1.7 MiB
1.7 MiB
import matplotlib.pyplot as plt
import numpy as np
import os
import PIL
import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers
from tensorflow.keras.models import Sequentialtf.config.list_physical_devices('GPU')[]
import cv2
import imghdr
import os
data_dir = 'trash/validate'
image_exts = ['jpeg','jpg', 'bmp', 'png']
for image_class in os.listdir(data_dir):
for image in os.listdir(os.path.join(data_dir, image_class)):
image_path = os.path.join(data_dir, image_class, image)
try:
img = cv2.imread(image_path)
tip = imghdr.what(image_path)
if tip not in image_exts:
print('Image not in ext list {}'.format(image_path))
os.remove(image_path)
except Exception as e:
print('Issue with image {}'.format(image_path))
# os.remove(image_path)[1;31m---------------------------------------------------------------------------[0m [1;31mKeyboardInterrupt[0m Traceback (most recent call last) Cell [1;32mIn[3], line 10[0m [0;32m 8[0m image_path [38;5;241m=[39m os[38;5;241m.[39mpath[38;5;241m.[39mjoin(data_dir, image_class, image) [0;32m 9[0m [38;5;28;01mtry[39;00m: [1;32m---> 10[0m img [38;5;241m=[39m [43mcv2[49m[38;5;241;43m.[39;49m[43mimread[49m[43m([49m[43mimage_path[49m[43m)[49m [0;32m 11[0m tip [38;5;241m=[39m imghdr[38;5;241m.[39mwhat(image_path) [0;32m 12[0m [38;5;28;01mif[39;00m tip [38;5;129;01mnot[39;00m [38;5;129;01min[39;00m image_exts: [1;31mKeyboardInterrupt[0m:
batch_size = 32
img_height = 180
img_width = 180train_ds = tf.keras.utils.image_dataset_from_directory(
"trash/train",
image_size = (img_height, img_width),
seed = 123,
batch_size = batch_size
)
val_ds = tf.keras.utils.image_dataset_from_directory(
"trash/validate",
image_size = (img_height, img_width),
seed = 123,
batch_size = batch_size
)
test_ds = tf.keras.utils.image_dataset_from_directory(
"trash/test",
image_size = (img_height, img_width),
seed = 123,
batch_size = batch_size
)Found 7524 files belonging to 4 classes. Found 1531 files belonging to 4 classes. Found 2562 files belonging to 4 classes.
class_names = train_ds.class_names
print(class_names)['biological', 'glass', 'paper', 'plastic']
import matplotlib.pyplot as plt
plt.figure(figsize=(10, 10))
for images, labels in train_ds.take(1):
for i in range(9):
ax = plt.subplot(3, 3, i + 1)
plt.imshow(images[i].numpy().astype("uint8"))
plt.title(class_names[labels[i]])
plt.axis("off")AUTOTUNE = tf.data.AUTOTUNE
train_ds = train_ds.cache().shuffle(1000).prefetch(buffer_size=AUTOTUNE)
val_ds = val_ds.cache().prefetch(buffer_size=AUTOTUNE)normalization_layer = layers.Rescaling(1./255)normalized_ds = train_ds.map(lambda x, y: (normalization_layer(x), y))
image_batch, labels_batch = next(iter(normalized_ds))
first_image = image_batch[0]
