Gender Recognition by Monitoring Walking Patterns via Smartwatch Sensors

Activity-based computing aims to monitor the physical state of a user by sending continuous signals through sensors attached to the subject's body. This research field is becoming very important with the increasing number of smart devices. The sensors embedded in these smart devices can be used to monitor user's activities. There has been significant work recognizing human activities. But in this paper, we propose a system to recognize the gender of the subject by monitoring the walking pattern. To capture the walking pattern of the subject, a smart watch was attached to the subject's wrist, while accelerometer and gyroscope signals was recorded at a fixed interval. After preprocessing the raw data, an artificial neural network was used to classify the gender of the subject. We reviewed the data mining technique to capture sensor signals, implemented an artificial neural network for classification purposes, analyzed the major challenges and introduced many real-life applications.