Shortcuts

NYUv2

class torch_uncertainty.datasets.NYUv2(root, split, transforms=None, min_depth=0.0, max_depth=10.0, download=False)[source]

NYUv2 depth dataset.

Parameters:
  • root (Path | str) – Root directory where dataset is stored.

  • split (Literal["train", "val"]) – Dataset split.

  • transforms (Callable | None) – Transform to apply to samples & targets. Defaults to None.

  • min_depth (float) – Minimum depth value. Defaults to 1e-3.

  • max_depth (float) – Maximum depth value. Defaults to 10.

  • download (bool) – Download dataset if not found. Defaults to False.