Cleaned up dataset script for report

Changed test_set_toggle to global variable from constant

Added plotting code for Loss over Epochs during training phase

- Swapped tensor conversion code to faster version
- Added preliminary CNN code (non functional)

- Added dataloader code
- Added (broken) image and landmark plotting code
- Reverted list to tensor conversion back to list to numpy to tensor conversion
- Added debug exports for final landmark, confidence score and mfcc arrays

-Separated landmarks, confidence scores and mfccs into different arrays.
- Dataset arrays now storing as Tensors
- Dataset now includes all 20 or 19 videos clips, instead of just clip1
- Cleaned up debug prints

Dataset class len() and getitem() overloading,
and variable name cleanup for functions

Added dataset class, added code to import landmark data and mfcc data from files. Needs further development to allow for the use of dataloader iterator for training the CNN.

Code is rough, i.e. needs a bit of clean up and proper commenting throughout... But it is fully functional... MFCC generation consistency with total number of frames is not checked yet however

Finished script using the face_alignment library, but the last three functions haven't been tested yet...

Need to do a code clean up, to remove old comments / code.

Need to turn the script into a function and add it to the main codebase...

Still need to implement bbox cropping!

Saved midterm data output into a backup folder, for later examination.

Saved debug prints for vid13 to vid20 as txt file

Added comments in PEP8 style to midterm submission codebase.

Renamed and backed up old script contents before changes.

Split up MFCC array into files holding each row... That correspond to their own individual frames...

Added debug print() statements for fps, framerate and length

Temporarily changed MFCC output to .txt file for debugging purposes

Unfinished, needs further testing and evaluation / understanding

Added MFCC code to codebase as audio_processing() function

Improved debug print() output

Removed vid_num argument from frame_processing() function

Will later be integrated into the existing codebase

Cleaned up print() debugging code

Added separate print() statements to delineate between video processing and frame processing stages