Creating Digital Life Stories through Activity Recognition with Image Filtering

摘要

This paper presents two algorithms that enables the MemoryLane system to support persons with mild dementia through creation of digital life stories. The MemoryLane system consists of a Logging Kit that captures context and image data, and a Review Client that recognizes activities and enables review of the captured data. The image filtering algorithm is based on image characteristics such as brightness, blurriness and similarity, and is a central component of the Logging Kit. The activity recognition algorithm is based on the captured contextual data together with concepts of persons and places. The initial results indicate that the MemoryLane system is technically feasible and that activity-based creation of digital life stories for persons with mild dementia is possible.