Given a target aspect ratio, how to crop into image, framing around a found face

I have 800x800 portrait image and want to crop into it, finding the face and "framing it" in the centre, while maintaining a 3:4 aspect ratio (outline select box).

So I almost need to calculate some kind of "centre point" of the calculated top, bottom, left and right and crop into that point (red dots).

I believe I'm someway off reaching this goal as my python isn't particularly good - here's what I have so far:

cropped_face_image = numpy.array(PIL.Image.open(BytesIO(image.content)))[
    top:bottom, left:right
]
pil_image = PIL.Image.fromarray(cropped_face_image)
pil_image.show()

This just hard-crops the face from the image.

Solution

You could do something like this:

from PIL import Image
im = Image.open('FlKyDm.png.jpeg')
top, bottom, left, right = 95, 177 , 70, 150

# Calculate centre of face and box
centreX, centreY = int((left+right)/2), int((top+bottom)/2)

# Calculate face width
faceW = right - left

# Calculate half width of box and half height of box
boxHalfW = faceW
boxHalfH = int((boxHalfW * 4) / 3)

# Crop accordingly
res = im.crop((centreX-boxHalfW, centreY-boxHalfH, centreX+boxHalfW, centreY+boxHalfH))
res.save('result.jpg')