{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"code","source":"# This Python 3 environment comes with many helpful analytics libraries installed\n# It is defined by the kaggle/python Docker image: https://github.com/kaggle/docker-python\n# For example, here's several helpful packages to load\n\nimport numpy as np # linear algebra\nimport pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n\n# Input data files are available in the read-only \"../input/\" directory\n# For example, running this (by clicking run or pressing Shift+Enter) will list all files under the input directory\n\nimport os\nfor dirname, _, filenames in os.walk('/kaggle/input'):\n    for filename in filenames:\n        print(os.path.join(dirname, filename))\n\n# You can write up to 20GB to the current directory (/kaggle/working/) that gets preserved as output when you create a version using \"Save & Run All\" \n# You can also write temporary files to /kaggle/temp/, but they won't be saved outside of the current session","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Data Manipulation\nimport numpy as np\nimport pandas as pd\n# Visualisation\nimport matplotlib.pyplot as plt\n# Dataset exploring\nimport os\n# Dataset generation\nfrom keras.preprocessing import image_dataset_from_directory\nfrom keras.preprocessing.image import ImageDataGenerator\n# Transfert learning\nfrom keras.applications import VGG16\n# Optimizer\nfrom keras.optimizers import Adam\n# Keras layers\nfrom keras.layers import Input, Dense, Dropout, Flatten, AveragePooling2D\n# Keras model\nfrom keras.models import Model","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"classes = []\nclass_counter = 0\n\nfor dirname, _, filenames in os.walk('../input/dlai5-bad-apples/train/train/'):\n    if dirname.endswith('/'):\n        continue\n    else:\n        classes.append({dirname.split('/')[-1]: 0})\n    file_count = 0\n    for filename in filenames:\n        file_count += 1\n    classes[class_counter][dirname.split('/')[-1]] = file_count\n    class_counter += 1\n    \nprint('{:<15} {:<15}'.format('Class', 'Number of instances'))\nprint()\nfor d in classes:\n    [(k, v)] = d.items()\n    print('{:<15} {:<15}'.format(k, v))","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"counts = []\nlabels = []\nfor d in classes:\n    [(k, v)] = d.items()\n    labels.append(k)\n    counts.append(v)\n\nplt.figure()\nplt.bar(range(len(counts)), counts, color = ['yellow', 'orange', 'orange', 'green', 'green', 'yellow'], alpha = .7)\nplt.xticks(range(len(counts)), labels, rotation = 30)\nplt.title('Count of each label in our training data')\nplt.show()","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"TRAIN_PATH = '../input/dlai5-bad-apples/train/train/'\nTEST_PATH = '../input/dlai5-bad-apples/test/test/'","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"datagen = ImageDataGenerator(\n    rotation_range = 30, \n    zoom_range = .3, \n    horizontal_flip = True, \n    vertical_flip = True, \n    validation_split = .3\n)\n\ntrain_ds = datagen.flow_from_directory(\n    directory = TRAIN_PATH,\n    target_size = (150, 150),\n    color_mode = 'rgb',\n    class_mode = 'categorical',\n    subset = 'training'\n)\n\nvalidation_ds = datagen.flow_from_directory(\n    directory = TRAIN_PATH,\n    target_size = (150, 150),\n    color_mode = 'rgb',\n    class_mode = 'categorical',\n    subset = 'validation'\n)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"vgg16 = VGG16(include_top = False, weights = 'imagenet', input_shape = (150, 150, 3))\nvgg16.trainable = False","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"X_input = Input(shape = (150, 150, 3))\nX = vgg16(X_input)\nX = AveragePooling2D(pool_size = (3, 3), strides = 2, padding = 'valid',name = 'AvgPool2D')(X)\nX = Flatten(name = 'Flatten')(X)\nX = Dense(200, activation = 