MultiPosture: A Dataset of body joints keypoints extracted using MediaPipe for multi-task sitting posture recognition with upper and lower body labels

This dataset contains skeletal pose data extracted from video recordings of 13 participants performing various sitting postures in home environments. The data was processed using MediaPipe Pose Heavy model and includes 4,800 frames of 3D skeletal coordinates (x, y, z) for 11 key body joints, with each frame manually labeled for both upper and lower body posture classifications.

The data is stored in CSV format with normalized coordinates relative to hip center, containing 33 input dimensions (11 joints × 3 coordinates) representing key skeletal points. To protect participant privacy, only the processed skeletal coordinates are included, with no raw video or image data due to privacy constraints.

Upper Body Labels:

TUP: Upright trunk position

TLB: Trunk leaning backward

TLF: Trunk leaning forward

TLR: Trunk leaning right

TLL: Trunk leaning left

Lower Body Labels:

LAP: Legs apart

LWA: Legs wide apart

LCS: Legs closed

LCR: Legs crossed right over left

LCL: Legs crossed left over right

LLR: Legs lateral right

LLL: Legs lateral left

Each frame in the dataset has been manually labeled and validated by experts, making it particularly suitable for developing and evaluating machine learning models for ergonomic monitoring systems, ambient assisted living applications, and general posture recognition research.

This dataset was collected as part of the study:

D. Carneros-Prado, L. Cabañero-Gómez, E. Johnson, I. González, J. Fontecha and R. Hervás, "A Comparison Between Multilayer Perceptrons and Kolmogorov-Arnold Networks for Multi-Task Classification in Sitting Posture Recognition," in IEEE Access, vol. 12, pp. 180198-180209, 2024, doi: 10.1109/ACCESS.2024.3510034. link