This dataset comprises multisensory signals obtained by a smartwatch during 10 different physical activities (drinking, eating, going up and down stairs, dancing, walking, jogging, running, sweeping the floor, sit-to-stand test, and drying up), classified by METs and organized into two experimental phases. Participants perform the tasks while smartwatch sensors capture movement and physiology data. Additionally, sociodemographic and physical fitness information is collected via questionnaire. The data aim to support research in physical activity classification, motor pattern analysis, and energy expenditure estimation. Due to privacy and ethical considerations, only a portion of the dataset may be publicly available. Researchers interested in accessing the full dataset must contact the authors and submit a formal request, which will be reviewed on a case-by-case basis to ensure compliance with privacy regulations and ethical standards. The data management plan is made available here.