Human activity recognition using tag-based localization

This paper provides a comparative study on the different techniques of classifying human activities using a tag-based radio-frequency (RF) localization system. Non-uniformly-sampled data containing position measurements of the tags on the body is first converted to a uniformly-sampled one using different curve-fitting algorithms. Then, the data is partitioned into segments. Finally, various classification techniques are applied to classify human activities. Curve-fitting, segmentation, and classification methods are compared using different cross-validation techniques and the combination resulting in the best performance is presented. The results indicate that the system demonstrates acceptable performance despite the fact that tag-based RF localization is not very accurate.