How to check the channel order of an image?

Question

Question

With an image loaded into Python as shown below, how do I know which order the channels are in? (e.g. BGR or RGB)

Code

from PIL import Image
import numpy as np

image_pil = Image.open("Stonehenge.jpg")
image_np = np.array(image_pil)
image_np[0][0]

Result

array([ 52, 123, 155], dtype=uint8)

Specific question

How do I know whether the 52 corresponds to the red channel, the blue channel, or a different channel? Or does this question not make sense on a conceptual level?

---

Notes

In a similar question for Java instead of Python, one person claims:

If you are reading in the image file, or you have access to the code that reads in the file, know it is:

• BGR order if you used cv2.imread(),
• RGB order if you used mpimg.imread(), (assuming import matplotlib.image as mpimg)

If you don't know how the file was opened, the accepted answer BufferedImage is great for Java.

Answer 1

Since you use PIL and you don't specify any other mode to load the Image with, you get R G B.

You could verify that by checking the "mode" attribute on the Image instance:

image_pil.mode   # should return the string 'RGB'

Pillow supports the array interface, via image_pil.__array_interface__ magic method, so when when you create the ndarray numpy just uses that. i.e., it doesn't know anything about the colour channel order. If you have an image file stored as BGR, and you load it like this, you will get blue data in the red channel and vice-versa, and it would look wrong when you display it.