Image RGB array must be uint8 or floating point; found int32

[ivan-uskov]

When using background=average, got this:

root@bfa4a21e5b8e:/app# python face_morpher/facemorpher/averager.py --work=/app/work/tmp/avg --out=work/result/1.png --background=average
<class 'numpy.ndarray'>
Averaged 3 work
Image RGB array must be uint8 or floating point; found int32

[Guml]

change

plt = plotter.Plotter(plot, num_images=1, out_filename=out_filename)
plt.save(dest_img)

to:
plt = plotter.Plotter(plot, num_images=1, out_filename=out_filename)
dest_img = dest_img.astype(np.uint8)
plt.save(dest_img)

from:
https://stackoverflow.com/questions/42044259/getting-black-plots-with-plt-imshow-after-multiplying-image-array-by-a-scalar

[nchlis]

Guml's answer above works.
Just don't forget to add "import numpy as np" in averager.py since it is not included by default.

[JonBoyleCoding]

This issue stems from the call to the locater here:

face_morpher/facemorpher/averager.py

Line 103 in 7a30611

average_background = locator.average_points(images)

The locator.average_points returns an int32 type as defined here:

face_morpher/facemorpher/locator.py

Lines 91 to 97 in 7a30611

	def average_points(point_set):
	""" Averages a set of face points from images
	:param point_set: *n* x *m* x 2 array of face points. \\
	*n* = number of images. *m* = number of face points per image
	"""
	return np.mean(point_set, 0).astype(np.int32)

The version of matplotlib uses requires it to be a uint8 when attempting to save the plot - hence the raised exception.