Closed vict0rsch closed 4 years ago
Train per-pixel linear regressors (42 metos => 3 rgb) then infer on data then at train time substract these inferences from the input to remove the least meaning variations
@krisrs1128 is that it?
Who's doing it?
This no longer seems relevant.
Train per-pixel linear regressors (42 metos => 3 rgb) then infer on data then at train time substract these inferences from the input to remove the least meaning variations
@krisrs1128 is that it?
Who's doing it?