import os
import numpy as np
from tensorflow.keras.utils import to_categorical
from tensorflow.keras.preprocessing.image import img_to_array, load_img
# 多类独热码
def load_dataset_CNN(base_dir, target_size, color_mode='rgb', shuffle=True):
X, y = [], []
total_num = 0
classes = os.listdir(base_dir)
for clas in classes:
label = classes.index(clas)
imgs_dir = os.path.join(base_dir, clas)
imgs_names = os.listdir(imgs_dir)
for img_name in imgs_names:
img_path = os.path.join(imgs_dir, img_name)
img_PIL = load_img(path=img_path, target_size=target_size, color_mode=color_mode)
img_array = img_to_array(img_PIL)
X.append(img_array)
y.append(label)
print("Found {} images in class '{}', setting as class[{}].".format(len(imgs_names), clas, label))
total_num += len(imgs_names)
print('{} images in total.'.format(total_num))
X = np.array(X)
y = to_categorical(np.array(y))
if shuffle is True:
state = np.random.get_state()
np.random.shuffle(X)
np.random.set_state(state)
np.random.shuffle(y)
return X, y
🌰:
from load_dataset import load_dataset_CNN
IMG_HEIGHT, IMG_WIDTH = 100, 100
X, y = load_dataset_CNN(base_dir='./train', target_size=(IMG_HEIGHT, IMG_WIDTH))
print(X)
print(y)
…
👍