geowatch.tasks.depth.predict module

geowatch.tasks.depth.predict.fake_model(batch2, tta=True)[source]
geowatch.tasks.depth.predict.run_inference(image, model, device=0)[source]

Example

>>> from geowatch.tasks.depth.predict import *  # NOQA
>>> import kwimage
>>> import kwarray
>>> src = kwimage.ensure_float01(kwimage.grab_test_image(dsize=(512, 512)))
>>> src = kwimage.Polygon.random(rng=None).scale(src.shape[0]).fill(src.copy(), np.nan)
>>> model = fake_model
>>> image = src * 255
>>> device = 'cpu'
>>> result = run_inference(image, model, device=device)
>>> # xdoctest: +REQUIRES(--show)
>>> import kwplot
>>> kwplot.autompl()
>>> kwplot.imshow(src, pnum=(1, 2, 1), doclf=True)
>>> kwplot.imshow(result, pnum=(1, 2, 2))
geowatch.tasks.depth.predict.anisotropic_diffusion(img, niter=1, kappa=50, gamma=0.1, voxelspacing=None, option=1)[source]

Edge-preserving, XD Anisotropic diffusion.

Parameters:

  • img (array_like) – Input image (will be cast to numpy.float).

  • niter (integer) – Number of iterations.

  • kappa (integer) – Conduction coefficient, e.g. 20-100. kappa controls conduction as a function of the gradient. If kappa is low small intensity gradients are able to block conduction and hence diffusion across steep edges. A large value reduces the influence of intensity gradients on conduction.

  • gamma (float) – Controls the speed of diffusion. Pick a value \(<= .25\) for stability.

  • voxelspacing (tuple of floats or array_like) – The distance between adjacent pixels in all img.ndim directions

  • option ({1, 2, 3}) – Whether to use the Perona Malik diffusion equation No. 1 or No. 2, or Tukey’s biweight function. Equation 1 favours high contrast edges over low contrast ones, while equation 2 favours wide regions over smaller ones. See [1] for details. Equation 3 preserves sharper boundaries than previous formulations and improves the automatic stopping of the diffusion. See [2] for details.

Returns:

anisotropic_diffusion – Diffused image.

Return type:

ndarray

Notes

Original MATLAB code by Peter Kovesi, School of Computer Science & Software Engineering, The University of Western Australia, pk @ csse uwa edu au, <http://www.csse.uwa.edu.au>

Translated to Python and optimised by Alistair Muldal, Department of Pharmacology, University of Oxford, <alistair.muldal@pharm.ox.ac.uk>

Adapted to arbitrary dimensionality and added to the MedPy library Oskar Maier, Institute for Medical Informatics, Universitaet Luebeck, <oskar.maier@googlemail.com>

June 2000 original version. - March 2002 corrected diffusion eqn No 2. - July 2012 translated to Python - August 2013 incorporated into MedPy, arbitrary dimensionality -

References