turi::object_detection::InputBatch Struct Reference

#include <toolkits/object_detection/od_model_trainer.hpp>

Public Attributes
neural_net::shared_float_array	images
std::vector< std::vector< neural_net::image_annotation > >	annotations
std::vector< std::pair< size_t, size_t > >	image_sizes

Detailed Description

Represents one batch of model-agnostic data, post-augmentation/resizing.

Definition at line 38 of file od_model_trainer.hpp.

Member Data Documentation

annotations

std::vector<std::vector<neural_net::image_annotation> > turi::object_detection::InputBatch::annotations

The raw annotations from the DataBatch.

Definition at line 46 of file od_model_trainer.hpp.

image_sizes

std::vector<std::pair<size_t, size_t> > turi::object_detection::InputBatch::image_sizes

The original height and width of each image, used to scale bounding-box predictions.

Definition at line 52 of file od_model_trainer.hpp.

images

neural_net::shared_float_array turi::object_detection::InputBatch::images

The (RGB) images from a DataBatch encoded as NHWC.

Definition at line 43 of file od_model_trainer.hpp.

---

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

• toolkits/object_detection/od_model_trainer.hpp