Presentation Information
[B-19-14]Preliminary Cognitive Load Estimation Using Heart Rate and Webcam-Derived Posture Features
◎△Hidaka Asai1, Hideaki Kimura1, Momona Kikuzawa1, Miki Matsuura1 (1. Department of Computer Science, Graduate School of Engineering, Chubu University)
Keywords:
Cognitive Load,Heart Rate,Webcam,Posture Features,Mental Arithmetic Task
Objective assessment of cognitive load during learning is important for adaptive learning support. This study preliminarily examined the relationship between heart-rate indices and webcam-derived relative motion features of the head and upper body during a mental arithmetic verification task. Ten male students in their twenties performed the task under three difficulty conditions: easy, normal, and hard. During the task, heart-rate data from a Polar H10, webcam video, task performance, and Raw NASA-TLX scores were synchronously recorded. For analysis, each task block was divided into 30-s windows, and mean heart rate, change in mean heart rate from baseline, the relative angle between the shoulder midpoint and face centroid, and face-centroid movement were calculated. The results showed that higher task difficulty was accompanied by lower accuracy, longer response time, and higher subjective workload. In addition, participants with lower accuracy in the hard condition tended to show larger changes in the relative shoulder-midpoint-to-face-centroid angle and greater face-centroid movement. These findings suggest that webcam-derived relative head–upper-body motion features may serve as candidate indicators for capturing subtle bodily responses associated with changes in cognitive load.