Real-Time Human Activity Recognition Using Phone Accelerometer Data
Jan 2026 – Apr 2026
Role: Team Member
Purpose: ELEC 292
As part of a team project, I worked on developing a real-time human activity recognition system using accelerometer data collected from a mobile phone (phyphox app). My work focused on processing and preparing the provided sensor data, exploring the learning curve and training a logistic regression model to recognize different activities based on historical data. I also contributed to the visualization and presentation of the model’s results, creating graphs and outputs to help evaluate its performance and communicate the activity predictions. This project gave me hands-on experience working with some basic machine learning, data processing, model evaluation and data visualization in Python.
App Activity Detection Sample 2
Feature distributions walking vs jumping
Acceleration Data
Raw and Preprocessed Signal Vizualization
Acceleration Data 2
Walking to Jumption Transition Graphs
Learning Curve 2
Learning Curve 1
Standard Deviation Feature Distributions
Mean Feature Distributions
Application Tabs
