I am currently working on a project using GAN. Searching for several models, I found a model called Conditional GAN and found the Face-Aging-with-Identity-Preserved-Conditional-Generative-Adversarial-Networks project on github. https://github.com/dawei6875797/Face-Aging-with-Identity-Preserved-Conditional-Generative-Adversarial-Networks
The Readme.md for this project described how to download the data and how to train the model. Viewing the results using the data provided by this project and the trained model was a success. But I failed to train from the data I wanted. I have tried many things about this, changing the path of the data, etc. but have not been successful. I'm not sure what I missed.
this is my code
import os.path
import os
os.environ['CUDA_DEVICES_ORDER'] = "PCI_BUS_ID"
os.environ["CUDA_VISIBLE_DEVICES"] = '0'
import numpy as np
import tensorflow as tf
from datetime import datetime
from models import FaceAging
import sys
sys.path.append('./tools/')
from source_input import load_source_batch3
from utils import save_images, save_source
from data_generator import ImageDataGenerator
flags = tf.app.flags
flags.DEFINE_float("learning_rate", 0.001, "Learning rate")
flags.DEFINE_integer("batch_size", 32, "The size of batch images")
flags.DEFINE_integer("image_size", 128, "the size of the generated image")
flags.DEFINE_integer("noise_dim", 256, "the length of the noise vector")
flags.DEFINE_integer("feature_size", 128, "image size after stride 2 conv")
flags.DEFINE_integer("age_groups", 5, "the number of different age groups")
flags.DEFINE_integer('max_steps', 200000, 'Number of batches to run')
flags.DEFINE_string("alexnet_pretrained_model", "pre_trained/alexnet.model-292000",
"Directory name to save the checkpoints")
flags.DEFINE_string("age_pretrained_model", "pre_trained/age_classifier.model-300000",
"Directory name to save the checkpoints")
flags.DEFINE_integer('model_index', None, 'the index of trained model')
flags.DEFINE_float("gan_loss_weight", 75, "gan_loss_weight")
flags.DEFINE_float("fea_loss_weight", 0.5e-4, "fea_loss_weight")
flags.DEFINE_float("age_loss_weight", 30, "age_loss_weight")
flags.DEFINE_float("tv_loss_weight", None, "face_loss_weight")
flags.DEFINE_string("checkpoint_dir", "checkpoints/age/0_conv5_lsgan_transfer_g75_0.5f-4_a30", "Directory name to save the checkpoints")
flags.DEFINE_string("source_checkpoint_dir", ' ', "Directory name to save the checkpoints")
flags.DEFINE_string("sample_dir", "age/0_conv5_lsgan_transfer_g75_0.5f-4_a30 ", "Directory name to save the sample images")
flags.DEFINE_string("fea_layer_name", 'conv5', "which layer to use for fea_loss")
flags.DEFINE_string("source_file", "train/", "source file path")
flags.DEFINE_string("root_folder", "CACD_cropped_400/", "folder that contains images")
