Automatic Lifelog media annotation based on heterogeneous sensor fusion

Personal Lifelog media system involves capturing of great amount of personal experiences in the form of digital multimedia during an entire lifespan. However, the usefulness of those data is limited by lack of adequate methods for accessing and indexing such a large database. It is important to manage those data systematically so that user can efficiently retrieve useful experiences whenever they need. In this paper, we focus on presenting how to create metadata, which is the core of the systematical approach, by the fusion of sensor data from a set of heterogeneous sensors. With this metadata, we can support users to find their life history efficiently in our system.