4 ms·
Yes, it's quite possible for ML to hallucinate ink, though it is on a much more local scale, like predicting a slightly longer stroke, filling in more of a char
by verditelabs 4mo ago
Yes, it's quite possible for ML to hallucinate ink, though it is on a much more local scale, like predicting a slightly longer stroke, filling in more of a character than is actually in the data, etc. Perhaps enough to change a reading of a character or show where ink isnt. It is difficult for ink detection to hallucinate grammatical and idiomatic greek and latin.
- im3w1l 4mo agoWhat is the input to the ML algorithm? Does it know the surrounding context so that it has a chance to deduce "if this stroke is slightly longer then the end result will be idiomatic greek and latin"?
- verditelabs 4mo agoThe input is 3d chunks of reconstructed CT data from our scans. I can't remember the specifics but maybe enough voxels for .5mm^3 at a time or so? They're all available for free from https://registry.opendata.aws/vesuvius-challenge-herculaneum-scrolls/ https://registry.opendata.aws/vesuvius-challenge-herculaneum... . Our trained models are all available at https://huggingface.co/scrollprize https://huggingface.co/scrollprize