bobev18 / gamcon Goto Github PK

digit_split_n_label.py -- produces "flattened_images.txt" and "flattened_images.txt" via interactive processing(splitting) and labling of digits 0 to 9 based on raw image "samples/score_digits.png"

flattened_images.txt -- flattened digit images classifications.txt -- labels for the "flattened_images.txt"

mydigits.ipynb -- uses flattened images of digits to add distortions single_digits.ipynb -- uses flattened images of digits to grid search ML params, saves to "score_model.pickle"

train_on_overlapping_digits.ipynb -- uses raw images of scores with overlap to == split digits == flatten images == train ML

unchecked == learn_alt.py environment.py

TODO:

• encapsulate keyboard interactions
• find a way to send signal to the app, while focus is on the game (i.e. via mouse)
• refactor score
• test overlapping model
• clean/refresh requirements