def main(_):
resized_width = 64
resized_height = 64
since = time.time()
image_lists = data_process.create_image_lists(FLAGS.images_dir)
test_datas, test_labels = data_process.get_batch_of_data(
image_lists, -1, FLAGS.images_dir, 'test', resized_width,
resized_height)
train_datas, train_labels = data_process.get_batch_of_data(
image_lists, -1, FLAGS.images_dir, 'train', resized_width,
resized_height)
print(train_datas.shape)
print(train_labels.shape)
model = svm.SVC(kernel='rbf', verbose=True, probability=True)
model.fit(train_datas, np.argmax(train_labels, axis=1))
pred_prob = model.predict_proba(test_datas)
test_acc = np.mean(
np.equal(np.argmax(pred_prob, axis=1), np.argmax(test_labels, axis=1)))
print('Test accuracy:', test_acc)
save_model_path = os.path.join(FLAGS.path, "save_model", "SVM",
"Nonlinear_SVM.pickle")
if not os.path.exists(save_model_path):
os.makedirs(save_model_path)
with open(save_model_path, 'wb') as f:
pickle.dump(model, f)
time_elapsed = time.time() - since
print("Runtime: {}min, {:0.2f}sec".format(int(time_elapsed // 60),
time_elapsed % 60))
save_results_path = os.path.join(FLAGS.path, "results/SVM", "results.txt")
if not os.path.exists(save_results_path):
os.makedirs(save_results_path)
with open(save_results_path, "w") as f:
f.write("Test accuracy = {}".format(test_acc) + "\n")
f.write("Runtime: " + str(int(time_elapsed // 60)) + "min," +
str(time_elapsed % 60) + "sec")
def predict(path, images_dir, resized_width, resized_height):
image_lists = data_process.create_image_lists(images_dir)
test_datas, test_labels = data_process.get_batch_of_data(
image_lists, -1, images_dir, 'test', resized_width, resized_height)
save_model_path = os.path.join(FLAGS.path, "save_model", "SVM",
"Nonlinear_SVM.pickle")
with open(save_model_path, 'rb') as f:
model = pickle.loads(f.read())
pred_prob = model.predict_proba(test_datas)
test_acc = np.mean(
np.equal(np.argmax(pred_prob, axis=1), np.argmax(test_labels, axis=1)))
print('Test accuracy:', test_acc)
pre_data_dir = os.path.join("../../predictData/SVM")
if not os.path.exists(pre_data_dir):
os.makedirs(pre_data_dir)
with h5py.File(os.path.join(pre_data_dir, "prediction_and_labels.h5"),
"w") as f:
f["prediction"] = pred_prob
f["truth"] = test_labels
def main(_):
resized_width = 128
resized_height = 128
model = build_CNN()
opt = Adam(lr=FLAGS.learning_rate)
model.compile(loss="categorical_crossentropy",
optimizer=opt,
metrics=["accuracy"])
model.load_weights(os.path.join(FLAGS.path, "save_model", "CNN.h5df"))
image_lists = data_process.create_image_lists(FLAGS.images_dir)
test_datas, test_labels = data_process.get_batch_of_data(
image_lists, -1, FLAGS.images_dir, "test", resized_width,
resized_height)
test_loss, test_acc = model.evaluate(test_datas, test_labels)
print("Test accuracy:{0:.4f}, test loss:{1:.4f}".format(
test_acc, test_loss))
prediction = model.predict(test_datas)
pre_data_dir = "../../predictData/CNN"
if not os.path.exists(pre_data_dir):
os.makedirs(pre_data_dir)
with h5py.File(os.path.join(pre_data_dir, "prediction_and_labels.h5"),
"w") as f:
f["prediction"] = prediction
f["truth"] = test_labels
def generate_train_data(image_lists, images_dir, batch_size, resized_width,
resized_height):
while (True):
train_datas, train_labels = data_process.get_batch_of_data(
image_lists, batch_size, images_dir, "train", resized_width,
resized_height)
yield (train_datas, train_labels)
def predict(path, images_dir, tree_num, resized_width, resized_height):
X = tf.placeholder(tf.float32, shape=[None, 128 * 128 * 3])
Y = tf.placeholder(tf.float32, shape=[None])
