16-05-2019, 10:37 PM
My point is by looking at all the gallows characters and by aggregating we can say what the shape looks like. This is why I referred to image recognition. What the shape actually looks like is important for my research. There is natural variation in how a shape is drawn.