'relu', name = 'Dense1')(X)\nX = Dropout(.1)(X)\nX = Dense(100, activation = 'relu', name = 'Dense2')(X)\nX = Dropout(.1)(X)\nX = Dense(6, activation = 'softmax', name = 'Dense3')(X)\n\nmodel = Model(inputs = X_input, outputs = X, name = 'Fruit_Classifer')\n\nmodel.summary()","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"optimizer = Adam(learning_rate = 0.001)\n\nmodel.compile(optimizer = optimizer, loss = 'categorical_crossentropy', metrics = ['accuracy'])\n\n_ = model.fit(train_ds, validation_data = validation_ds, epochs = 5, batch_size = 32)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"test_ds = image_dataset_from_directory(\n    TEST_PATH,\n    label_mode = 'categorical',\n    color_mode = 'rgb',\n    image_size = (150, 150)\n)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"results = model.evaluate(test_ds)\n\nprint('{:<20} {:<20}'.format('Test loss', 'Test accuracy'))\nprint('{:<20} {:<20}'.format(np.round(results[0], 2), np.round(results[1], 2)))","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"model.save('fruit_preds')","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from keras.preprocessing import image as image_utils\nfrom keras.applications.imagenet_utils import preprocess_input\n\nimport matplotlib.image as mpimg\n\ndef make_predictions(image_path):\n    image = image_utils.load_img(image_path, target_size=(150, 150))\n    image = image_utils.img_to_array(image)\n    image = image.reshape(1,150,150,3)\n    image = preprocess_input(image)\n    preds = model.predict(image)\n    return preds.argmax()\n\n\nimport os\nfile = open('submission.csv','w')\nfile.write(f'ID,Code\\n')\nfor dirname, _, filenames in os.walk('../input/dlai5-bad-apples/validation/'):\n    for filename in filenames:\n        p = os.path.join(dirname, filename)\n        r = make_predictions(p) + 1\n        file.write(f'{filename},{r}\\n')\nfile.close()\nprint('output done.')","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from keras.preprocessing import image as image_utils\nfrom keras.applications.imagenet_utils import preprocess_input\nimport matplotlib.pyplot as plt\nimport numpy as np\nimport matplotlib.image as mpimg\nimport os\nimport random\n\ndef make_predictions(image_path):\n    image = image_utils.load_img(image_path, target_size=(150, 150))\n    image = image_utils.img_to_array(image)\n    image = image.reshape(1,150,150,3)\n    image = preprocess_input(image)\n    preds = (model.predict(image)).argmax()\n    return preds+1\n\ndef get_label(pred):\n    if pred==1:\n        return 'freshapples'\n    if pred==2:\n        return 'freshbananas'\n    if pred==3:\n        return 'freshoranges'\n    if pred==4:\n        return 'rottenapples'\n    if pred==5:\n        return 'rottenbananas'\n    if pred==6:\n        return 'rottenoranges'\n\nbase_dir = '../input/dlai5-bad-apples'\ntrain_dir = os.path.join(base_dir,'train/train')\ntest_dir = os.path.join(base_dir, 'test/test')\nvalid_dir = os.path.join(base_dir, 'validation')\n\ndef show_samples(folder, rows=3, cols=5):\n    images = []\n    for i in range(rows*cols):\n        fruit = random.choice(os.listdir(folder))\n        image = random.choice(os.listdir(os.path.join(folder, fruit)))\n        images.append([fruit, os.path.join(folder, fruit, image)])\n        p = os.path.join(folder, fruit, image)\n        print(get_label(make_predictions(p)))\n    _, axes = plt.subplots(rows,cols, figsize=(18,5))\n\n    axes = axes.flatten()\n    for img, ax in zip(images, axes):\n        ax.imshow(mpimg.imread(img[1]))\n        ax.set_title(img[0])\n        ax.set_axis_off()\n        \n    plt.show()\n\nshow_samples(valid_dir)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import pandas\ndf = pandas.read_csv('submission.csv')\nprint(df.head)","metadata":{"trusted":true},"execution_count":null,"outputs":[]}]}