def test(features_path, save_path, test_features, csv_path, name):
model = Classifier(len(test_features))
n_batches = np.load(os.path.join(features_path, 'number.npy'))[0]
graph = tf.Graph()
with graph.as_default():
model.build(is_training=False)
saver = tf.train.Saver()
model_path = os.path.join(save_path, name, 'catdog.ckpt')
results = {}
count = 0
with tf.Session(graph=graph) as sess:
saver.restore(sess, model_path)
for i in range(n_batches):
features = None
labels = None
for test_feature in test_features:
feature_path = os.path.join(features_path, test_feature,
'batch_{}'.format(i))
feature = np.load(os.path.join(feature_path, 'features.npy'))
if labels is None:
labels = np.load(os.path.join(feature_path, 'labels.npy'))
if features is None:
features = feature
else:
features = np.hstack((features, feature))
prediction = sess.run(model.prediction,
feed_dict={model.inputs: features})
for j, label in enumerate(labels):
results[label] = prediction[j]
count += 1
print('Processed {} images'.format(count))
keys = sorted(results.keys())
with open(os.path.join(csv_path, name + '.csv'), 'w', newline='') as f:
writer = csv.writer(f)
writer.writerow(['id', 'label'])
for key in keys:
writer.writerow([key, results[key]])
image_URL = "../data/valid/handwritten/00002.png"
class_dict = {0: "handwritten",
1: "printed"}
model = Classifier()
if phase == "train":
model.compile(
loss="binary_crossentropy",
optimizer="adam",
metrics=["accuracy"]
)
model.build((None,) + image_size)
model.summary()
train_loader, val_loader = get_loaders(data_dir, "train", image_size[:2], batch_size)
model.fit(
train_loader,
steps_per_epoch=train_loader.samples // batch_size,
epochs=5,
validation_data=val_loader,
validation_steps=val_loader.samples // batch_size
)
model.save_weights("../printed_and_handwritten_classifier.h5")