Inspiration We wanted a simple, efficient way to track fitness using just a phone, without needing extra devices or complex setups.
What it does It uses phone sensors to track distance, activity levels (light, moderate, or intense), and extrapolate burned calories.
How we built it We trained a machine learning model in MATLAB using phone sensor data collected using MATLAB mobile (accelerometer, gyroscope, GPS etc.), which finds the activity level of the person exercising. A neural network was trained using a FitBit data dataset to estimate the calories burned from steps travelled and activity level.
Challenges we ran into Limited time to collect data and train models. Didn't have time to train a good "activity level" model.
Accomplishments that we're proud of A tracker that just about works. With more time to train better models it would work very well.
What we learned We got hands-on with sensor data, machine learning, and working under strict time constraints.
Log in or sign up for Devpost to join the conversation.