The meet easy project is looking to democratise video conferencing solutions by providing as system at low cost that is easy to use, with high quality and level of flexibility.
Results obtained earlier in the project; machine learning to automatically count the number of people in the room, automatic framing of video, microphone array systems - has been exploited to develop more advanced services such as tracking people using a combination of audio and video, as well as automated whiteboard summarization by understanding when the user has added text or images to the board.
The project includes conferencing cameras and whiteboard cameras. By connecting the microphone array and camera data the project demonstrated an improvement in speaker following. The project has also finished experiments on automatic room geometry understanding and a light-weight framework for determining actual meeting participants. These functionalities are intended to be used for automatic tuning of parameters to improve the perceived quality of sound and video.
The project has also financed a post-doc at NTNU. This research has resulted in papers on automatic quality improvement of video conferencing. The work spans from metrics used for deciding which improvements lead to perceived increase in meeting quality to methods for automatic improvements of video quality. In preparation are works on content-aware rescaling of images and supervised image distortion correction.
Traditional videoconferencing solutions are often expensive, cumbersome to install, and hard to use. "Meet Easy" solves these problems by creating a camera that is easy to use, cost efficient, flexible and all this while maintaining high quality.