Predict the coordinates of vowels from 2D sights.
Extend the base class to predict the coordinates of vowels from 2D sights. It overrides methods to provide specific calculations for different vowel positions based on facial landmarks.
It adds the calculation of a radius for each vowel: it allows to define an area in which the target is acceptable.
This model was created from observed values in a corpus.
- Collected data: 3769 values
- Filtered data: 3752 values (outliers removed)
Statistics by Hand Position - b: Count = 230, Mean = (144, 397), Std Dev = (70.45, 92.26) - c: Count = 529, Mean = (425, 888), Std Dev = (112.48, 76.10) - m: Count = 676, Mean = (260, 682), Std Dev = (79.36, 90.67) - s: Count = 1750, Mean = (-193, 635), Std Dev = (241.45, 232.91) - t: Count = 584, Mean = (435, 1419), Std Dev = (143.24, 223.21)
