Working with preprocessing layers

Authors: Francois Chollet, Mark Omernick

View on TensorFlow.org

Run in Google Colab

View source on GitHub

View on keras.io

Keras preprocessing

The Keras preprocessing layers API allows developers to build Keras-native input processing pipelines. These input processing pipelines can be used as independent preprocessing code in non-Keras workflows, combined directly with Keras models, and exported as part of a Keras SavedModel.

With Keras preprocessing layers, you can build and export models that are truly end-to-end: models that accept raw images or raw structured data as input; models that handle feature normalization or feature value indexing on their own.

Available preprocessing

Text preprocessing

• tf.keras.layers.TextVectorization: turns raw strings into an encoded representation that can be read by an Embedding layer or Dense layer.

Numerical features preprocessing

• tf.keras.layers.Normalization: performs feature-wise normalization of input features.
• tf.keras.layers.Discretization: turns continuous numerical features into integer categorical features.

Categorical features preprocessing

• tf.keras.layers.CategoryEncoding: turns integer categorical features into one-hot, multi-hot, or count dense representations.
• tf.keras.layers.Hashing: performs categorical feature hashing, also known as the "hashing trick".
• tf.keras.layers.StringLookup: turns string categorical values into an encoded representation that can be read by an Embedding layer or Dense layer.
• tf.keras.layers.IntegerLookup: turns integer categorical values into an encoded representation that can be read by an Embedding layer or Dense layer.

Image preprocessing

These layers are for standardizing the inputs of an image model.

• tf.keras.layers.Resizing: resizes a batch of images to a target size.
• tf.keras.layers.Rescaling: rescales and offsets the values of a batch of images (e.g. go from inputs in the [0, 255] range to inputs in the [0, 1] range.
• tf.keras.layers.CenterCrop: returns a center crop of a batch of images.

Image data augmentation

These layers apply random augmentation transforms to a batch of images. They are only active during training.

• tf.keras.layers.RandomCrop
• tf.keras.layers.RandomFlip
• tf.keras.layers.RandomTranslation
• tf.keras.layers.RandomRotation
• tf.keras.layers.RandomZoom
• tf.keras.layers.RandomContrast

The adapt() method

Some preprocessing layers have an internal state that can be computed based on a sample of the training data. The list of stateful preprocessing layers is:

• TextVectorization: holds a mapping between string tokens and integer indices
• StringLookup and IntegerLookup: hold a mapping between input values and integer indices.
• Normalization: holds the mean and standard deviation of the features.
• Discretization: holds information about value bucket boundaries.

Crucially, these layers are non-trainable. Their state is not set during training; it must be set before training, either by initializing them from a precomputed constant, or by "adapting" them on data.

You set the state of a preprocessing layer by exposing it to training data, via the adapt() method:

import numpy as np
import tensorflow as tf
from tensorflow import keras