Wearable devices typically rely on reflective photoplethysmography (PPG) to track heart rate, using light to measure blood flow under the skin. This approach has become the default sensing technique in consumer wearables, but its accuracy can be substantially impacted by real-world conditions such as motion, skin tone, or poor sensor contact.
The newly released WildPPG dataset addresses this gap by offering long, continuous PPG recordings collected in naturalistic settings. Most existing datasets are short or collected in controlled conditions, limiting their use for building robust heart-rate algorithms. WildPPG instead provides sustained raw signals that better reflect the challenges of everyday wear.
By making this resource available, the authors aim to support the development of more reliable heart-rate estimation methods. Researchers can use the dataset to test how well algorithms handle the noise and variability inherent in real-world wearable sensing, potentially leading to improvements in fitness tracking, health monitoring, and clinical research that depends on accurate PPG-based heart rate.