FLAGS = flags.FLAGS
# How often to run a batch through the validation model.
VAL_INTERVAL = 5000
# How often to save a model checkpoint
SAVE_INTERVAL = 10000
d_iter = 1
g_iter = 1
config = tf.ConfigProto()
config.gpu_options.allow_growth = True
# Initalize the data generator seperately for the training and validation set
train_generator = ImageDataGenerator(batch_size=FLAGS.batch_size, height=FLAGS.feature_size, width=FLAGS.feature_size,
z_dim=FLAGS.noise_dim, scale_size=(FLAGS.image_size, FLAGS.image_size), mode='train')
def my_train():
with tf.Graph().as_default():
sess = tf.Session(config=config)
model = FaceAging(sess=sess, lr=FLAGS.learning_rate, keep_prob=1., model_num=FLAGS.model_index, batch_size=FLAGS.batch_size,
age_loss_weight=FLAGS.age_loss_weight, gan_loss_weight=FLAGS.gan_loss_weight,
fea_loss_weight=FLAGS.fea_loss_weight, tv_loss_weight=FLAGS.tv_loss_weight)
imgs = tf.placeholder(tf.float32, [FLAGS.batch_size, FLAGS.image_size, FLAGS.image_size, 3])
true_label_features_128 = tf.placeholder(tf.float32, [FLAGS.batch_size, 128, 128, FLAGS.age_groups])
true_label_features_64 = tf.placeholder(tf.float32, [FLAGS.batch_size, 64, 64, FLAGS.age_groups])
false_label_features_64 = tf.placeholder(tf.float32, [FLAGS.batch_size, 64, 64, FLAGS.age_groups])
age_label = tf.placeholder(tf.int32, [FLAGS.batch_size])
source_img_227, source_img_128, face_label = load_source_batch3(FLAGS.source_file, FLAGS.root_folder, FLAGS.batch_size)
model.train_age_lsgan_transfer(source_img_227, source_img_128, imgs, true_label_features_128,
true_label_features_64, false_label_features_64, FLAGS.fea_layer_name, age_label)
ge_samples = model.generate_images(imgs, true_label_features_128, reuse=True, mode='train')
# Create a saver.
model.saver = tf.train.Saver(model.save_d_vars + model.save_g_vars, max_to_keep=200)
model.alexnet_saver = tf.train.Saver(model.alexnet_vars)
model.age_saver = tf.train.Saver(model.age_vars)
d_error = model.d_loss/model.gan_loss_weight
g_error = model.g_loss/model.gan_loss_weight
fea_error = model.fea_loss/model.fea_loss_weight
age_error = model.age_loss/model.age_loss_weight
# Start running operations on the Graph.
sess.run(tf.global_variables_initializer())
tf.train.start_queue_runners(sess)
model.alexnet_saver.restore(sess, FLAGS.alexnet_pretrained_model)
model.age_saver.restore(sess, FLAGS.age_pretrained_model)
if model.load(FLAGS.checkpoint_dir, model.saver):
print(" [*] Load SUCCESS")
else:
print(" [!] Load failed...")
print("{} Start training...")
# Loop over max_steps
for step in range(FLAGS.max_steps):
images, t_label_features_128, t_label_features_64, f_label_features_64, age_labels = \
train_generator.next_target_batch_transfer2()
dict = {imgs: images,
true_label_features_128: t_label_features_128,
true_label_features_64: t_label_features_64,
false_label_features_64: f_label_features_64,
age_label: age_labels
}
for i in range(d_iter):
_, d_loss = sess.run([model.d_optim, d_error], feed_dict=dict)
for i in range(g_iter):
_, g_loss, fea_loss, age_loss = sess.run([model.g_optim, g_error, fea_error, age_error],
feed_dict=dict)
format_str = ('%s: step %d, d_loss = %.3f, g_loss = %.3f, fea_loss=%.3f, age_loss=%.3f')
print(format_str % (datetime.now(), step, d_loss, g_loss, fea_loss, age_loss))
