preprocess Namespace Reference

Functions
Tuple	preprocess (DictConfig cfg=None)
def	apply_rescaling (tf.data.Dataset dataset=None, float scale=None, float offset=None)
tf.Tensor	preprocess_input (np.ndarray image, dict input_details)

Function Documentation

apply_rescaling()

def preprocess.apply_rescaling	(	tf.data.Dataset	dataset = None,
		float	scale = None,
		float	offset = None
	)

Applies rescaling to a dataset using a tf.keras.Sequential model.

Args:
    dataset (tf.data.Dataset): The dataset to be rescaled.
    scale (float): The scaling factor.
    offset (float): The offset factor.

Returns:
    The rescaled dataset.

Definition at line 87 of file preprocess.py.

Referenced by evaluate.evaluate(), and quantize.quantize().

Here is the caller graph for this function:

preprocess()

Tuple preprocess.preprocess

(

DictConfig

cfg = None

)

Preprocesses the data based on the provided configuration.

Args:
    cfg (DictConfig): Configuration object containing the settings.

Returns:
    Tuple: A tuple containing the following:
        - data_augmentation (object): Data augmentation object.
        - augment (bool): Flag indicating whether data augmentation is enabled.
        - pre_process (object): Preprocessing object.
        - train_ds (object): Training dataset.
        - valid_ds (object): Validation dataset.

Definition at line 36 of file preprocess.py.

References data_loader.load_dataset().

Here is the call graph for this function:

preprocess_input()

tf.Tensor preprocess.preprocess_input	(	np.ndarray	image,
		dict	input_details
	)

Preprocesses an input image according to input details.

Args:
    image: Input image as a NumPy array.
    input_details: Dictionary containing input details, including quantization and dtype.

Returns:
    Preprocessed image as a TensorFlow tensor.

Definition at line 111 of file preprocess.py.