How can I segment already augmented images into smaller ones?

Question

I have been trying to figure out a way to segment a large already annotated image which I have the .json for into smaller images with python for the deep learning model as the image is very high in resolution and I need it to be 256x256, how can this be done?

Answer 1

If you read image with cv2 then you have numpy.array and you can use indexes

img[ row_start:row_end, col_start:col_end ]

to get subimage - ie. img[0:256,0:256]

And you can repeate it in for-loops like

all_small = []

for row in range(0, height, 256):
    for col in range(0, width, 256):
        small = img[ row:row+256, col:col+256 ]
        all_small.append(small)