caer.preprocessing

caer.preprocessing.compute_mean(data, channels, per_channel_subtraction=True)

Computes mean per channel over the train set and returns a tuple of dimensions=channels Train should not be normalized

Return type

Optional[Tuple]

caer.preprocessing.compute_mean_from_dir(DIR, channels, per_channel_subtraction=True, recursive=True)

Computes mean per channel Mean must be computed ONLY on the train set

Return type

Optional[tuple]

caer.preprocessing.subtract_mean(data, channels, mean_sub_values)

Per channel subtraction values computed from compute_mean() or compute_mean_from_dir() Subtracts mean from the validation set

Return type

List[str]