emnist-train-simple-hcr-model-using-keras.ipynb

{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"name":"python","version":"3.7.9","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"}},"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\nimport cv2\nimport matplotlib.pyplot as plt\n\nfrom sklearn.model_selection import train_test_split\n\nimport tensorflow as tf\n\nfrom keras.models import Sequential, load_model\nfrom keras.callbacks import EarlyStopping, ModelCheckpoint\nfrom keras import layers\nfrom keras.layers import *\nfrom keras.utils import np_utils\n\nfrom tqdm import tqdm\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":1,"outputs":[{"name":"stdout","text":"/kaggle/input/emnist/emnist-letters-mapping.txt\n/kaggle/input/emnist/emnist-letters-test.csv\n/kaggle/input/emnist/emnist-mnist-mapping.txt\n/kaggle/input/emnist/emnist-digits-train.csv\n/kaggle/input/emnist/emnist-bymerge-mapping.txt\n/kaggle/input/emnist/emnist-balanced-train.csv\n/kaggle/input/emnist/emnist-digits-test.csv\n/kaggle/input/emnist/emnist-balanced-test.csv\n/kaggle/input/emnist/emnist-mnist-test.csv\n/kaggle/input/emnist/emnist-letters-train.csv\n/kaggle/input/emnist/emnist-byclass-train.csv\n/kaggle/input/emnist/emnist-bymerge-test.csv\n/kaggle/input/emnist/emnist-balanced-mapping.txt\n/kaggle/input/emnist/emnist-mnist-train.csv\n/kaggle/input/emnist/emnist-digits-mapping.txt\n/kaggle/input/emnist/emnist-bymerge-train.csv\n/kaggle/input/emnist/emnist-byclass-test.csv\n/kaggle/input/emnist/emnist-byclass-mapping.txt\n/kaggle/input/emnist/emnist_source_files/emnist-digits-test-labels-idx1-ubyte\n/kaggle/input/emnist/emnist_source_files/emnist-bymerge-train-labels-idx1-ubyte\n/kaggle/input/emnist/emnist_source_files/emnist-letters-test-labels-idx1-ubyte\n/kaggle/input/emnist/emnist_source_files/emnist-byclass-train-images-idx3-ubyte\n/kaggle/input/emnist/emnist_source_files/emnist-byclass-test-images-idx3-ubyte\n/kaggle/input/emnist/emnist_source_files/emnist-mnist-train-images-idx3-ubyte\n/kaggle/input/emnist/emnist_source_files/emnist-digits-train-labels-idx1-ubyte\n/kaggle/input/emnist/emnist_source_files/emnist-bymerge-test-images-idx3-ubyte\n/kaggle/input/emnist/emnist_source_files/emnist-bymerge-test-labels-idx1-ubyte\n/kaggle/input/emnist/emnist_source_files/emnist-mnist-test-labels-idx1-ubyte\n/kaggle/input/emnist/emnist_source_files/emnist-balanced-test-labels-idx1-ubyte\n/kaggle/input/emnist/emnist_source_files/emnist-balanced-test-images-idx3-ubyte\n/kaggle/input/emnist/emnist_source_files/emnist-mnist-test-images-idx3-ubyte\n/kaggle/input/emnist/emnist_source_files/emnist-bymerge-train-images-idx3-ubyte\n/kaggle/input/emnist/emnist_source_files/emnist-letters-train-labels-idx1-ubyte\n/kaggle/input/emnist/emnist_source_files/emnist-digits-train-images-idx3-ubyte\n/kaggle/input/emnist/emnist_source_files/emnist-byclass-train-labels-idx1-ubyte\n/kaggle/input/emnist/emnist_source_files/emnist-balanced-train-images-idx3-ubyte\n/kaggle/input/emnist/emnist_source_files/emnist-balanced-train-labels-idx1-ubyte\n/kaggle/input/emnist/emnist_source_files/emnist-letters-test-images-idx3-ubyte\n/kaggle/input/emnist/emnist_source_files/emnist-byclass-test-labels-idx1-ubyte\n/kaggle/input/emnist/emnist_source_files/emnist-letters-train-images-idx3-ubyte\n/kaggle/input/emnist/emnist_source_files/emnist-mnist-train-labels-idx1-ubyte\n/kaggle/input/emnist/emnist_source_files/emnist-digits-test-images-idx3-ubyte\n","output_type":"stream"}]},{"cell_type":"code","source":"train_df = pd.read_csv('/kaggle/input/emnist/emnist-balanced-train.csv', header=None)\ntrain_df.head()","metadata":{"trusted":true},"execution_count":2,"outputs":[{"execution_count":2,"output_type":"execute_result","data":{"text/plain":"   0    1    2    3    4    5    6    7    8    9    ...  