classes_from_labels¶
- classes_from_labels(labels: ndarray, delim: str = '-', which: int = 0, crop: slice = slice(None, None, None), cats: dict = None) tuple[dict, ndarray][source][source]¶
Extract class IDs from string labels.
- This function processes string labels to extract class IDs using a
- delimiter, and returns a dictionary mapping class names to indices and an
array of class indices.
- Parameters:
labels (
np.ndarray) – Array of string labels to process.delim (
str, optional) – Delimiter to split the labels, by default ‘-‘.which (
int, optional) – Which part of the split label to use, by default 0.crop (
slice, optional) – Slice to apply to each label part, by default slice(None).cats (
dict, optional) – Existing category mapping to use. If None, a new mapping is created.
- Returns:
A tuple containing: - Dictionary mapping class names to indices - Array of class indices corresponding to the input labels
- Return type:
tuple[dict,np.ndarray]
Examples
>>> labels = np.array(['cat-dog', 'dog-cat', 'cat-bird']) >>> classes_from_labels(labels, delim='-') ({'cat': 0, 'dog': 1}, array([0, 1, 0]))