# Notice the pixel values are now in `[0,1]`.
print(np.min(first_image), np.max(first_image))0.0 1.0
num_classes = len(class_names)
model = Sequential([
layers.Rescaling(1./255, input_shape=(img_height, img_width, 3)),
layers.Conv2D(16, 3, padding='same', activation='relu'),
layers.MaxPooling2D(),
layers.Conv2D(32, 3, padding='same', activation='relu'),
layers.MaxPooling2D(),
layers.Conv2D(64, 3, padding='same', activation='relu'),
layers.MaxPooling2D(),
layers.Flatten(),
layers.Dense(128, activation='relu'),
layers.Dense(num_classes)
])model.compile(optimizer='adam',
loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),
metrics=['accuracy'])model.summary()Model: "sequential"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
rescaling_1 (Rescaling) (None, 180, 180, 3) 0
conv2d (Conv2D) (None, 180, 180, 16) 448
max_pooling2d (MaxPooling2D (None, 90, 90, 16) 0
)
conv2d_1 (Conv2D) (None, 90, 90, 32) 4640
max_pooling2d_1 (MaxPooling (None, 45, 45, 32) 0
2D)
conv2d_2 (Conv2D) (None, 45, 45, 64) 18496
max_pooling2d_2 (MaxPooling (None, 22, 22, 64) 0
2D)
flatten (Flatten) (None, 30976) 0
dense (Dense) (None, 128) 3965056
dense_1 (Dense) (None, 4) 516
=================================================================
Total params: 3,989,156
Trainable params: 3,989,156
Non-trainable params: 0
_________________________________________________________________
epochs=10
history = model.fit(
train_ds,
validation_data=val_ds,
epochs=epochs
)Epoch 1/10 236/236 [==============================] - 101s 426ms/step - loss: 0.9672 - accuracy: 0.5795 - val_loss: 0.8036 - val_accuracy: 0.6656 Epoch 2/10 236/236 [==============================] - 118s 499ms/step - loss: 0.6830 - accuracy: 0.7228 - val_loss: 0.6583 - val_accuracy: 0.7466 Epoch 3/10 236/236 [==============================] - 103s 433ms/step - loss: 0.5218 - accuracy: 0.7920 - val_loss: 0.5555 - val_accuracy: 0.7884 Epoch 4/10 236/236 [==============================] - 98s 415ms/step - loss: 0.3826 - accuracy: 0.8542 - val_loss: 0.6888 - val_accuracy: 0.7688 Epoch 5/10 236/236 [==============================] - 97s 413ms/step - loss: 0.2760 - accuracy: 0.8970 - val_loss: 0.5302 - val_accuracy: 0.8295 Epoch 6/10 236/236 [==============================] - 97s 410ms/step - loss: 0.1951 - accuracy: 0.9269 - val_loss: 0.6208 - val_accuracy: 0.8184 Epoch 7/10 236/236 [==============================] - 96s 408ms/step - loss: 0.1234 - accuracy: 0.9545 - val_loss: 0.6179 - val_accuracy: 0.8354 Epoch 8/10 236/236 [==============================] - 98s 414ms/step - loss: 0.0947 - accuracy: 0.9680 - val_loss: 0.6647 - val_accuracy: 0.8406 Epoch 9/10 236/236 [==============================] - 103s 436ms/step - loss: 0.0624 - accuracy: 0.9805 - val_loss: 0.7285 - val_accuracy: 0.8321 Epoch 10/10 236/236 [==============================] - 99s 421ms/step - loss: 0.0649 - accuracy: 0.9793 - val_loss: 0.8687 - val_accuracy: 0.8374
acc = history.history['accuracy']
val_acc = history.history['val_accuracy']
loss = history.history['loss']
val_loss = history.history['val_loss']
epochs_range = range(epochs)
plt.figure(figsize=(8, 8))
plt.subplot(1, 2, 1)
plt.plot(epochs_range, acc, label='Training Accuracy')
plt.plot(epochs_range, val_acc, label='Validation Accuracy')
plt.legend(loc='lower right')
plt.title('Training and Validation Accuracy')
plt.subplot(1, 2, 2)
plt.plot(epochs_range, loss, label='Training Loss')
plt.plot(epochs_range, val_loss, label='Validation Loss')
plt.legend(loc='upper right')
plt.title('Training and Validation Loss')
plt.show()data_augmentation = keras.Sequential(
[
layers.RandomFlip("horizontal",
input_shape=(img_height,
img_width,
3)),
layers.RandomRotation(0.1),
layers.RandomZoom(0.1),
]
)plt.figure(figsize=(10, 10))
for images, _ in train_ds.take(1):
for i in range(9):
augmented_images = data_augmentation(images)
ax = plt.subplot(3, 3, i + 1)
plt.imshow(augmented_images[0].numpy().astype("uint8"))
plt.axis("off")model = Sequential([
data_augmentation,
layers.Rescaling(1./255),
layers.Conv2D(16, 3, padding='same', activation='relu'),
layers.MaxPooling2D(),
layers.Conv2D(32, 3, padding='same', activation='relu'),
layers.MaxPooling2D(),
layers.Conv2D(64, 3, padding='same', activation='relu'),
layers.MaxPooling2D(),
layers.Dropout(0.2),
layers.Flatten(),
layers.Dense(128, activation='relu'),
layers.Dense(num_classes)
])model.compile(optimizer='adam',
loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),
metrics=['accuracy'])model.summary()
