def test_model():
templates = load_templates()
model = load_model_nn()
x = model['x']
keep_prob = model['keep_prob']
saver = model['saver']
prediction = model['prediction']
graph = model['graph']
model_ckpt_path, _ = find_model_ckpt()
print("Used the model:", model_ckpt_path)
with tf.Session(graph=graph) as session:
tf.global_variables_initializer().run()
saver.restore(session, model_ckpt_path)
dataset, labels = gen_dataset(1, templates) # generate one image
label = prediction.eval(feed_dict={x: dataset, keep_prob: 1.0}, session=session)[0]
print("predict label:", label)
show_im(dataset)
print("actual label:", vec2str(labels[0]))
from gen.gen_captcha import load_templates, create_captcha from model.nn import weight_variable import tensorflow as tf templates = load_templates() c, cs = create_captcha(templates) c.show()
def train():
templates = load_templates()
model = load_model_nn()
x = model["x"]
y = model['y']
loss = model['loss']
optimizer = model['optimizer']
accuracy = model['accuracy']
keep_prob = model['keep_prob']
saver = model['saver']
graph = model['graph']
save_dir = os.path.join(".", '.checkpoint')
print("Model saved path: ", save_dir)
def save_model(_step):
saver.save(
session,
os.path.join(save_dir, 'hjyd-model.ckpt'),
global_step=_step
)
with tf.Session(graph=graph) as session:
merged = tf.summary.merge_all()
writer = tf.summary.FileWriter(graph_log_dir, session.graph)
tf.global_variables_initializer().run()
step = 0
try:
model_ckpt_path, global_step = find_model_ckpt() # try to continue ....
except IOError:
print("Initialized")
else: # try continue to train
saver.restore(session, model_ckpt_path)
step = global_step
print('found %s, step from %d' % (model_ckpt_path, step))
while True:
batch_data, batch_labels = gen_dataset(64, templates)
_, l = session.run(
[optimizer, loss],
feed_dict={
x: batch_data,
y: batch_labels,
keep_prob: 0.75
}
)
step += 1
print(("Step: %d, Loss: %4f" % (step, l)))
if step % 50 == 0:
test_dataset, test_labels = gen_dataset(100, templates)
test_accuracy = session.run(
accuracy,
feed_dict={
x: test_dataset,
y: test_labels,
keep_prob: 1.0
}
)
print(("Step: %d, Test Accuracy: %s" % (step, test_accuracy)))
save_model(step) # save the model every 50 step
if test_accuracy >= 0.92 or step >= 10000: # stop when accuracy above 92%
save_model(step)
break
print("Test accuracy: %g" %
session.run(
accuracy,
feed_dict={
x: test_dataset,
y: test_labels,
keep_prob: 1.0
})
)
# coding=utf-8
from __future__ import print_function
from gen.gen_captcha import gen_dataset, load_templates
import cPickle as pickle
from PIL import Image
import numpy as np
from gen.utils import vec2str
def check_dataset(dataset, labels, index):
data = np.uint8(dataset[index]).reshape((40, 100)) * 255
im = Image.fromarray(data)
im.show()
print("label:", vec2str(labels[index]))
if __name__ == '__main__':
templates = load_templates()
dataset, labels = gen_dataset(1, templates) # generate one image
check_dataset(dataset, labels, 0)