Main Menpo Widgets

These are top-level functions for visualizig the various menpo objects with high-level interactive widgets.

  1. Basics
  2. PointClouds/PointGraphs Widget
  3. LandmarkGroups Widget
  4. Landmarks Widget
  5. Images Widget
  6. Patches Widget
  7. Features Widget
  8. Plot Graph Widget
  9. Save Figure Widget
  10. View Widget Methods

We highly recommend that you render all matplotlib figures inline the Jupyter notebook for the best menpowidgets experience. This can be done by running
%matplotlib inline
in a cell. Note that you only have to run it once and not in every rendering cell.

1. Basics

Let's first import all the widgets:

from menpowidgets import (visualize_pointclouds, visualize_landmarkgroups,
                          visualize_landmarks, visualize_images, visualize_patches,
                          plot_graph, save_matplotlib_figure, features_selection)

They are all functions which have some common arguments:

  • style
    It can be either 'coloured' or 'minimal'. The 'coloured' style uses a colouring theme that is different for each widget. The 'minimal' style is very simple with black and white colours.
  • figure_size
    This argument is a tuple that defines the size of the rendered figure in inches. The figure size can also be controlled from the Renderer options within the widgets.
  • browser_style
    It can be either 'buttons' or 'slider'. This argument exists in widgets that visualize a list of objects (e.g. images or pointclouds). If 'buttons', then the object selection will be done by the / buttons. If 'slider', then the object selection is done using a slider widget.

Note that all widgets can get as input a list of objects with totally different attributes between them. For example, list of images with different number of channels or list of LandmarkGroups with different number of points. Finally, they all have a Renderer tab that has many rendering-related options (such as lines, markers, axes, legend, grid, image) and an Export tab that allows the user to save the figure to file.

Before moving on, let's create some lists of objects to visualize with different properties. First let's import what is needed.

%matplotlib inline
import menpo.io as mio
from menpo.landmark import (labeller, face_ibug_68_to_face_ibug_49,
                            face_ibug_68_to_face_ibug_66,
                            face_ibug_68_to_face_ibug_68_trimesh,
                            face_ibug_68_to_face_ibug_68)
from menpo.feature import igo, hog, lbp

Now let's load the builtin assets and differentiate them:

im1 = mio.import_builtin_asset.breakingbad_jpg()
im1 = im1.crop_to_landmarks_proportion(0.2)
labeller(im1, 'PTS', face_ibug_68_to_face_ibug_68)

im2 = mio.import_builtin_asset.einstein_jpg()
im2 = im2.crop_to_landmarks_proportion(0.2)
im2 = igo(im2, double_angles=True)
labeller(im2, 'PTS', face_ibug_68_to_face_ibug_49)

im3 = mio.import_builtin_asset.lenna_png()
im3 = im3.crop_to_landmarks_proportion(0.2)
im3 = hog(im3)

im4 = mio.import_builtin_asset.takeo_ppm()
im4 = im4.crop_to_landmarks_proportion(0.2)
labeller(im4, 'PTS', face_ibug_68_to_face_ibug_68_trimesh)

im5 = mio.import_builtin_asset.tongue_jpg()
im5 = im5.crop_to_landmarks_proportion(0.2)
im5 = im5.as_greyscale()

im6 = mio.import_builtin_asset.menpo_thumbnail_jpg()

2. PointClouds/PointGraphs Widget

First, we need to add the objects in a list

pointclouds = [im1.landmarks['PTS'].lms,
               im2.landmarks['face_ibug_49'].lms,
               im3.landmarks['LJSON'].lms,
               im4.landmarks['PTS'].lms,
               im5.landmarks['PTS'].lms]

and then get the widget

visualize_pointclouds(pointclouds)

By pressing the button, the widget automatically goes through all and visualizes the objects. The animation speed can be controlled by / . The repeat mode can be controlled by / .

The 1st tab prints some information regarding each PointCloud.

The 2nd tab has rendering options regarding the markers, lines, numbering, axes and zoom. There are also two radiobuttons that define the axes mode.

The 3rd tab allows the user to save a figure to file.

3. LandmarkGroups Widget

Let's group all LandmarkGroups in a list

landmark_groups = [im1.landmarks['PTS'],
                   im2.landmarks['face_ibug_49'],
                   im3.landmarks['LJSON'],
                   im4.landmarks['PTS'],
                   im5.landmarks['PTS'],
                   im3.landmarks['LJSON']]

and visualize them using the visualize_landmarkgroups() widget with a slider browser style

visualize_landmarkgroups(landmark_groups, browser_style='slider')

The animation works as before, with the difference that the buttons are replaced by a slider.

