Posture Pattern Recognition Analysis in Lectures
Weverson da Silva Pereira, Fernando Pujaico Rivera, Leila Cristina Carneiro Bergamasco, Paulo Sergio Rodrigues · 2023
The Study of Posture Analysis and Non-Verbal Communication plays a pivotal role in enhancing communication among individuals in various contexts. The ability to decode and comprehend messages conveyed through gestures, facial expressions, and body movements is crucial for fostering more effective and meaningful interactions. Accordingly, this present work aims to conduct an exploratory analysis of posture patterns among speakers worldwide. To achieve this, the Openpifpaf algorithm was employed in videos of lectures for pose extraction, and the K-means clustering algorithm was utilized to distinguish commonly adopted postures during this lectures. The evaluation regarding the representativeness of keyposes involved an online questionnaire in which participants were asked to classify certain speaker poses into one of the clusters. The results revealed that the K-means algorithm achieved an accuracy rate of 85.71%.