Abstract
In this preliminary study, we explore the potential for automatically recognizing activities of daily living by identifying the folding patterns of various laundry items such as shirts, jeans, and towels. Laundry is an important aspect of personal hygiene for the elderly population. The study is performed by creating a model to identify these laundry items with high accuracy using data collected from wearable sensors. Time-domain and frequency domain features are extracted from data sets and are fed to Support Vector classifiers using fine Gaussian kernels in MATLAB. Multiple locations on the body and multiple sets of data at each location provide ample data for training the system. The overall classification rate is 78% for detecting all laundry items. Activities of Daily Living will be further explored to recognize cognitive health states by analyzing the activity pattern changes over an extended period.