Hacker News new | ask | show | jobs
by Giho 2222 days ago
I wonder how it "thinks"/"sees". Because the AI clearly couldn't recognize my version (putting out dots until it became like a continuous line).
1 comments

I think this is an important point. How is the data processed, given and saved? -Are the movements and start to stop positions saved when doing the strokes? -Or is a picture taken after the stroke is released and the pixels compared one by one? - is it made to vector art and then compared? -or just take mouse input, button pressed and not in a time log sort of and compare?