Cognitive assessment in children through motion capture and computer vision
Saif Sayed, Konstantinos Tsiakas, Morris David Bell, Vassilis Athitsos, Fillia S. Makedon · 2019
This paper focuses on creating video-based human activity recognition methods towards an automated cognitive assessment system for children. We present the Activate Test for Embodied Cognition (ATEC), which assesses executive functioning in children through physical/cognitive tasks. Detecting activities for children is challenging due to high amount of random motion and variability. This paper focuses on creating a ubiquitous and non-intrusive activity recognition system for upper-body movements. Our proposed methods are evaluated on real-world data from children performing the Cross-your-Body task. The dataset includes 15 children performing 8 types of activities, resulting to 1900 annotated video samples.