Reading Circle 2010-03/08

Date: 03/08-2010
Participants: Azadeh, Lian, Liting, Miao
Today's Tasks: Disscus the second paper, "Learning Spatial Event Models from Multiple-Camera Perspectives in an Intelligent Room"

Azadeh:

Lian:
This article talks about creating an Intelligent Environments (IE) system, which can track people in an intelligent room, whit mupit-cameras. The system combines the video streams of these cameras to creat a synchronous system which knows the invents occours in the room.
But this article is mainly focus on the multi-camera part. The cameras work in different levels and states, some ones work in the low-level to process the detailed areas such as face, while the global cameras work to detect if some invents occour in somewhere. But in our projects, our original idea is to use two independent cameras to detect the game projection and user`s position. But now we just use one camera, not muti-camera, so we don`t have to consider communication between cameras. Only we have to consider is how to avoid colour confusion and shadow.

Liting:
From this article, it stated that multiple-camera is commonly used for teleconferencing, which it needs to know who, gesture, motion, facial expression. Using multiple-camera encourages interacting naturally, which people do not need to worry about which direction they are facing or whether they stay in the field of the view. The example is illustrated from the article that is using the multiple-camera in the human’s normal life. The cameras are placed into the environment, such as on the shelf, on the TV, on the table, etc. People might not be aware the cameras. However, for our project, we are going to provide a dark room to play the game, which is easy to be aware that the cameras existence in the environment. From human’s cognition, people may be cognitive that he/she has to stand in the front of the camera, or for shy person that he/she may avoid to be captured by the camera. The author states that using multiple-camera provides much richer descriptions of human activities, which supports natural interactions. For our concept we only need to capture if the player is standing on or squatting down, and due the environment limitation, we have to consider about the color conflict, and shadow problem.

Miao:
This article describes an Intelligent Environments (IEs) that the system and environment can interact with people better. In this IEs, there are 6 cameras can detect the human’s actions and depends on user’s action to run some functions automatically. It is also talk about how to let different systems to share the same camera and how can those systems cooperation with each others.