Body landmark point estimation for near-duplicate human motion

This dataset comprises time-series posture estimation data collected from 34 participants performing three represenative culinary tasks: slicing, chopping, and sawing. Each participant’s data is organized into three CSV files, one for each task, capturing detailed positional information of 33 body landmark points using the MediaPipe framework. The data for each landmark includes x, y, and z coordinates, corresponding to the participant's body position in each recorded frame, at a frequency of 30 frames per second. The dataset can be used for studying human motion, for example in occupational settings, where posture analysis is important for understanding ergonomic risks and improving task performance.