TOWARDS 3D CROWDSOURCING: EASING WEB3D NAVIGATION USING USER TRACES

摘要

This paper is the first step towards easing Web3D user navigation using crowdsourcing. We collect user interactions to determine informative regions (the most popular regions) within a 3D object. We show that subsequent users can benefit from previous user navigation to improve their Web3D experience: a set of 3D poses is computed and used for recommendations to subsequent users. The recommended poses aim at simplifying the 3D navigation, that is, reducing the number of pan, zoom and rotate events the user may need to reach his/her desired regions. An experimental user study shows that our system redirects user to interesting areas quicker, with fewer interactions required.