Benchmarking Annotation Procedures for Multi-channel Time Series HAR Dataset

This contribution evaluates semi-automated annotation for generating high-quality data for multi-channel time series Human Activity Recognition. For this purpose, time series data that consists of Optical Motion Capturing and inertial measurements from on-body devices for industrial deployment is created, annotated, and revised by four individuals. The semi-automated annotation consists of predictions from a temporal convolutional neural-network, and manual revisions for generating high-quality and fine-grained annotations for Human Activity Recognition. This annotation approach reduces the annotation effort keeping similar quality as compared to manual annotation.