Exploiting User Generated Content for Mountain Peak Detection
Download as PPT, PDF1 like1,938 views
The document discusses a method for utilizing user-generated content to identify mountain peaks and build environmental models through passive human computation. It outlines a matching algorithm comprising steps such as scaling, rendering, edge extraction, filtering, and matching, achieving a maximum accuracy of 62% and 81% accuracy in the top-10 matches. Potential applications include augmented reality and creating environmental models.
Exploiting User Generated Content for Mountain Peak Detection
- 1. 1/05/2013 SoHuman 2013 at ACM Web Science 2013 1 Exploiting User Generated Content for Mountain Peak Detection Roman Fedrov, Davide Martinenghi, Marco Tagliasacchi, Andrea Castelletti
- 2. 2 Passive Human Computation User Effort SoHuman 2013 at ACM Web Science 20131/05/2013
- 3. 3 Passive Human Computation Collective Intelligence SoHuman 2013 at ACM Web Science 20131/05/2013 Our Goal Retrieve the collections of the mountain appearances in different time instants and build environmental models. 201320122011201020092008 Key Problem Identify the mountains
- 4. 4 Key Problem: Object Identification Input SoHuman 2013 at ACM Web Science 20131/05/2013 46° 0’ 48.51” N 7° 48’ 6.62” E http://www.udeuschle.de/Panoramen.html
- 5. 5 Key Problem: Object Identification Output SoHuman 2013 at ACM Web Science 20131/05/2013
- 6. 6 Matching Algorithm SoHuman 2013 at ACM Web Science 20131/05/2013 Steps Scale Render Extract Edges Filter Match
- 7. 7 Matching Algorithm SoHuman 2013 at ACM Web Science 20131/05/2013 Steps Scale Render Extract Edges Filter Match
- 8. 8 Matching Algorithm SoHuman 2013 at ACM Web Science 20131/05/2013 Steps Scale Render Extract Edges Filter Match
- 9. 9 Matching Algorithm SoHuman 2013 at ACM Web Science 20131/05/2013 Steps Scale Render Extract Edges Filter Match
- 10. 10 Matching Algorithm SoHuman 2013 at ACM Web Science 20131/05/2013 Steps Scale Render Extract Edges Filter Match Adapted from matching technique by Baboud et al. “Automatic photo-to-terrain alignment for the annotation of mountain pictures”, CVPR 2011
- 11. 11 Results SoHuman 2013 at ACM Web Science 20131/05/2013 62% of correct match as maximum result 81% of correct match in top-10 positions
- 12. 12 Potential Applications SoHuman 2013 at ACM Web Science 20131/05/2013 http://www.nohrsc.noaa.gov/ Augmented Reality Environmental Models Colle dei BreuilFurgggrat