# Save the model checkpoint periodically.
if step % SAVE_INTERVAL == SAVE_INTERVAL-1 or (step + 1) == FLAGS.max_steps:
checkpoint_path = os.path.join(FLAGS.checkpoint_dir)
model.save(checkpoint_path, step, 'acgan')
if step % VAL_INTERVAL == VAL_INTERVAL-1:
if not os.path.exists(FLAGS.sample_dir):
os.makedirs(FLAGS.sample_dir)
path = os.path.join(FLAGS.sample_dir, str(step))
if not os.path.exists(path):
os.makedirs(path)
source = sess.run(source_img_128)
save_source(source, [4, 8], os.path.join(path, 'source.jpg'))
for j in range(train_generator.n_classes):
true_label_fea = train_generator.label_features_128[j]
dict = {
imgs: source,
true_label_features_128: true_label_fea
}
samples = sess.run(ge_samples, feed_dict=dict)
save_images(samples, [4, 8], './{}/test_{:01d}.jpg'.format(path, j))
def main(argv=None):
my_train()
if __name__ == '__main__':
tf.app.run()
this is error msg
Traceback (most recent call last):
File "C:/Users/admin/Desktop/Face/Face-Aging-with-Identity-Preserved-Conditional-Generative-Adversarial-Networks-master/age_lsgan_transfer.py", line 167, in <module>
tf.app.run()
File "C:\Users\admin\PycharmProjects\Deep\venv\lib\site-packages\tensorflow\python\platform\app.py", line 40, in run
_run(main=main, argv=argv, flags_parser=_parse_flags_tolerate_undef)
File "C:\Users\admin\PycharmProjects\Deep\venv\lib\site-packages\absl\app.py", line 300, in run
_run_main(main, args)
File "C:\Users\admin\PycharmProjects\Deep\venv\lib\site-packages\absl\app.py", line 251, in _run_main
sys.exit(main(argv))
File "C:/Users/admin/Desktop/Face/Face-Aging-with-Identity-Preserved-Conditional-Generative-Adversarial-Networks-master/age_lsgan_transfer.py", line 163, in main
my_train()
File "C:/Users/admin/Desktop/Face/Face-Aging-with-Identity-Preserved-Conditional-Generative-Adversarial-Networks-master/age_lsgan_transfer.py", line 88, in my_train
source_img_227, source_img_128, face_label = load_source_batch3(FLAGS.source_file, FLAGS.root_folder, FLAGS.batch_size)
File "C:\Users\admin\Desktop\Face\Face-Aging-with-Identity-Preserved-Conditional-Generative-Adversarial-Networks-master\source_input.py", line 128, in load_source_batch3
img_list, label_list = get_imgAndlabel_list2(filename, img_folder)
File "C:\Users\admin\Desktop\Face\Face-Aging-with-Identity-Preserved-Conditional-Generative-Adversarial-Networks-master\source_input.py", line 173, in get_imgAndlabel_list2
f = open(filename, 'r')
PermissionError: [Errno 13] Permission denied: 'train/'
this code is source_input.py
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import argparse
import os
import sys
import tensorflow as tf
#from read_image import *
from tensorflow.python.platform import gfile
from tensorflow.python.platform import flags
import numpy as np
import scipy.io as scio
from tensorflow.python.framework import ops
from PIL import Image
FLAGS = flags.FLAGS
T = 1
IM_HEIGHT = 400
IM_WIDTH = 400
IM_CHANNELS = 3
def _int64_feature(value):
return tf.train.Feature(int64_list=tf.train.Int64List(value=[value]))
def _bytes_feature(value):
return tf.train.Feature(bytes_list=tf.train.BytesList(value=[value]))
def read_images(filename_queue, new_height=None, new_width=None):
reader = tf.WholeFileReader()
key, value = reader.read(filename_queue)
image = tf.image.decode_jpeg(value) # use png or jpeg decoder based on your files
image = tf.reshape(image, [IM_HEIGHT, IM_WIDTH, IM_CHANNELS])
if new_height and new_width:
image = tf.image.resize_images(image, [new_height, new_width])
image = tf.cast(image, tf.float32) - np.array([104., 117., 124.])
return image
def read_images2(filename_queue):
reader = tf.WholeFileReader()
key, value = reader.read(filename_queue)
image = tf.image.decode_jpeg(value) # use png or jpeg decoder based on your files
image = tf.reshape(image, [IM_HEIGHT, IM_WIDTH, IM_CHANNELS])
image_227 = tf.image.resize_images(image, [227, 227])
image_227 = tf.cast(image_227, tf.float32) - np.array([104., 117., 124.])
image_128 = tf.image.resize_images(image, [128, 128])
image_128 = tf.cast(image_128, tf.float32) - np.array([104., 117., 124.])
