turi::neural_net::float_array_image_augmenter Class Reference

#include <ml/neural_net/image_augmentation.hpp>

Classes
struct	float_array_result
struct	labeled_float_image

Public Member Functions
const options &	get_options () const override
result	prepare_images (std::vector< labeled_image > source_batch) override

Detailed Description

An abstract class that inherits from image_augmenter used to convert input images, annotations and predictions to shared_float_arrays for tf_image_augmenter. Subclass must be written for it if needed. The subclass must implement the pure virtual method prepare_augmented_images.

Definition at line 262 of file image_augmentation.hpp.

Member Function Documentation

get_options()

const options& turi::neural_net::float_array_image_augmenter::get_options ( ) const

inlineoverridevirtual

Returns the options parameterizing this instance.

Implements turi::neural_net::image_augmenter.

Definition at line 266 of file image_augmentation.hpp.

prepare_images()

result turi::neural_net::float_array_image_augmenter::prepare_images ( std::vector< labeled_image >  source_batch )

overridevirtual

Performs augmentation on a batch of images (and their annotations).

If the source batch is smaller than the batch size specified in the options, then the result is padded with zeroes as needed.

Implements turi::neural_net::image_augmenter.

---

The documentation for this class was generated from the following file:

• ml/neural_net/image_augmentation.hpp