edflow.data.util package

Submodules:

• edflow.data.util.cached_dset module
• edflow.data.util.util_dsets module

Summary

Reference

edflow.data.util.flow2hsv(flow)

Given a Flowmap of shape [W, H, 2] calculates an hsv image, showing the relative magnitude and direction of the optical flow.

Parameters

flow (np.array) – Optical flow with shape [W, H, 2].

Returns

Containing the hsv data.

Return type

np.array

edflow.data.util.cart2polar(x, y)

Takes two array as x and y coordinates and returns the magnitude and angle.

edflow.data.util.hsv2rgb(hsv)

color space conversion hsv -> rgb. simple wrapper for nice name.

edflow.data.util.flow2rgb(flow)

converts a flow field to an rgb color image.

Parameters

flow (np.array) – optical flow with shape [W, H, 2].

Returns

Containing the rgb data. Color indicates orientation, intensity indicates magnitude.

Return type

np.array

edflow.data.util.get_support(image)

… warning: This function makes a lot of assumptions that need not be met!

Assuming that there are three categories of images and that the image_array has been properly constructed, this function will estimate the support of the given image.

Parameters

image (np.ndarray) – Some properly constructed image like array. No assumptions need to be made about the shape of the image, we simply assme each value is some color value.

Returns

The support. Either ‘0->1’, ‘-1->1’ or ‘0->255’

Return type

str

edflow.data.util.sup_str_to_num(support_str)

Converts a support string into usable numbers.

edflow.data.util.adjust_support(image, future_support, current_support=None, clip=False)

Will adjust the support of all color values in image.

Parameters

• image (np.ndarray) – Array containing color values. Make sure this is properly constructed.

• future_support (str) – The support this array is supposed to have after the transformation. Must be one of ‘-1->1’, ‘0->1’, or ‘0->255’.

• current_support (str) – The support of the colors currentl in image. If not given it will be estimated by get_support().

• clip (bool) – By default the return values in image are simply coming from a linear transform, thus the actual support might be larger than the requested interval. If set to True the returned array will be cliped to future_support.

Returns

The given image with transformed support.

Return type

same type as image

edflow.data.util.clip_to_support(image, supp_str)

edflow.data.util.add_im_info(image, ax)

Adds some interesting facts about the image.

edflow.data.util.im_fn(key, im, ax)

Plot an image. Used by plot_datum().

edflow.data.util.heatmap_fn(key, im, ax)

Assumes that heatmap shape is [H, W, N]. Used by plot_datum().

edflow.data.util.keypoints_fn(key, keypoints, ax)

Plots a list of keypoints as a dot plot.

edflow.data.util.flow_fn(key, im, ax)

Plot an flow. Used by plot_datum().

edflow.data.util.other_fn(key, obj, ax)

Print some text about the object. Used by plot_datum().

edflow.data.util.default_heuristic(key, obj)

Determines the kind of an object. Used by plot_datum().

edflow.data.util.plot_datum(nested_thing, savename='datum.png', heuristics=<function default_heuristic>, plt_functions={'flow': <function flow_fn>, 'heat': <function heatmap_fn>, 'image': <function im_fn>, 'keypoints': <function keypoints_fn>, 'other': <function other_fn>})

Plots all data in the nested_thing as best as can.

If heuristics is given, this determines how each leaf datum is converted to something plottable.

Parameters

• nested_thing (dict or list) – Some nested object.

• savename (str) – Path/to/the/plot.png.

• heuristics (Callable) – If given this should produce a string specifying the kind of data of the leaf. If None determinde automatically. See default_heuristic().

• plt_functions (dict of Callables) – Maps a kind to a function which can plot it. Each callable must be able to receive a the key, the leaf object and the Axes to plot it in.