Computer vision for 3D environment recognitionin augmented reality mobile games

We propose to develop computer vision technologies for recognizing 3D environment for augmented reality (AR) mobile games. Given an image/video taken captured a mobile device (e.g. iPad), we would like to be able to recognize some useful information about the 3D environment of the game player. For example, one of our goals is to recognize whether the player is in an indoor or outdoor environment. This information will allow the game to project appropriate virtual objects and characters that interact with the user. Another goal is to recognize certain important objects (e.g. building, road, sky) in the natural scene. This information will be critical to provide a more realistic and immersive gaming experience for the players. We will evaluate current state-of-the-art computer vision techniques for this application and adapt them for the mobile platform.

Mrigank Rochan
Faculty Supervisor: 
Dr. Yang Wang