Our inspiration for this project came just days before the event. We were looking over potential hardware to use at the hackathon, when Andy saw accelerometers. Knowing that Frank wanted to do a machine learning project, he thought about using accelerometers to learn someones step and stride patterns.
What it does
Stride is a proof of concept for an application that can detect and alert people when their possessions are taken by others based on their walking patterns, detected by accelerometer hardware in smart devices.
How we built it
We built the application using the Android Studio API and TensorFlow machine learning on python. All of the hardware we used was from our own sources, despite the cool stuff at NWHacks.
Challenges we ran into
One of the biggest challenges was getting JSON files written for machine learning data. Another issue was linking between the machine learning python scripts and the Android Studio scripts.
Accomplishments that we're proud of
We're most proud that our concept of learning stride patterns actually worked, and a personal victory for all of us was getting accelerometer data to output (James), interfacing between the mobile device and the watch (Andy), and setting up the machine learning and having it work (Frank) We're proud that walking patterns can be distinguished by our M-88 neural network model.
What we learned
We learned about interfacing with Android Studios to create Android applications, as well as wirelessly connecting via the ADB bridge, as well as becoming more familiarized with machine learning. We also learned how to collect data to train machine learning.
What's next for Stride
The next step is getting the machine learning integrated into our smart devices, so it will become a viable product and alert people of thievery.