Skip to content

Uses 4 simultaneous EMG channels to collect muscle data during exercise. Data is processed using ASP and DSP and sent through Bluetooth where a machine learning model classifies the exercise.

Notifications You must be signed in to change notification settings

jerryliu3/EMG-Exercise-Analysis

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

77 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

EMG-Exercise-Analysis

Overview:

  • Circuits and PCB for 4 channel EMG data acquisition
  • Bluno beetle and Android app interfacing for wireless real-time transfer of data
  • Machine learning and implemention of Random Forest Classifier for 9 exercises

Current Project Functionalities:

  1. Take EMG signals from 4 channels and pass them through the PCB to remove noise and amplify the signal

  2. Transfer the data over Bluetooth to the Android app which receives and updates multiline graphs in real time.

  3. Save the data on a local file or online database

  4. Put the data through a random forest classifier for exercise classification of 9 different exercises

Problems We Encountered:

  • Low quality circuit boards are near impossible to work with
  • Natural inconsistencies in the hardware leading to hours of debugging
  • Lack of guides for using a multiplexer
  • Training the random forest classifier

To Do:

  1. Attach the “nervous system” to the “skin”

  2. Pass data through seglearn

    • Option 1: get seglearn working on a phone
    • Option 2: get the data off the phone and run seglearn on a computer
  3. Improve machine learning and database handling

  4. Test interfacing of the app and the Bluno Beetle instead of the app and Arduino for communication

Maintainers:

About

Uses 4 simultaneous EMG channels to collect muscle data during exercise. Data is processed using ASP and DSP and sent through Bluetooth where a machine learning model classifies the exercise.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published