added neural network class; made necessary changes to project structure

.gitignore

@@ -143,3 +143,5 @@ cython_debug/
 
 # local sandbox
 sandbox/
+/algorithms/learn/decision_tree/decistion_tree.png
+/resources/data/neural_network/train

algorithms/learn/decision_tree/decision_tree.py

@@ -18,7 +18,7 @@ class DecisionTree:
             self.vec = None
 
     def build(self, training_file: str, depth: int):
-        path = os.path.join("..", "..", "resources", "data", training_file)
+        path = os.path.join("../..", "..", "resources", "data", training_file)
 
         samples = list()
         results = list()
@@ -87,4 +87,4 @@ if __name__ == "__main__":
     # generate_data("training_set.txt", 12000)
     decision_tree = DecisionTree()
     decision_tree.build("training_set.txt", 15)
-    decision_tree.test()
+    decision_tree.save()

algorithms/learn/neural_network/neural_network.py

@@ -0,0 +1,124 @@
+import numpy as np
+import joblib
+import pathlib
+import tensorflow as tf
+
+from tensorflow import keras
+from tensorflow.keras import layers
+from tensorflow.keras.models import Sequential
+
+
+class NeuralNetwork:
+    def __init__(self, saved_model_path=None, classes_path=None, img_height=180, img_width=180):
+        self.training_data_dir = pathlib.Path(r"../../../resources/data/neural_network/train")
+        self.img_height = img_height
+        self.img_width = img_width
+
+        if saved_model_path is not None and classes_path is not None:
+            self.load(saved_model_path, classes_path)
+        else:
+            self.model = None
+            self.class_names = None
+
+    def build(self, num_classes=3, epochs=15, batch_size=32):
+        train_ds = tf.keras.preprocessing.image_dataset_from_directory(
+            self.training_data_dir,
+            validation_split=0.2,
+            subset="training",
+            seed=123,
+            image_size=(self.img_height, self.img_width),
+            batch_size=batch_size)
+
+        self.class_names = train_ds.class_names
+
+        val_ds = tf.keras.preprocessing.image_dataset_from_directory(
+            self.training_data_dir,
+            validation_split=0.2,
+            subset="validation",
+            seed=123,
+            image_size=(self.img_height, self.img_width),
+            batch_size=batch_size)
+
+        autotune = tf.data.AUTOTUNE
+
+        train_ds = train_ds.cache().shuffle(1000).prefetch(buffer_size=autotune)
+        val_ds = val_ds.cache().prefetch(buffer_size=autotune)
+
+        data_augmentation = keras.Sequential(
+            [
+                layers.experimental.preprocessing.RandomFlip("horizontal",
+                                                             input_shape=(self.img_height,
+                                                                          self.img_width,
+                                                                          3)),
+                layers.experimental.preprocessing.RandomRotation(0.1),
+                layers.experimental.preprocessing.RandomZoom(0.1),
+            ]
+        )
+
+        model = Sequential([
+            data_augmentation,
+            layers.experimental.preprocessing.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.fit(
+            train_ds,
+            validation_data=val_ds,
+            epochs=epochs
+        )
+
+        self.model = model
+
+    def save(self):
+        self.model.save("saved_model.h5")
+        joblib.dump(self.class_names, "saved_model_classes.joblib")
+
+    def load(self, model_path, classes_path):
+        self.model = tf.keras.models.load_model(model_path)
+        self.class_names = joblib.load(classes_path)
+
+    def get_answer(self, image_path):
+        img = keras.preprocessing.image.load_img(
+            image_path, target_size=(self.img_height, self.img_width)
+        )
+
+        img_array = keras.preprocessing.image.img_to_array(img)
+        img_array = tf.expand_dims(img_array, 0)
+
+        predictions = self.model.predict(img_array)
+        score = tf.nn.softmax(predictions[0])
+
+        print(
+            "This image most likely belongs to {} with a {:.2f} percent confidence."
+            .format(self.class_names[np.argmax(score)], 100 * np.max(score))
+        )
+
+
+if __name__ == "__main__":
+    # Building and saving a new model:
+
+    # (requires a valid training set in resources/data/neural_network/train)
+    # neural_network = NeuralNetwork()
+    # neural_network.build(epochs=10)
+    # neural_network.save()
+
+    # Loading a model from file:
+    neural_network = NeuralNetwork("saved_model.h5", "saved_model_classes.joblib")
+
+    # Test
+    image = r"../../../resources/data/neural_network/disarm/ships/0e91bae301a5f5b136c2a0d20ef97cf6.jpg"
+    neural_network.get_answer(image)