How do I convert data type of images from uint8 to torch.float

Question

I have multiple folders which have PNG images in it. There datatype is of uint8. I ty to run a for loop but it loops only through the array values of 1st image.

I tried this code:

import torch
for i in range(NUMBER OF IMAGES):
    image = cv2.imread(f'DIR/FOLDER/{i}')
    image = torch.to_numpy(image[i]).to(dtype = torch)

Answer 1

According to my understanding it seems you want to get the images from a directory where the images are named like 1.png, 2.png and then change the type to torch.float, then:

import cv2
import numpy as np
import torch

directory = '/DIR/FOLDER/'
total_imgs = 10

# Initializing an empty list to store the images
images = []

# Iterating over the images
for i in range(0, total_imgs):
    # I've used i+1 to start from 1 as the image would be 1.png
    file_path = directory + str(i+1) + '.png'

    # Reading the image
    image = cv2.imread(file_path)

    image_uint8 = np.array(image, dtype=np.uint8)

    # Converting the image to torch.float
    image_float = torch.from_numpy(image_uint8.astype(np.float32))

    # Appending the converted image to the list
    images.append(image_float)

Hope that helps.