MoLiDAR-LP102: A Mobile LiDAR Depth Image Dataset for Lateral Posture Assessment

MoLiDAR-LP102 is an expert-annotated mobile LiDAR depth-image dataset for lateral posture assessment. The dataset contains 1,020 raw depth frames acquired from 102 anonymized adult participants using a mobile-device LiDAR sensor under a standardized right and left lateral imaging protocol. Each participant has 10 depth frames: five right-lateral and five left-lateral acquisitions. Raw files are provided as CSV files containing x, y, and metric depth values in meters. Each CSV can be reconstructed as a 240 x 320 depth matrix. Invalid or unmeasurable pixels are encoded as 0.0. Labels were assigned at the lateral-profile level by an experienced physiotherapist using a standardized plumb-line posture assessment procedure. The binary labels are healthy posture and postural deviation. The dataset contains no RGB images, facial images, names, contact information, student numbers, or other directly identifying fields. Use of the dataset is subject to the ethics approval, the data-use notice included in the release package, and applicable personal data protection regulations, including KVKK and GDPR.