Datasets API Reference

PyTorch Dataset classes for loading and processing data.

Module: `datasets.base`

Class: `BaseNoiseDataset`

Base dataset with common functionality.

Constructor:

python

BaseNoiseDataset(image_paths, patch_size=None, patches_per_image=None)

Parameters:

image_paths (list): List of image file paths
patch_size (int, optional): Patch size
patches_per_image (int, optional): Patches per image

Methods:

__len__(): Returns dataset length
load_image(idx): Load and normalize image at index
preprocess_for_roi(img_path, scale_factor): Preprocess for ROI detection
get_roi_patches(preprocessed_img, patch_size, threshold, above_threshold, scale_factor): Get ROI patch coordinates

Module: `datasets.training`

Class: `TrainingDataset`

Training dataset with masking and augmentation.

Constructor:

python

TrainingDataset(image_paths, patch_size, kernel_size, mask, mask_percentage, 
                mask_strat=0, prediction_kernel=None, patches_per_image=100,
                use_roi=True, scale_factor=0.25, roi_threshold=0.5, 
                select_background=True, use_augmentation=True)

Parameters:

image_paths (list): Image paths
patch_size (int): Patch size
kernel_size (int): Kernel size for local mean
mask (np.ndarray): Boolean mask array
mask_percentage (float): Mask percentage (0-100)
mask_strat (int): Masking strategy (0=local mean, 1=zeros, 2=random)
prediction_kernel (np.ndarray): Prediction kernel
patches_per_image (int): Patches per image
use_roi (bool): Use ROI selection
scale_factor (float): ROI scale factor
roi_threshold (float): ROI threshold
select_background (bool): Select background vs structures
use_augmentation (bool): Apply augmentation

Methods:

__getitem__(idx): Get training sample
- Returns: (input_tensor, target_tensor, mask_tensor)

Example:

python

from autoStructN2V.datasets import TrainingDataset

dataset = TrainingDataset(
    image_paths=train_paths,
    patch_size=32,
    kernel_size=3,
    mask=mask_array,
    mask_percentage=15.0,
)

Module: `datasets.validation`

Class: `ValidationDataset`

Validation dataset without masking.

Constructor:

python

ValidationDataset(image_paths, patch_size, patches_per_image=100,
                  use_roi=False, scale_factor=0.25, roi_threshold=0.5, 
                  select_background=True)

Methods:

__getitem__(idx): Get validation sample
- Returns: (input_tensor, input_tensor, ones_mask)

Module: `datasets.testing`

Class: `TestDataset`

Test dataset for full image processing.

Constructor:

python

TestDataset(image_paths)

Methods:

__getitem__(idx): Get full test image
- Returns: (input_tensor, input_tensor, ones_mask)

Example:

python

from autoStructN2V.datasets import TestDataset

test_dataset = TestDataset(image_paths=test_paths)
test_loader = DataLoader(test_dataset, batch_size=1)

See Also:

Datasets API Reference ​

Module: datasets.base ​

Class: BaseNoiseDataset ​

Module: datasets.training ​

Class: TrainingDataset ​

Module: datasets.validation ​

Class: ValidationDataset ​

Module: datasets.testing ​

Class: TestDataset ​

Datasets API Reference

Module: `datasets.base`

Class: `BaseNoiseDataset`

Module: `datasets.training`

Class: `TrainingDataset`

Module: `datasets.validation`

Class: `ValidationDataset`

Module: `datasets.testing`

Class: `TestDataset`