775  776  777  778  \\\n0   45    0    0    0    0    0    0    0    0    0  ...    0    0    0    0   \n1   36    0    0    0    0    0    0    0    0    0  ...    0    0    0    0   \n2   43    0    0    0    0    0    0    0    0    0  ...    0    0    0    0   \n3   15    0    0    0    0    0    0    0    0    0  ...    0    0    0    0   \n4    4    0    0    0    0    0    0    0    0    0  ...    0    0    0    0   \n\n   779  780  781  782  783  784  \n0    0    0    0    0    0    0  \n1    0    0    0    0    0    0  \n2    0    0    0    0    0    0  \n3    0    0    0    0    0    0  \n4    0    0    0    0    0    0  \n\n[5 rows x 785 columns]","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>0</th>\n      <th>1</th>\n      <th>2</th>\n      <th>3</th>\n      <th>4</th>\n      <th>5</th>\n      <th>6</th>\n      <th>7</th>\n      <th>8</th>\n      <th>9</th>\n      <th>...</th>\n      <th>775</th>\n      <th>776</th>\n      <th>777</th>\n      <th>778</th>\n      <th>779</th>\n      <th>780</th>\n      <th>781</th>\n      <th>782</th>\n      <th>783</th>\n      <th>784</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>45</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>36</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>43</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>15</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>4</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n  </tbody>\n</table>\n<p>5 rows × 785 columns</p>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"train_df.shape","metadata":{"trusted":true},"execution_count":3,"outputs":[{"execution_count":3,"output_type":"execute_result","data":{"text/plain":"(112800, 785)"},"metadata":{}}]},{"cell_type":"markdown","source":"### Split train dataframe into X & y","metadata":{}},{"cell_type":"code","source":"X_train = train_df.loc[:, 1:]\ny_train = train_df.loc[:, 0]\n\nX_train.shape, y_train.shape","metadata":{"trusted":true},"execution_count":4,"outputs":[{"execution_count":4,"output_type":"execute_result","data":{"text/plain":"((112800, 784), (112800,))"},"metadata":{}}]},{"cell_type":"code","source":"X_train.head()","metadata":{"trusted":true},"execution_count":5,"outputs":[{"execution_count":5,"output_type":"execute_result","data":{"text/plain":"   1    2    3    4    5    6    7    8    9    10   ...  775  776  777  778  \\\n0    0    0    0    0    0    0    0    0    0    0  ...    0    0    0    0   \n1    0    0    0    0    0    0    0    0    0    0  ...    0    0    0    0   \n2    0    0    0    0    0    0    0    0    0    0  ...    0    0    0    0   \n3    0    0    0    0    0    0    0    0    0    0  ...    0    0    0    0   \n4    0    0    0    0    0    0    0    0    0    0  ...    0    0    0    0   \n\n   779  780  781  782  783  784  \n0    0    0    0    0    0    0  \n1    0    0    0    0    0    0  \n2    0    0    0    0    0    0  \n3    0    0    0    0    0    0  \n4    0    0    0    0    0    0  \n\n[5 rows x 784 columns]","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>1</th>\n      <th>2</th>\n      <th>3</th>\n      <th>4</th>\n      <th>5</th>\n      <th>6</th>\n      <th>7</th>\n      <th>8</th>\n      <th>9</th>\n      <th>10</th>\n      <th>...</th>\n      <th>775</th>\n      <th>776</th>\n      <th>777</th>\n      <th>778</th>\n      <th>779</th>\n      <th>780</th>\n      <th>781</th>\n      <th>782</th>\n      <th>783</th>\n      <th>784</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n  </tbody>\n</table>\n<p>5 rows × 784 columns</p>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"y_train.head()","metadata":{"trusted":true},"execution_count":6,"outputs":[{"execution_count":6,"output_type":"execute_result","data":{"text/plain":"0    45\n1    36\n2    43\n3    15\n4     4\nName: 0, dtype: int64"},"metadata":{}}]},{"cell_type":"markdown","source":"### Create