recyclebot/train.py at master · kashiish/recyclebot · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
import tensorflow as tf
import tensorflow_hub as hub
from tensorflow.keras import layers

import numpy as np
import os
import time

train_dataset_dir = 'images/train'
val_dataset_dir = 'images/validation'

batch_size = 8
image_res = 224

labels = np.array(sorted(os.walk(train_dataset_dir).next()[1]))


URL = "https://tfhub.dev/google/tf2-preview/mobilenet_v2/feature_vector/2"
feature_extractor = hub.KerasLayer(URL, input_shape=(image_res, image_res,3))


def get_images(dataset_dir):
	#list of image paths and the images' corresponding label ids
	image_names, image_label_ids = [], []

	label_id = 0

	for label in labels:
		label_dir = os.path.join(dataset_dir, label)

		image_paths = os.walk(label_dir).next()


		for image_path in image_paths[2]:
			if image_path.endswith('jpg') or image_path.endswith('jpeg'):
				image_names.append(os.path.join(label_dir, image_path))
				image_label_ids.append(label_id)

		label_id += 1

	return image_names, image_label_ids


def parser(image_name, label_id):
	decoded_image = tf.image.decode_jpeg(tf.io.read_file(image_name), channels=3)
	image = tf.cast(decoded_image, tf.float32)
	image = tf.image.resize(image, (image_res, image_res))/255.0
	return image, label_id


def save_model(model):
	t = time.time()

	export_path_sm = "./{}".format(int(t))
	print(export_path_sm)

	tf.saved_model.save(model, export_path_sm)


training_images, training_labels = get_images(train_dataset_dir)

training_dataset = tf.data.Dataset.from_tensor_slices((tf.constant(training_images), tf.constant(training_labels)))
training_dataset = training_dataset.map(parser)
training_dataset = training_dataset.shuffle(len(training_images))
training_dataset = training_dataset.batch(batch_size)


val_images, val_labels = get_images(val_dataset_dir)

val_dataset = tf.data.Dataset.from_tensor_slices((tf.constant(val_images), tf.constant(val_labels)))
val_dataset = val_dataset.map(parser)
val_dataset = val_dataset.shuffle(len(val_images))
val_dataset = val_dataset.batch(batch_size)


feature_extractor.trainable = False

model = tf.keras.Sequential([
  feature_extractor,
  layers.Dense(4, activation='softmax')
])

model.summary()

model.compile(
  optimizer='adam',
  loss=tf.losses.SparseCategoricalCrossentropy(),
  metrics=['accuracy'])

EPOCHS = 6
history = model.fit(training_dataset,
                    epochs=EPOCHS,
                    validation_data=val_dataset)

save_model(model)

# image_batch, label_batch = next(iter(training_dataset.take(1)))
# image_batch = image_batch.numpy()
# label_batch = label_batch.numpy()

# predicted_batch = model.predict(image_batch)
# predicted_batch = tf.squeeze(predicted_batch).numpy()
# predicted_ids = np.argmax(predicted_batch, axis=-1)
# predicted_class_names = labels[predicted_ids]
# print(predicted_class_names)

# print("Labels: ", label_batch)
# print("Predicted labels: ", predicted_ids)


#RELOAD
reload_sm_keras = tf.keras.models.load_model(
  '1578826768',
  custom_objects={'KerasLayer': hub.KerasLayer})


# img_name = val_dataset_dir + '/paper/19774.jpg'
# img, label = parser(img_name, 3)

# img = np.expand_dims(img, axis=0)

# predict = reload_sm_keras.predict(img)
# predict =  tf.squeeze(predict).numpy()
# predict_id = np.argmax(predict, axis=-1)
# predicted_class_name = labels[predict_id]
# print(predicted_class_name)

EPOCHS = 3
history = reload_sm_keras.fit(training_dataset,
                    epochs=EPOCHS,
                    validation_data=val_dataset)

save_model(reload_sm_keras)

# image_batch, label_batch = next(iter(training_dataset.take(1)))
# image_batch = image_batch.numpy()
# label_batch = label_batch.numpy()

# predicted_batch = reload_sm_keras.predict(image_batch)
# predicted_batch = tf.squeeze(predicted_batch).numpy()
# predicted_ids = np.argmax(predicted_batch, axis=-1)
# predicted_class_names = labels[predicted_ids]
# print(predicted_class_names)

# print("Labels: ", label_batch)
# print("Predicted labels: ", predicted_ids)