The 1st tab prints some information regarding each LandmarkGroup.

The 2nd tab has options related to the landmarks. Specifically, the user can select to render specific labels.

The 3nd tab has rendering options regarding the lines, markers, numbering of points, legend, zoom and axes. Once again, there are also two radiobuttons that define the axes mode.

Finally, the last tab allows the user to save a figure to file.

4. Landmarks Widget

First we need to create a list of LandmarkManager objects and then visualize them using a 'minimal' style

landmarks = [im1.landmarks,
             im2.landmarks,
             im3.landmarks,
             im4.landmarks,
             im5.landmarks,
             im6.landmarks]

visualize_landmarks(landmarks, style='minimal')

This widget has the exact same structure and functionality as the previous one for LandmarkGroups. The only difference is that in the Landmarks tab, the group can also be specified in case a LandmarkManager has more than one groups.

5. Images Widget

Let's now create a list with the actual Image objects and visualize them

images = [im1, im2, im3, im4, im5, im6]
visualize_images(images)

Compared to the visualize_landmarks() widget, this one has an additional tab for Channels options. In this tab, the user has several options, such as to visualize the image's channels individually or in subplots, to render the image as a glyph and to render the sum of the image's channels.

6. Patches Widget

Let's now extract some patches from the images and create two lists: one with the patches and one with the patche centers.

patches1 = im1.extract_patches_around_landmarks(group='PTS')
pc1 = im1.landmarks['PTS'].lms
patches2 = im2.extract_patches_around_landmarks(group='face_ibug_49')
pc2 = im2.landmarks['face_ibug_49'].lms
patches3 = im3.extract_patches_around_landmarks(group='LJSON')
pc3 = im3.landmarks['LJSON'].lms
patches4 = im4.extract_patches_around_landmarks(group='face_ibug_68_trimesh')
pc4 = im4.landmarks['face_ibug_68_trimesh'].lms
patches5 = im5.extract_patches_around_landmarks(group='PTS')
pc5 = im5.landmarks['PTS'].lms

patches = [patches1, patches2, patches3, patches4, patches5]
patch_centers = [pc1, pc2, pc3, pc4, pc5]

visualize_patches(patches, patch_centers)

The visualize_patches is exactly the same with visualize_images, with the difference that it has a patch-related options tab.

7. Features Widget

menpowidgets also has a widget for selecting a specific feature with some options. In order to facilitate the options selection, the widget has a Preview tab that applies the feature on the Lenna image. The widget returns a function as a single element of a list. This can be done as

feat = features_selection()

Note that by pressing the Select button, the widget automatically closes. Then the function can be applied to a new image as:

feat[0](im1).view(channels=[0, 1]);

8. Plot Graph Widget

There is a widget that facilitates the graph plotting. It supports plotting of multiple curves. It expects a list of values for x_axis and a list of lists of values for y_axis, each one corresponding to a different curve. Let's generate the values of sin and cos

import numpy as np

x_axis = [x * np.pi / 10.0 for x in range(21)]

y_sin = list(np.sin(x_axis))
y_cos = list(np.cos(x_axis))
y_axis = [y_sin, y_cos]

The widget has a legend_enrties argument.

plot_graph(x_axis, y_axis, legend_entries=['sin', 'cos'])

In the 1st tab, the user can set the X/Y labels, the Title and the Legend entries.

In the 2nd tab, there are Renderer options that can be defined for each curve separately.

Finally, as always, the last tab allows the user to save the plot to file.

9. Save Figure Widget

The widget to save a figure to file can be called independently. Specifically, assume that we have figure from which we kept the rendering object:

renderer = im1.view_landmarks(group='PTS')
save_matplotlib_figure(renderer)

Note that if the Overwrite if file exists checkbox is not ticked, then a warning will appear when trying to save to a file that already exists.

10. View Widget Methods

Finally, all menpo objects have a .view_widget() method. Below are some examples:

images[0].view_widget(figure_size=(6, 4))
images[-1].view_widget(figure_size=(6, 4))
pointclouds[0].view_widget(figure_size=(6, 4))
landmark_groups[0].view_widget(figure_size=(6, 4))
landmarks[0].view_widget(figure_size=(6, 4))

results matching ""

    No results matching ""