Why is the numpy array immutable after being read using matplotlib.pyplot.imread? What is the reasoning behind that?
>>> import matplotlib
>>> test=matplotlib.pyplot.imread('download.jpeg')
>>> test.flags
C_CONTIGUOUS : True
F_CONTIGUOUS : False
OWNDATA : False
WRITEABLE : False
ALIGNED : True
WRITEBACKIFCOPY : False
UPDATEIFCOPY : False
In terms of what happens in the code, matplotlib.pyplot.imread delegates to Pillow to read the file and then calls np.asarray on the resulting PIL.Image.Image object. When converting a PIL.Image.Image to a NumPy array, the Image object builds a bytestring to use as the array's buffer. Bytestrings are immutable, so the resulting array is unwriteable.
In terms of why anyone on the Matplotlib or Pillow dev team chose an implementation that results in an unwriteable array, I don't know. It's not clear whether this was deliberate at all.
If you want a writeable array, call the array's copy method:
mutable_array = matplotlib.pyplot.imread('download.jpeg').copy()
test.