Schrifterkennung/convert_from_big_dataset.py

29 lines
776 B
Python

import os
from PIL import Image
import numpy as np
directory = "/home/mia/Schule/KISY/schrifterkennung/out"
files = os.listdir(f"{directory}/BigDataSet")
print(files)
labels = []
images = []
for letter in files:
dir = f"{directory}/BigDataSet/{letter}"
i = 0
os.makedirs(f"{directory}/Scaled/{letter}/", exist_ok=True)
for image in os.listdir(dir)[:]:
print(image)
img = Image.open(os.path.join(dir, image))
res = img.resize((32,32))
res = res.convert("L")
images.append(res)
labels.append(letter) # Label ist der Ordnername
res.save(f"{directory}/Scaled/{letter}/miakieler_{i}.png")
i += 1
X = np.array(images)
Y = np.array(labels)
np.save('images_big.npy', X)
np.save('labels_big.npy', Y)