JINetModel

class aydin.nn.models.jinet.JINetModel(*args, **kwargs)[source]

The JINet model is a hybrid CNN-perceptron model that leverages dilated convolutions to incorporate a built-in ‘blind-spot’ as required for N2S denoising. Supports both 2D and 3D images.

fit(input_image, target_image, batch_size, callbacks, verbose=None, max_epochs=None, total_num_patches=None, img_val=None, create_patches_for_validation=None, train_valid_ratio=None)[source]
Parameters
input_image
target_image
batch_size
callbacks
verbose
max_epochs
total_num_patches
img_val
create_patches_for_validation
train_valid_ratio
Returns
loss_history
predict(x, batch_size=None, verbose=0, steps=None, callbacks=None, max_queue_size=10, workers=1, use_multiprocessing=False)[source]

Overwritten model predict method.

Parameters
x
batch_size
verbose
steps
callbacks
max_queue_size
workers
use_multiprocessing
size()[source]

Returns size of the model in bytes