return image_227, image_128
def read_images3(input_queue):
label = input_queue[1]
file_contents = tf.read_file(input_queue[0])
image = tf.image.decode_image(file_contents, channels=3)
image = tf.reshape(image, [IM_HEIGHT, IM_WIDTH, IM_CHANNELS])
image_227 = tf.image.resize_images(image, [227, 227])
image_227 = tf.cast(image_227, tf.float32) - np.array([104., 117., 124.])
image_128 = tf.image.resize_images(image, [128, 128])
# image_128 = tf.cast(image_128, tf.float32)
image_128 = tf.cast(image_128, tf.float32) - np.array([104., 117., 124.])
return image_227, image_128, label
def load_source_batch(filename, img_folder, batch_size, img_size, shuffle=True):
filenames = get_imgAndlabel_list(filename, img_folder)
print('%d images to train' %(len(filenames)))
if not filenames:
raise RuntimeError('No data files found.')
with tf.name_scope('input'):
filename_queue = tf.train.string_input_producer(filenames, shuffle=shuffle)
# Even when reading in multiple threads, share the filename queue.
image = read_images(filename_queue, new_height=img_size, new_width=img_size)
image_batch = tf.train.shuffle_batch(
[image],
batch_size=batch_size,
num_threads=4,
capacity=1280,
min_after_dequeue=640)
# image_batch = tf.train.batch(
# [image],
# batch_size=batch_size,
# num_threads=4,
# capacity=1280)
#
return image_batch
def load_source_batch2(filename, img_folder, batch_size, shuffle=True):
filenames = get_imgAndlabel_list(filename, img_folder)
print('%d images to train' % (len(filenames)))
if not filenames:
raise RuntimeError('No data files found.')
with tf.name_scope('input'):
filename_queue = tf.train.string_input_producer(filenames, shuffle=shuffle)
# Even when reading in multiple threads, share the filename queue.
image_227, image_128 = read_images2(filename_queue)
image_227_batch, image_128_batch = tf.train.shuffle_batch(
[image_227, image_128],
batch_size=batch_size,
num_threads=4,
capacity=1280,
min_after_dequeue=640)
return image_227_batch, image_128_batch
def load_source_batch3(filename, img_folder, batch_size, shuffle=True):
img_list, label_list = get_imgAndlabel_list2(filename, img_folder)
print('%d images to train' % (len(img_list)))
images = ops.convert_to_tensor(img_list, dtype=tf.string)
labels = ops.convert_to_tensor(label_list, dtype=tf.int32)
# Makes an input queue
input_queue = tf.train.slice_input_producer([images, labels], shuffle=shuffle)
# Even when reading in multiple threads, share the filename queue.
image_227, image_128, label = read_images3(input_queue)
image_227_batch, image_128_batch, label_batch = tf.train.shuffle_batch(
[image_227, image_128, label],
batch_size=batch_size,
num_threads=4,
capacity=1280,
min_after_dequeue=640)
return image_227_batch, image_128_batch, label_batch
def get_imgAndlabel_list(filename, img_folder):
"""
:param filename:
each line in filename is img_name \space label
:return:
img names list
label list
"""
f = open(filename, 'r')
lines = f.readlines()
f.close()
imgname_lists = []
for i in range(len(lines)):
img_name = lines[i].split()[0]
imgname_lists.append(os.path.join(img_folder, img_name))
return imgname_lists
def get_imgAndlabel_list2(filename, img_folder):
"""
:param filename:
each line in filename is img_name \space label
:return:
img names list
label list
"""
f = open(filename, 'r')
lines = f.readlines()
f.close()
imgname_lists = []
label_lists = []
for i in range(len(lines)):
img_name, label = lines[i].split()
imgname_lists.append(os.path.join(img_folder, img_name))
label_lists.append(int(label))
return imgname_lists, label_lists
