# plt.axis('off')
# plt.show()
with tf.Graph().as_default() as g:
# gray_images = []
# for image in face_images:
# grayed_image = tf.image.rgb_to_grayscale(face_image)
# gray_images.append(grayed_image)
images = np.reshape(face_images, [len(faces), 42, 42, 1])
# images = tf.pack(face_images)
# images = tf.cast(tf.reshape(images, [len(faces), 42, 42, 1]), tf.float32)
# print(images)
with tf.Session() as sess:
cnn = CNN.CNN_net(False)
sess.run(tf.global_variables_initializer())
saver = tf.train.Saver(max_to_keep=1)
model_file = tf.train.latest_checkpoint('ckpt1/')
saver.restore(sess, model_file)
feed_dict_test = {cnn.images: images}
labels = sess.run(cnn.logits, feed_dict=feed_dict_test)
# print(labels)
im = Image.open(image_path)
for i in range(len(face_images)):
emotion = emotions[np.where(labels[i] == np.max(labels[i]))[0][0]]
import tensorflow as tf
import datetime
from timer import Timer
from fer2013 import Fer2013
import numpy as np
import CNN
# img1 = utils.load_image("./test_data/tiger.jpeg")
# img1_true_result = [1 if i == 292 else 0 for i in range(7)] # 1-hot result for tiger
# batch1 = img1.reshape((1, 224, 224, 3))
with tf.Session() as sess:
cnn = CNN.CNN_net()
sess.run(tf.global_variables_initializer())
saver = tf.train.Saver(max_to_keep=1)
train_timer = Timer()
load_timer = Timer()
fer2013 = Fer2013('train')
fer2013_test = Fer2013('val')
max_iter = 100000
summary_iter = 10
epoch = 0
for step in range(1, max_iter + 1):
load_timer.tic()