image_lists = data_process.create_image_lists(images_dir)
train_step, randomforest_loss, accuracy, output = randomforest(
X, Y, 128 * 128 * 3, 6, tree_num)
saver = tf.train.Saver()
init = tf.group(
tf.global_variables_initializer(),
resources.initialize_resources(resources.shared_resources()))
with tf.Session(config=config) as sess:
sess.run(init)
ckpt = tf.train.latest_checkpoint(os.path.join(path, 'save_model'))
print(ckpt)
saver.restore(sess, ckpt)
test_datas, test_labels = data_process.get_batch_of_data(
image_lists, -1, images_dir, 'test', resized_width, resized_height)
labels = np.argmax(test_labels, axis=1)
print(test_datas.shape)
print(test_labels.shape)
print(labels.shape)
test_acc, prediction = sess.run([accuracy, output],
feed_dict={
X: test_datas,
Y: labels
})
print('Test accuracy:{0:.4f}'.format(test_acc))
pre_data_dir = os.path.join("../../predictData/RandomForest")
if not os.path.exists(pre_data_dir):
os.makedirs(pre_data_dir)
with h5py.File(os.path.join(pre_data_dir, "prediction_and_labels.h5"),
"w") as f:
f["prediction"] = prediction
f["truth"] = test_labels
def main(_):
resized_width = 128
resized_height = 128
model = build_CNN()
opt = Adam(lr=FLAGS.learning_rate)
model.compile(loss="categorical_crossentropy",
optimizer=opt,
metrics=["accuracy"])
model.load_weights(os.path.join(FLAGS.path,"weights/weights/", "model.h5df"))
image_lists = data_process.create_image_lists(FLAGS.images_dir)
test_datas, test_labels = data_process.get_batch_of_data(
image_lists, -1, FLAGS.images_dir, "test", resized_width, resized_height)
test_loss, test_acc = model.evaluate(test_datas, test_labels)
print("Test accuracy:{0:.4f}, test loss:{1:.4f}".format(test_acc, test_loss))
def main(_):
resized_width = 128
resized_height = 128
since = time.time()
model = build_CNN()
model.summary()
opt = Adam(lr=FLAGS.learning_rate)
model.compile(loss="categorical_crossentropy",
optimizer=opt,
metrics=["accuracy"])
result = get_result()
checkpoint_path = os.path.join(FLAGS.path, "save_model", "CNN.h5df")
checkpoint = ModelCheckpoint(filepath=checkpoint_path,
save_best_only=True,
save_weights_only=True,
monitor="val_acc",
mode=max)
tb = TensorBoard(log_dir=os.path.join(FLAGS.path, "results/results/logs"))
callbacks = [result, checkpoint, tb]
image_lists = data_process.create_image_lists(FLAGS.images_dir)
with gfile.FastGFile(os.path.join(FLAGS.path, "results/output_labels.txt"),
"w") as f:
f.write("\n".join(image_lists.keys()) + "\n")
val_datas, val_labels = data_process.get_batch_of_data(
image_lists, -1, FLAGS.images_dir, "val", resized_width,
resized_height)
model.fit_generator(generate_train_data(image_lists, FLAGS.images_dir,
FLAGS.batch_size, resized_width,
resized_height),
epochs=FLAGS.epochs,
steps_per_epoch=100,
validation_data=(val_datas, val_labels),
callbacks=callbacks)
test_datas, test_labels = data_process.get_batch_of_data(
image_lists, -1, FLAGS.images_dir, "test", resized_width,
resized_height)
test_loss1, test_acc1 = model.evaluate(test_datas, test_labels)
print("Test accuracy:{0:.4f}, test loss:{1:.4f}".format(
test_acc1, test_loss1))
model.load_weights(os.path.join(FLAGS.path, "save_model", "CNN.h5df"))
test_loss2, test_acc2 = model.evaluate(test_datas, test_labels)
time_elapsed = time.time() - since
print("Test accuracy with best validation = {}".format(test_acc2 * 100))
print("Final test accuracy = {}".format(test_acc1 * 100))
print("Total Model Runtime: {}min, {:0.2f}sec".format(
int(time_elapsed // 60), time_elapsed % 60))
with open(os.path.join(FLAGS.path, "results/results/results.txt"),
