neg_pixel_prop

hybrid_learning.datasets.data_visualization.neg_pixel_prop(data, max_num_samples=10, show_progress_bar=False)[source]

Collect the mean proportion of negative pixels in the binary segmentation mask data. The proportion is estimated from the first max_num_samples samples in the dataset. Set the value to None to take into account all samples in the dataset.

Parameters

  • data – a typing.Sequence that yields tuples of input (arbitrary) and masks (as torch.Tensor).

  • max_num_samples (Optional[int]) – the maximum number of samples to take into account for the estimation

  • show_progress_bar (bool) – whether to show a progress bar for loading the masks

Returns

the proportion of negative pixels in all (resp. the first num_samples) binary segmentation masks contained in the given data

Return type

float