A Multi-wavelength Wrist-worn Photoplethysmography Dataset For Heart Rate Monitoring

The Multi-wavelength PPG Heart Rate Estimation (MW-PPG-HR) dataset was developed to provide a comprehensive benchmark for photoplethysmography-based heart rate estimation. It addresses key limitations of existing resources by incorporating a diverse participant cohort, multiple LED wavelengths, and a wide range of activities designed to elicit realistic motion artefacts. Data were collected using a research-grade wrist-worn biosensor with synchronised multi-wavelength PPG and accelerometry, alongside ECG ground truth. The protocol balanced rest, moderate, and vigorous exercise, capturing over 26,000 samples across the full physiological heart rate range. MW-PPG-HR therefore offers a robust, generalisable dataset for evaluating modern deep learning methods.