determine whether the photo of surveillance camera is grayscale

ifGrayImg.py

'''
Some surveillance camera has 2 modes, RGB and Infrared. However, 
the some frames under infrared mode is not completely "gray", they
are still 3 channels and value of three channels are not equal.
This function accepts image read by opencv(3-d numpy.ndarray), and
method ('rgb' or 'hsv'). 'rgb': consider 'gray' means r==g==b of
one pixel, smaller difference means geayer. 'hsv': consider gray
image is almost black in saturation channel.
'''
import cv2
import numpy as np

# img is numpy.ndarray read by opencv
def isGrayImg(img, method='rgb'):
    if method =='rgb':
        # for one pixel [r,g,b], diff=(r-g)+(r-b)+(g-b)
        diff0 = img[:,:,0]-img[:,:,1]
        diff1 = img[:,:,2]-img[:,:,1]
        diff2 = img[:,:,0]-img[:,:,2]
        count = np.count_nonzero(diff0)
        count += np.count_nonzero(diff1)
        count += np.count_nonzero(diff2)

        h,w,c = img.shape
        percent = count/(h*w*c)
        return percent < 0.2 # 20% is threshold percentage (0~100%)
    
    elif method == 'hsv':
        imghsv = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)
        h,w,c = img.shape
        # low saturation means gray image
        avg_saturation = imghsv[:,:,1].sum()/(h*w)
        return avg_saturation < 10 # 10 is threshold value (0~255)
    
path = 'path/to/image'
test_img = cv2.imread(path)
print(f'Image {path} is{isGrayImg(test_img, mode='hsv') ? ' ' : ' not '}gray image')