v1.0

main.py

@@ -0,0 +1,158 @@
+import os
+
+import matplotlib.pyplot as plt
+import numpy as np
+import tensorflow as tf
+
+from tensorflow import keras
+from tensorflow.keras import layers
+from tensorflow.keras.models import Sequential
+
+import pathlib
+
+from tensorflow.python.client import device_lib
+print(device_lib.list_local_devices())
+
+data_dir = "C:/Users/jilek/Downloads/AAT+"
+data_dir = pathlib.Path(data_dir).with_suffix('')
+
+data_dir_test = "C:/Users/jilek/Downloads/AAT+_TEST"
+data_dir_test = pathlib.Path(data_dir_test).with_suffix('')
+
+image_count = len(list(data_dir.glob('*/*.jpg')))
+print(image_count)
+
+batch_size = 1
+img_height = 1024
+img_width = 1024
+
+train_ds = tf.keras.utils.image_dataset_from_directory(
+    data_dir,
+    validation_split=0.0,
+    #subset="training",
+    seed=123,
+    labels='inferred',
+    label_mode='categorical',
+    class_names=["C100", "C095", "C090", "C085", "C080", "C070", "C060", "C040", "C020"],
+    color_mode="grayscale", #grayscale
+    shuffle=True,
+    image_size=(img_height, img_width),
+    batch_size=batch_size)
+
+val_ds = tf.keras.utils.image_dataset_from_directory(
+    data_dir_test,
+    validation_split=0.0,
+    #subset="validation",
+    seed=123,
+    labels='inferred',
+    label_mode='categorical',
+    class_names=["C100", "C095", "C090", "C085", "C080", "C070", "C060", "C040", "C020"],
+    color_mode="grayscale",
+    image_size=(img_height, img_width),
+    batch_size=batch_size)
+
+class_names = train_ds.class_names
+print(class_names)
+
+for image_batch, labels_batch in train_ds:
+    print(image_batch.shape)
+    print(labels_batch.shape)
+    break
+
+AUTOTUNE = tf.data.AUTOTUNE
+
+data_augmentation = keras.Sequential(
+    [
+        layers.RandomFlip("horizontal_and_vertical",
+                          input_shape=(img_height,
+                                       img_width,
+                                       1)), #rgb
+        #layers.RandomRotation(0.5),
+        #layers.RandomZoom(0.5),
+    ]
+)
+
+train_ds = train_ds.shuffle(buffer_size=900).prefetch(buffer_size=AUTOTUNE) #.cache()
+val_ds = val_ds.prefetch(buffer_size=AUTOTUNE) #.cache()
+
+
+num_classes = len(class_names)
+print(str(num_classes))
+
+model = Sequential([
+  layers.Rescaling(1.0/255, input_shape=(img_height, img_width, 1)), #rgb
+  #layers.Dropout(0.0),
+  #layers.MaxPooling2D(pool_size=(8, 8)),
+  layers.Conv2D(4, (4, 4), strides=(2, 2), padding='valid', dilation_rate=(1, 1), groups=1, input_shape=(1024, 1024, 1), activation='relu'),
+  layers.Conv2D(8, (4, 4), strides=(2, 2), padding='valid', dilation_rate=(1, 1), groups=1, input_shape=(512, 512, 4), activation='relu'),
+  layers.Conv2D(16, (4, 4), strides=(4, 4), padding='valid', dilation_rate=(1, 1), groups=1, input_shape=(256, 256, 8), activation='relu'),
+  layers.Conv2D(32, (4, 4), strides=(4, 4), padding='valid', dilation_rate=(1, 1), groups=1, input_shape=(64, 64, 16), activation='relu'),
+  layers.Conv2D(64, (4, 4), strides=(4, 4), padding='valid', dilation_rate=(1, 1), groups=1, input_shape=(16, 16, 32), activation='relu'),
+  #layers.Conv2D(128, (4, 4), strides=(1, 1), padding='valid', dilation_rate=(1, 1), groups=1, input_shape=(8, 8, 64), activation='relu'),
+  #layers.Dropout(0.1),
+  layers.Flatten(),
+  layers.Dense(32, activation='relu'),
+  layers.Dense(num_classes, activation='softmax')
+])
+model.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=1e-5),
+              loss=tf.keras.losses.CategoricalCrossentropy(),
+              metrics=['accuracy'])
+
+model.summary()
+model.save("./model/AAT+")
+
+epochs = 130
+history = model.fit(
+  train_ds,
+  validation_data=val_ds,
+  epochs=epochs
+)
+
+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()
+
+test_dir = "C:/Users/jilek/Downloads/AAT_T/"
+for file_name in os.listdir(test_dir):
+    file_path = os.path.join(test_dir, file_name)
+    img = tf.keras.utils.load_img(
+        file_path, target_size=(img_height, img_width), color_mode="grayscale" #grayscale
+    )
+    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(file_name)
+    print(
+        "This image most likely belongs to {} with a {:.2f} percent confidence."
+        .format(class_names[np.argmax(score)], 100 * np.max(score))
+    )
+
+
+
+
+
+
+
+
+

model/AAT+/keras_metadata.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:429c89709452f69986c9cc2c9549be3e6e6525d467876355348d373241545d32
+size 23711

model/AAT+/saved_model.pb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bf5d62d94e882bf080ea5aa8f5ba9c4635bddd6a2cfd03d4be9ba875f2b5fa1a
+size 148529