visualize_segmentation_data

hybrid_learning.datasets.data_visualization.visualize_segmentation_data(dataset, save_as=None, max_num_samples=5, skip_none=True, skip_empty=True, shuffle=False)[source]

Visualize a dataset yielding tuples of the form (input, target_mask).

Both input and target_mask must be images as torch.Tensor of the same width and height.

Parameters

  • dataset (BaseDataset) – the dataset to visualize

  • save_as (Optional[str]) – file path to save the image under using matplotlib.pyplot.savefig(); not saved if None

  • max_num_samples (int) – the maximum number of samples to show for each dataset

  • skip_none (bool) – whether to ignore when a None value in input or target is encountered

  • skip_empty (bool) – whether to ignore samples with all-black target masks

  • shuffle (bool) – whether to randomly select samples from the dataset