the repo is incomple. you can find how it look like in directory_structure.txt

The goal of this project is to use data collect by MIC and IMU data to do reconization, you can find more info here https://docs.google.com/presentation/d/1lcTPnxFe2JT8OKlVd3GqUNlWWhBYevT8/edit?usp=sharing&ouid=109552043465733562677&rtpof=true&sd=true

I did 15 recordings, and each recording consists of 12 sets of transitions to the gesture.

Additionally, I had a video for each recording.

The data at the beginning comprises 15*2 RAW files. First,

I extracted the data and created CSV, PLOT, and audio files for all of them.

Since all 12 sets of transitions of the gesture are within one recordingthe,

I extracted each different set of data into separate CSV files for different recording

and extracted each different set segmant of video for ground truth.

I reformatted the IMU CSV files,

performed MFCC on the MIC data,

reformatted the MIC CSV files,

combined both, and reformatted them for training.

Then, I trained the SVM model.