"w") as f:
f.write("Test accuracy with best validation:" + str(test_acc2) + "\n")
f.write("Final test accuracy: " + str(test_acc1) + "\n")
f.write("Total Model Runtime: " + str(int(time_elapsed // 60)) +
"min," + str(time_elapsed % 60) + "sec")
def main(_):
resized_width = 128
resized_height = 128
since = time.time()
X = tf.placeholder(tf.float32, shape=[None, 128 * 128 * 3])
Y = tf.placeholder(tf.float32, shape=[None])
image_lists = data_process.create_image_lists(FLAGS.images_dir)
class_count = len(image_lists.keys())
train_step, randomforest_loss, accuracy = randomforest(
X, Y, 128 * 128 * 3, class_count, FLAGS.tree_num)
saver = tf.train.Saver()
save_model_path = os.path.join(FLAGS.path, "save_model", "RandomForest")
if not os.path.exists(save_model_path):
os.makedirs(save_model_path)
save_results_path = os.path.join(FLAGS.path, "results", "results.txt")
if not os.path.exists(save_results_path):
os.makedirs(save_results_path)
init = tf.group(
tf.global_variables_initializer(),
resources.initialize_resources(resources.shared_resources()))
with tf.Session(config=config) as sess:
sess.run(init)
merged = tf.summary.merge_all()
train_writer = tf.summary.FileWriter(FLAGS.path + "results/logs/train",
sess.graph)
validation_writer = tf.summary.FileWriter(FLAGS.path +
"results/logs/validation")
val_datas, val_labels = data_process.get_batch_of_data(
image_lists, -1, FLAGS.images_dir, "val", resized_width,
resized_height)
val_labels = np.argmax(val_labels, axis=1)
test_datas, test_labels = data_process.get_batch_of_data(
image_lists, -1, FLAGS.images_dir, "test", resized_width,
resized_height)
test_labels = np.argmax(test_labels, axis=1)
best_acc = 0.0
for i in range(FLAGS.iters):
train_datas, train_labels = data_process.get_batch_of_data(
image_lists, FLAGS.batch_size, FLAGS.images_dir, "train",
resized_width, resized_height)
train_labels = np.argmax(train_labels, axis=1)
sess.run(train_step, feed_dict={X: train_datas, Y: train_labels})
if (i % 10 == 0):
train_loss, train_acc = sess.run([randomforest_loss, accuracy],
feed_dict={
X: train_datas,
Y: train_labels
})
train_summary = sess.run(merged,
feed_dict={
X: train_datas,
Y: train_labels
})
val_loss, val_acc = sess.run([randomforest_loss, accuracy],
feed_dict={
X: val_datas,
Y: val_labels
})
val_summary = sess.run(merged,
feed_dict={
X: val_datas,
Y: val_labels
})
train_writer.add_summary(train_summary, i)
validation_writer.add_summary(val_summary, i)
if (val_acc > best_acc):
best_acc = val_acc
test_loss1, test_acc1 = sess.run(
[randomforest_loss, accuracy],
feed_dict={
X: test_datas,
Y: test_labels
})
saver = tf.train.Saver()
saver.save(sess, save_model_path)
print(
"Iteration {0:d}: train loss:{1:f}, train acc:{2:f}, val loss:{3:f}, val acc:{4:f}"
.format(i, train_loss, train_acc, val_loss, val_acc))
test_loss2, test_acc2 = sess.run([randomforest_loss, accuracy],
feed_dict={
X: test_datas,
Y: test_labels
})
print("Best validation accuracy = {}".format(best_acc * 100))
print("Test accuracy with best validation = {}".format(test_acc1 *
100))
print("Final test accuracy = {}".format(test_acc2 * 100))
time_elapsed = time.time() - since
print("Runtime: {}min, {:0.2f}sec".format(int(time_elapsed // 60),
time_elapsed % 60))
with open(save_results_path, "w") as f:
f.write("Best validation accuracy: " + str(best_acc) + "\n")
f.write("Test accuracy with best validation: " + str(test_acc1) + "\n")
f.write("Final test accuracy = {}".format(test_acc2 * 100) + "\n")
f.write("Runtime: " + str(int(time_elapsed // 60)) + "min," +
str(time_elapsed % 60) + "sec")
