read_images.py

import os
import numpy as np
from PIL import Image
import h5py
import cv2

# Global variables
MALE = np.arange(0, 5)
FEMALE = np.arange(5, 10)

# Folder path
PATH = "MSFDE"
FILE_FORMAT = (".tif", ".jpg")

# Get first three digits
def getImageId(name):
	return name[:3]

images = []
imagesResized = []
sex = []
ethnic = []
emotion = []

for subdir, dirs, files in os.walk(PATH):
	for file in files:
		if file.endswith(FILE_FORMAT):
			name = os.path.join(subdir, file)
			im = cv2.imread(name, cv2.IMREAD_COLOR)
			im = cv2.cvtColor(im, cv2.COLOR_BGR2GRAY)
			im = cv2.cvtColor(im, cv2.COLOR_GRAY2RGB)
			
			# im.show()
			images.append(np.array(im))

			im = cv2.resize(im, (224, 224))
			imagesResized.append(np.array(im))
			
			imageId = getImageId(file)
			if int(imageId[1]) in MALE:
				sex.append(1)
			else:
				sex.append(0)
			ethnic.append(int(imageId[0]) - 1)
			if (imageId[2].isdigit()):
				emotion.append(int(imageId[2]))
			else:
				emotion.append(0)


# Concatenate
images = np.float64(np.stack(images))
print(images.shape)
imagesResized = np.float64(np.stack(imagesResized))
sex = np.stack(sex)
ethnic = np.stack(ethnic)
emotion = np.stack(emotion)

			
	
# Normalize data
images /= 255.0
imagesResized /= 255.0
# Save to disk
f = h5py.File("images.h5", "w")
# Create dataset to store images
X_dset = f.create_dataset('data', images.shape, dtype='f')
X_dset[:] = images
X_dset = f.create_dataset('dataResized', imagesResized.shape, dtype='f')
X_dset[:] = imagesResized

# Create dataset to store labels
y_dset = f.create_dataset('sex', sex.shape, dtype='i')
y_dset[:] = sex
y_dset = f.create_dataset('ethnic', ethnic.shape, dtype='i')
y_dset[:] = ethnic
y_dset = f.create_dataset('emotion', emotion.shape, dtype='i')
y_dset[:] = emotion
    
f.close()