Model: "sequential_2"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
sequential_1 (Sequential) (None, 180, 180, 3) 0
rescaling_2 (Rescaling) (None, 180, 180, 3) 0
conv2d_3 (Conv2D) (None, 180, 180, 16) 448
max_pooling2d_3 (MaxPooling (None, 90, 90, 16) 0
2D)
conv2d_4 (Conv2D) (None, 90, 90, 32) 4640
max_pooling2d_4 (MaxPooling (None, 45, 45, 32) 0
2D)
conv2d_5 (Conv2D) (None, 45, 45, 64) 18496
max_pooling2d_5 (MaxPooling (None, 22, 22, 64) 0
2D)
dropout (Dropout) (None, 22, 22, 64) 0
flatten_1 (Flatten) (None, 30976) 0
dense_2 (Dense) (None, 128) 3965056
dense_3 (Dense) (None, 4) 516
=================================================================
Total params: 3,989,156
Trainable params: 3,989,156
Non-trainable params: 0
_________________________________________________________________
epochs = 15
history = model.fit(
train_ds,
validation_data=val_ds,
epochs=epochs
)Epoch 1/15 236/236 [==============================] - 109s 459ms/step - loss: 1.0434 - accuracy: 0.5354 - val_loss: 0.8396 - val_accuracy: 0.6251 Epoch 2/15 236/236 [==============================] - 109s 463ms/step - loss: 0.8393 - accuracy: 0.6411 - val_loss: 0.7897 - val_accuracy: 0.6558 Epoch 3/15 236/236 [==============================] - 117s 496ms/step - loss: 0.7672 - accuracy: 0.6809 - val_loss: 0.7359 - val_accuracy: 0.7035 Epoch 4/15 236/236 [==============================] - 120s 508ms/step - loss: 0.7099 - accuracy: 0.7027 - val_loss: 0.7405 - val_accuracy: 0.6871 Epoch 5/15 236/236 [==============================] - 115s 489ms/step - loss: 0.6702 - accuracy: 0.7240 - val_loss: 0.7335 - val_accuracy: 0.7211 Epoch 6/15 236/236 [==============================] - 109s 461ms/step - loss: 0.6064 - accuracy: 0.7582 - val_loss: 0.7130 - val_accuracy: 0.7302 Epoch 7/15 236/236 [==============================] - 109s 460ms/step - loss: 0.5666 - accuracy: 0.7727 - val_loss: 0.6493 - val_accuracy: 0.7564 Epoch 8/15 236/236 [==============================] - 110s 464ms/step - loss: 0.5299 - accuracy: 0.7939 - val_loss: 0.5655 - val_accuracy: 0.7936 Epoch 9/15 236/236 [==============================] - 109s 460ms/step - loss: 0.5043 - accuracy: 0.8069 - val_loss: 0.5580 - val_accuracy: 0.7956 Epoch 10/15 236/236 [==============================] - 109s 460ms/step - loss: 0.4594 - accuracy: 0.8266 - val_loss: 0.6662 - val_accuracy: 0.7720 Epoch 11/15 236/236 [==============================] - 111s 471ms/step - loss: 0.4437 - accuracy: 0.8359 - val_loss: 0.5410 - val_accuracy: 0.8034 Epoch 12/15 236/236 [==============================] - 116s 492ms/step - loss: 0.4287 - accuracy: 0.8381 - val_loss: 0.5495 - val_accuracy: 0.8106 Epoch 13/15 236/236 [==============================] - 126s 534ms/step - loss: 0.3930 - accuracy: 0.8511 - val_loss: 0.5732 - val_accuracy: 0.8027 Epoch 14/15 236/236 [==============================] - 114s 481ms/step - loss: 0.3790 - accuracy: 0.8624 - val_loss: 0.5392 - val_accuracy: 0.8197 Epoch 15/15 236/236 [==============================] - 111s 468ms/step - loss: 0.3558 - accuracy: 0.8650 - val_loss: 0.5117 - val_accuracy: 0.8217
acc = history.history['accuracy']
val_acc = history.history['val_accuracy']
loss = history.history['loss']
val_loss = history.history['val_loss']
epochs_range = range(epochs)
plt.figure(figsize=(8, 8))
plt.subplot(1, 2, 1)
plt.plot(epochs_range, acc, label='Training Accuracy')
plt.plot(epochs_range, val_acc, label='Validation Accuracy')
plt.legend(loc='lower right')
plt.title('Training and Validation Accuracy')
plt.subplot(1, 2, 2)
plt.plot(epochs_range, loss, label='Training Loss')
plt.plot(epochs_range, val_loss, label='Validation Loss')
plt.legend(loc='upper right')
plt.title('Training and Validation Loss')
plt.show()model.evaluate(test_ds)81/81 [==============================] - 9s 110ms/step - loss: 0.6138 - accuracy: 0.8009
[0.6138237714767456, 0.800936758518219]
model.save("m_with_data_augmentation.h5")sunflower_path = "trash/test/biological/biological476.jpg"
img = tf.keras.utils.load_img(
sunflower_path, target_size=(img_height, img_width)
)
img_array = tf.keras.utils.img_to_array(img)
img_array = tf.expand_dims(img_array, 0) # Create a batch
predictions = model.predict(img_array)
score = tf.nn.softmax(predictions[0])
print(
"This image most likely belongs to {} with a {:.2f} percent confidence."
.format(class_names[np.argmax(score)], 100 * np.max(score))
)1/1 [==============================] - 0s 22ms/step This image most likely belongs to biological with a 90.06 percent confidence.