`sparkocr.transformers.visual_encoder_decoder.Dit`#

Module Contents#

Classes#

`Dit`
`DitDatasetIterable`	Creates an IterableDataset from a PySpark Dataframe

Functions#

normalize_bbox(bbox, width, height)

class Dit(base_model_path, params)#

base_model_path#

batch_size#

checkpoint_model#

device#

label2id#

model#

params#

patch_size#

smoothing = 0.1#

window_size#

finish_model(checkpoint_model)#

to_onnx(onnx_model_path, validate=False, atol=0.01)#

train(df)#

update_config_labels(labels)#

class DitDatasetIterable(df, image_col='image', labels_col='label', n_width=224, n_height=224, label2id={})#

Bases: torch.utils.data.IterableDataset

Creates an IterableDataset from a PySpark Dataframe It returns images and labels.

dataset#

df#

image_col = 'image'#

label2id#

label_col = 'label'#

n_height = 224#

n_width = 224#

normalize_bbox(bbox, width, height)#