Salvi Dario, Velardo Carmelo, Brynes Jamieson, Tarassenko Lionel
Annu Int Conf IEEE Eng Med Biol Soc. 2018 Jul;2018:4423-4427. doi: 10.1109/EMBC.2018.8513319.
Step counting from smart-phones allows a wide range of applications related to fitness and health. Estimating steps from phones' accelerometers is challenging because of the multitude of ways a smart-phone can be carried. We focus our work on the windowed peak detection algorithm, which has previously been shown to be accurate and efficient and thus suitable for mobile devices. We explore and optimise further the algorithm and its parameters making use of data collected by three volunteers holding the phone in six different positions. In order to simplify the analysis of the data, we also built a novel device for the detection of the ground truth steps. Over the collected data set, the algorithm reaches 95% average accuracy. We implemented the algorithm for the Android OS and released it as an open source project. A separate dataset was collected with the algorithm running on the smart-phone for further validation. The validation confirms the accuracy of the algorithm in real-time conditions.
智能手机计步功能催生了大量与健身和健康相关的应用。由于智能手机的携带方式多种多样，通过手机加速度计估算步数具有挑战性。我们的工作重点是窗口峰值检测算法，该算法此前已被证明准确高效，因此适用于移动设备。我们利用三名志愿者在六个不同位置手持手机收集的数据，对该算法及其参数进行了进一步探索和优化。为了简化数据分析，我们还构建了一种用于检测实际步数的新型设备。在收集的数据集上，该算法平均准确率达到95%。我们为安卓操作系统实现了该算法，并将其作为开源项目发布。通过在智能手机上运行该算法收集了一个单独的数据集用于进一步验证。验证结果证实了该算法在实时条件下的准确性。
