Active Learning Base Class

all active learning methods are inherited from ActiveLearning class. Only initialize methods are different. It means that you can use any active learning method by calling sample method of ActiveLearning class.

sample(image_dir, num_images, result_dir, save_images=True, hold=True)

Sample images from the image directory

Parameters:

Name	Type	Description	Default
sampled_image_dir	Union[Path, str]	image directory	required
num_images	int	number of images to sample	required
result_dir	Union[Path, str]	result directory	required
save_images	bool	save sampled images. Defaults to True.	True
hold	bool	hold process. Defaults to True.	True

Returns:

Name	Type	Description
str	str	result directory

RandomSampling

Bases: ActiveLearning

__init__(seed=0)

EntropySampling

Bases: ActiveLearning

__init__(hub, image_size=None, letter_box=None, batch_size=32, num_workers=4, device='0')

Entropty Sampling

Parameters:

Name	Type	Description	Default
hub	Hub	Hub	required
image_size	int	Image size. Defaults to None.	None
letter_box	bool	Letter box. Defaults to None.	None
batch_size	int	Batch size. Defaults to 32.	32
num_workers	int	Number of workers. Defaults to 4.	4
device	str	Device. Defaults to "0".	'0'

PL2NSampling

Bases: ActiveLearning

__init__(hub, diversity_sampling=False, image_size=None, letter_box=None, batch_size=32, num_workers=4, device='0')

PL2N Sampling

Parameters:

Name	Type	Description	Default
hub	Hub	Hub	required
diversity_sampling	bool	Diversity sampling. Defaults to False.	False
image_size	int	Image size. Defaults to None.	None
letter_box	bool	Letter box. Defaults to None.	None
batch_size	int	Batch size. Defaults to 32.	32
num_workers	int	Number of workers. Defaults to 4.	4
device	str	Device. Defaults to "0".	'0'