Fall detection

1. data_conversion
   • Converts data between .bin, .gif, and .npy
2. ml
   • Contains functions that supports or enables ML functionalities
   • filters contains functions that
     • are applied before train and test sets are generated
     • maintains the shape of the input data and
     • maintains the meanings of each dimension (i.e. 3rd dimension represents time, 1st dimension represent doppler etc.)
   • feature_extract contains functions that
     • are applied before train and test sets are generated
     • are not filters
   • misc contains functions that are neither filters

Logs

v4.0 (git branch "gui")

• Added ml_final - storing the code used to train weights and run live demo (and their respective tests)

v3.0

• Add global and local normalization of range_profile() and doppler_profile()
• New filter - Downsampling of doppler axis by factor of 2
• Added ml_training - having only the filters/feature_extraction that is required to train SVM for final implementation, and most importantly transposes each frame before training and testing, and saves trained model weights

v2.0

• Restructuring of source code
  • three phases:
    • preprocessing: Output data must still have the same shape as input data
      • Each data point should have features and labels, and should look something like the format below:

             [
               [DATA OF ONE FRAME],
               [DATA OF ONE FRAME]
             ], 
             [LABEL_NUMBER]]```
        

    • feature extraction: Output data shape can vary depending on model
    • model
  • each phases are called as imported functions instead of being executed as individual files
    • allow for shared variables in the overall wrapping function
    • simpler setup and teardown for testing of functions
  • labelling will be embedded into each datapoint (not depending on parent dir names)
• Make remove center 3 lines a filter in the preprocessing phase
• Write function to label data from csv file
• Write functions to retrieve energy and range features (given by xavier)