University of Southern California

Institute for Creative Technologies



 Perceptual Interfaces for Immersive Activities







Project Description

The simulation of human activity and its rendering has to take into account their interactions with the objects or people in the scene in order to create a realistic virtual experience. Nowadays, most of the simulation software available track the position of the head in order to infer the person's gaze and accurately the part of the scene that the participant is looking at. These approaches were focusing on simulating scenarios for a single person evolving in a scene. The main use of such simulations is training, however very little research work was performed in training a small group of persons. In this case, individual actions have to be characterized and understood by the system in order to integrate the feedback from each person. Human activity in the broad sense is very complex to analyze, we will focus on human gestures in order to understand the activity of the person and provide a feedback to the simulator. Gestures are commonly used to communicate, point to specific locations or attract other participants' attention. The use of these gestures has a large impact on simulations since it provides a natural feedback from the person to the simulation engine.
We propose to automatically extract and understand these human motion and gestures from a set of cameras observing the person in the simulation room. This approach does not require the person to wear any additional sensors and therefore create a more realistic simulation of real scenarios. This gesture based communications will improve the training of single persons or group of persons where the feedback from each person will be taken into account for evaluating the performances of the user or a group of persons in performing a specific task.


Research Topics