This is my school project from 2015 on using accelerometer data for human activity recognition task.
School Project done at CIC, University of Delhi. I will add all the notes/pages/etc I have soon before archiving it. Note that this paper was never submitted anywhere, but I plan to put it on arxiv.
In the repo, there is:
- MPU–9250 fw, in few flavours, eg. SD card data logging.
- Schematics of the arduino data logger
- Python code for training a SVM classifier
- Some pre-built models to classify the accelerometer data
- Raw dataset i created with motion data
- Project report Submitted in school
- Pretty graphs and illustrations
- An iphone app (in xcode!) for streaming accelerometer data over wifi in udp packets
- A python server which takes the streaming data and shows classified predictions in console
- A data logger script to put the streaming data into InfluxDB time series
- Raw dataset obtained from two universities And some 3D point cloud visualisation
Based heavily on MPU-9250 example code by kriswiner/MPU-9250.
- Code License: MIT
- Dataset License: CC0, Public Domain
- Illustrations License: CC-BY-SA-NC
- Report License: Property of Delhi University
- Report source License: CC0
- Trained Model License: WTFPL