内存错误将大RGB图像转换为灰度

Question

I am trying to load original color fundus images in python for some learning project. These are RGB images of sizes approx. 4000*2000*3. I want to convert each image to grayscale before sending them for training into my model. For this, I've tried the following 2 approaches:

Method 1:

import matplotlib.image as mpimg

def rgb2gray(rgb):
    return np.dot(rgb[...,:3], [0.299, 0.587, 0.114])

def load_images(folder):
    images = []
    for filename in os.listdir(folder):
        img = mpimg.imread(os.path.join(folder, filename))
        gray = rgb2gray(img)
        images.append(gray)
    return images

Method 2:

from PIL import Image

def load_images(folder):
    images = []
    for filename in os.listdir(folder):
        img = Image.open(os.path.join(folder, filename))
        gray = img.convert('L')
        images.append(gray)
    return images

In both the methods, I get the same Memory Error at lines gray = rgb2gray(img) and gray = img.convert('L') respectively.

Is there any way I can convert such large images to grayscale one at a time and store it as a list?

Answer 1

Turns out that the problem is not with the commands but with the size of my data. As @user894763 correctly pointed out, my size of data amounts upto 10 GB which is the cause of Memory Error.