imlab.train_classifier.Train

class imlab.train_classifier.Train(datas: dict, model_type: str, image_size: int, weight: Optional[str] = None, mod: str = 'new')

Bases: Wrapper

Train. Trainer class

__init__(datas: dict, model_type: str, image_size: int, weight: Optional[str] = None, mod: str = 'new') → None

__init__.

Parameters

• datas (dict) – {label:[image]}

• model_type (str) –

• image_size (int) –

• weight (str) –

• mod (str) –

Return type

None

Methods

__init__(datas, model_type, image_size[, ...])	__init__.
count_classes(data)	count_classes.
gen_data([test_size])	gen_data.
generator(datas)	generator.
get_model_train()	get_model_train.
norm_input(image)	norm_input.
train([steps, batch_size, learning_rate, ...])	train.

count_classes(data: dict) → int

count_classes. count number of classes

Parameters

data (dict) –

Return type

int

gen_data(test_size: float = 0.1) -> (<class 'object'>, <class 'object'>)

gen_data. generate data to tensorflow format

Parameters

test_size (float) –

Return type

(object, object)

generator(datas: list[tuple]) -> (<class 'numpy.ndarray'>, <class 'str'>)

generator.

Parameters

datas (list[tuple]) –

Return type

(np.ndarray, str)

get_model_train() → object

get_model_train. get model to be trained

Return type

object

norm_input(image: object) → ndarray

norm_input. normalize input

Parameters

image (object) –

Return type

np.ndarray

train(steps: int = 200, batch_size: int = 32, learning_rate: float = 0.001, path_chkp: str = '') → object

train.

Parameters

• steps (int) –

• batch_size (int) –

• learning_rate (float) –

• path_chkp (str) –

Return type

object