Our research focuses on analysing human activities according to a knownbehaviorist scenario, in case of noisy and high dimensional collected data. Thedata come from the monitoring of patients with dementia diseases by wearablecameras. We define a structural model of video recordings based on a HiddenMarkov Model. New spatio-temporal features, color features and localizationfeatures are proposed as observations. First results in recognition ofactivities are promising.
展开▼
