. POI #1: Downtown, Central Singapore salient tag set exemplary photos view light marinabay Theme 1 skyline sands Photo Theme Salient Tag Extraction Assignment extract the salient tags that sky sunset clouds reservoir theme photo and tag set well describe this theme landscape nofollow {ℐ, 𝒰} merlion,skyline,CBD Theme 2 ,flyer,esplanade, merlion symbol stb Theme 3 statue tourism logo lion … … Theme 4
POI OI Th Theme me Discovery scovery Downtown, Central Singapore Central Catchment Reserve Sentosa, South West, Singapore view light marinabay skyline sands reservoir sunset peirce hdr macritchie clouds sunset park beach clouds sky rocks merlion Theme 1 Theme 1 Theme 1 zoo white tiger animal monastery gene sky sunset clouds reservoir landscape nofollow food braise meal cuisine restaurant lunch Theme 2 Theme 2 Theme 2 flower flowers garden beautiful art blue red megan kavadi lady goddess kali girls merlion symbol stb statue tourism logo lion Theme 3 Theme 3 Theme 3
Location cation Th Theme eme Aggregatio regation Vector Space Model using the salient tags Theme 1 water reservoir park sunset river peirce dreiseitl atelier Theme 1 bird park birds jurongbirdpark birdpark parrot Theme 2 gardens botanical botanic flowers flower Theme 2 flower chinesegarden garden plant I Theme 1 merlion symbol stb statue tourism logo lion Theme 2 view light marinabay skyline sands Theme 1 sunset park beach clouds sky rocks Theme 2 flower flowers garden beautiful art blue red
Location cation Th Theme eme Aggregatio regation Vector Space Model using the salient tags gardens botanical botanic flowers flower Location Theme #1 flower flowers garden beautiful art blue red chinesegarden Jurong Town Central Catchment Reserve Sentosa, south west
Singapore ngapore Visualization sualization
Th Thank ank you ou!
Recommend
More recommend
