Our model now collects only X and Y coordinates for the both eyes. Reasearch if we can add new columns to our training csv to improve our results.