label dictionary","metadata":{}},{"cell_type":"code","source":"label_map = pd.read_csv(\"/kaggle/input/emnist/emnist-balanced-mapping.txt\", \n                        delimiter = ' ', \n                        index_col=0, \n                        header=None, \n                        squeeze=True)\nlabel_map.head()","metadata":{"trusted":true},"execution_count":7,"outputs":[{"execution_count":7,"output_type":"execute_result","data":{"text/plain":"0\n0    48\n1    49\n2    50\n3    51\n4    52\nName: 1, dtype: int64"},"metadata":{}}]},{"cell_type":"code","source":"label_dictionary = {}\nfor index, label in enumerate(label_map):\n    label_dictionary[index] = chr(label)\n\nlabel_dictionary","metadata":{"trusted":true},"execution_count":8,"outputs":[{"execution_count":8,"output_type":"execute_result","data":{"text/plain":"{0: '0',\n 1: '1',\n 2: '2',\n 3: '3',\n 4: '4',\n 5: '5',\n 6: '6',\n 7: '7',\n 8: '8',\n 9: '9',\n 10: 'A',\n 11: 'B',\n 12: 'C',\n 13: 'D',\n 14: 'E',\n 15: 'F',\n 16: 'G',\n 17: 'H',\n 18: 'I',\n 19: 'J',\n 20: 'K',\n 21: 'L',\n 22: 'M',\n 23: 'N',\n 24: 'O',\n 25: 'P',\n 26: 'Q',\n 27: 'R',\n 28: 'S',\n 29: 'T',\n 30: 'U',\n 31: 'V',\n 32: 'W',\n 33: 'X',\n 34: 'Y',\n 35: 'Z',\n 36: 'a',\n 37: 'b',\n 38: 'd',\n 39: 'e',\n 40: 'f',\n 41: 'g',\n 42: 'h',\n 43: 'n',\n 44: 'q',\n 45: 'r',\n 46: 't'}"},"metadata":{}}]},{"cell_type":"markdown","source":"### Visualize sample data","metadata":{}},{"cell_type":"code","source":"# Sample entry number 42\nsample_image = X_train.iloc[42]\nsample_label = y_train.iloc[42]\nsample_image.shape, sample_label","metadata":{"trusted":true},"execution_count":9,"outputs":[{"execution_count":9,"output_type":"execute_result","data":{"text/plain":"((784,), 23)"},"metadata":{}}]},{"cell_type":"code","source":"W = 28\nH = 28","metadata":{"trusted":true},"execution_count":10,"outputs":[]},{"cell_type":"code","source":"print(\"Label entry 42:\", label_dictionary[sample_label])\nplt.imshow(sample_image.values.reshape(W, H), cmap=plt.cm.gray)\nplt.show()","metadata":{"trusted":true},"execution_count":11,"outputs":[{"name":"stdout","text":"Label entry 42: N\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<Figure size 432x288 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":"def reshape_and_rotate(image):\n    W = 28\n    H = 28\n    image = image.reshape(W, H)\n    image = np.fliplr(image)\n    image = np.rot90(image)\n    return image\n\nprint(\"Label entry 42:\", label_dictionary[sample_label])\nplt.imshow(reshape_and_rotate(sample_image.values), cmap=plt.cm.gray)\nplt.show()","metadata":{"trusted":true},"execution_count":12,"outputs":[{"name":"stdout","text":"Label entry 42: N\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<Figure size 432x288 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"markdown","source":"Next we want to apply reshape_and_rotate to all images in X_train","metadata":{}},{"cell_type":"code","source":"# note: np.apply_along_axis returns a numpy array, X_train is not a pandas.DataFrame anymore\nX_train = np.apply_along_axis(reshape_and_rotate, 1, X_train.values)\nX_train.shape","metadata":{"trusted":true},"execution_count":13,"outputs":[{"execution_count":13,"output_type":"execute_result","data":{"text/plain":"(112800, 28, 28)"},"metadata":{}}]},{"cell_type":"markdown","source":"## Visualize more sample","metadata":{}},{"cell_type":"code","source":"sample_image = X_train[42]\nsample_label = y_train.iloc[42]\nprint(\"Label entry 42:\", label_dictionary[sample_label])\nplt.imshow(sample_image.reshape(W, H), cmap=plt.cm.gray)\nplt.show()\n\n\nfor i in range(100, 106):\n    plt.subplot(390 + (i+1))\n    plt.imshow(X_train[i], cmap=plt.cm.gray)\n    plt.title(label_dictionary[y_train[i]])","metadata":{"trusted":true},"execution_count":14,"outputs":[{"name":"stdout","text":"Label entry 42: N\n","output_type":"stream"},{"output_type":"display_data","data":{"text/plain":"<Figure size 432x288 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}},{"output_type":"display_data","data":{"text/plain":"<Figure size 432x288 with 6 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"markdown","source":"Looks good\n\n### Normalize Data","metadata":{}},{"cell_type":"code","source":"X_train = X_train.astype('float32') / 255","metadata":{"trusted":true},"execution_count":15,"outputs":[]},{"cell_type":"markdown","source":"### One Hot Encode Label","metadata":{}},{"cell_type":"code","source":"number_of_classes = y_train.nunique()\nnumber_of_classes","metadata":{"trusted":true},"execution_count":16,"outputs":[{"execution_count":16,"output_type":"execute_result","data":{"text/plain":"47"},"metadata":{}}]},{"cell_type":"code","source":"y_train = np_utils.to_categorical(y_train, number_of_classes)\ny_train.shape","metadata":{"trusted":true},"execution_count":17,"outputs":[{"execution_count":17,"output_type":"execute_result","data":{"text/plain":"(112800, 47)"},"metadata":{}}]},{"cell_type":"code","source":"y_train","metadata":{"trusted":true},"execution_count":18,"outputs":[{"execution_count":18,"output_type":"execute_result","data":{"text/plain":"array([[0., 0., 0., ..., 0., 1., 0.],\n       [0., 0., 0., ..., 0., 0., 0.],\n       [0., 0., 0., ..., 0., 0., 0.],\n       ...,\n       [0., 0., 0., ..., 0., 0., 0.],\n       [0., 0., 0., ..., 0., 0., 0.],\n       [0., 0., 0., ..., 0., 0., 0.]], dtype=float32)"},"metadata":{}}]},{"cell_type":"code","source":"# Reshape to fit model input shape\n# Tensorflow (batch, width, height, channels)\nX_train = X_train.reshape(-1, W, H, 1)\nX_train.shape","metadata":{"trusted":true},"execution_count":19,"outputs":[{"execution_count":19,"output_type":"execute_result","data":{"text/plain":"(112800, 28, 28, 1)"},"metadata":{}}]},{"cell_type":"code","source":"# Split 10% validation \nX_train, X_val, y_train, y_val = train_test_split(X_train, \n                                                  y_train, \n                                                  test_size= 0.1, \n                                                  random_state=88)","metadata":{"trusted":true},"execution_count":20,"outputs":[]},{"cell_type":"markdown","source":"### Define model\nLeNet 5","metadata":{}},{"cell_type":"code","source":"model = Sequential()\n\nmodel.add(layers.Conv2D(filters=32, kernel_size=(5,5), padding='same', activation='relu', input_shape=(W, H, 1)))\nmodel.add(layers.MaxPool2D(strides=2))\nmodel.add(layers.Conv2D(filters=48, kernel_size=(5,5), padding='valid', activation='relu'))\nmodel.add(layers.MaxPool2D(strides=2))\nmodel.add(layers.Flatten())\nmodel.add(layers.Dense(256, activation='relu'))\nmodel.add(layers.Dense(84, activation='relu'))\nmodel.add(layers.Dense(number_of_classes, activation='softmax'))\n\nmodel.summary()","metadata":{"trusted":true},"execution_count":21,"outputs":[{"name":"stdout","text":"Model: \"sequential\"\n_________________________________________________________________\nLayer (type)                 Output Shape              Param #   \n=================================================================\nconv2d (Conv2D)              (None, 28, 28, 32)        832       \n_________________________________________________________________\nmax_pooling2d (MaxPooling2D) (None, 14, 14, 32)        0         \n_________________________________________________________________\nconv2d_1 (Conv2D)            (None, 10, 10, 48)        38448     \n_________________________________________________________________\nmax_pooling2d_1 (MaxPooling2 (None, 5, 5, 48)          0         \n_________________________________________________________________\nflatten (Flatten)            (None, 1200)              0         \n_________________________________________________________________\ndense (Dense)                (None, 256)               307456    \n_________________________________________________________________\ndense_1 (Dense)              (None, 84)                21588     \n_________________________________________________________________\ndense_2 (Dense)              (None, 47)                3995      \n=================================================================\nTotal params: 372,319\nTrainable params: 372,319\nNon-trainable params: 0\n_________________________________________________________________\n","output_type":"stream"}]},{"cell_type":"code","source":"optimizer_name = 'adam'\n\nmodel.compile(loss='categorical_crossentropy', optimizer=optimizer_name, metrics=['accuracy'])\n\nearly_stopping = EarlyStopping(monitor='val_loss', patience=5, verbose=1, mode='min')\nmcp_save = ModelCheckpoint('my_model.h5', save_best_only=True, monitor='val_loss', verbose=1, mode='auto')","metadata":{"trusted":true},"execution_count":22,"outputs":[]},{"cell_type":"code","source":"history = model.fit(X_train,\n                    y_train, \n                    epochs=30, \n                    batch_size=32, \n                    verbose=1, \n                    validation_split=0.1,\n                    callbacks=[early_stopping, mcp_save])","metadata":{"trusted":true},"execution_count":23,"outputs":[{"name":"stdout","text":"Epoch 1/30\n2856/2856 [==============================] - 11s 3ms/step - loss: 1.0924 - accuracy: 0.6779 - val_loss: 0.4187 - val_accuracy: 0.8545\n\nEpoch 00001: val_loss improved from inf to 0.41875, saving model to my_model.h5\nEpoch 2/30\n2856/2856 [==============================] - 9s 3ms/step - loss: 0.3849 - accuracy: 0.8624 - val_loss: 0.3908 - val_accuracy: 0.8615\n\nEpoch 00002: val_loss improved from 0.41875 to 0.39075, saving model to my_model.h5\nEpoch 3/30\n2856/2856 [==============================] - 9s 3ms/step - loss: 0.3195 - accuracy: 0.8815 - val_loss: 0.3522 - val_accuracy: 0.8753\n\nEpoch 00003: val_loss improved from 0.39075 to 0.35222, saving model to my_model.h5\nEpoch 4/30\n2856/2856 [==============================] - 9s 3ms/step - loss: 0.2774 - accuracy: 0.8947 - val_loss: 0.3563 - val_accuracy: 0.8759\n\nEpoch 00004: val_loss did not improve from 0.35222\nEpoch 5/30\n2856/2856 [==============================] - 9s 3ms/step - loss: 0.2486 - accuracy: 0.9033 - val_loss: 0.3487 - val_accuracy: 0.8790\n\nEpoch 00005: val_loss improved from 0.35222 to 0.34871, saving model to my_model.h5\nEpoch 6/30\n2856/2856 [==============================] - 9s 3ms/step - loss: 0.2223 - accuracy: 0.9098 - val_loss: 0.3623 - val_accuracy: 0.8796\n\nEpoch 00006: val_loss did not improve from 0.34871\nEpoch 7/30\n2856/2856 [==============================] - 8s 3ms/step - loss: 0.2055 - accuracy: 0.9148 - val_loss: 0.3818 - val_accuracy: 0.8791\n\nEpoch 00007: val_loss did not improve from 0.34871\nEpoch 8/30\n2856/2856 [==============================] - 9s 3ms/step - loss: 0.1835 - accuracy: 0.9239 - val_loss: 0.3902 - val_accuracy: 0.8753\n\nEpoch 00008: val_loss did not improve from 0.34871\nEpoch 9/30\n2856/2856 [==============================] - 8s 3ms/step - loss: 0.1732 - accuracy: 0.9274 - val_loss: 0.4182 - val_accuracy: 0.8772\n\nEpoch 00009: val_loss did not improve from 0.34871\nEpoch 10/30\n2856/2856 [==============================] - 8s 3ms/step - loss: 0.1588 - accuracy: 0.9347 - val_loss: 0.4487 - val_accuracy: 0.8726\n\nEpoch 00010: val_loss did not improve from 0.34871\nEpoch 00010: early stopping\n","output_type":"stream"}]},{"cell_type":"code","source":"# plot accuracy and loss\ndef plotgraph(epochs, acc, val_acc):\n    # Plot training & validation accuracy values\n    plt.plot(epochs, acc, 'b')\n    plt.plot(epochs, val_acc, 'r')\n    plt.title('Model accuracy')\n    plt.ylabel('Accuracy')\n    plt.xlabel('Epoch')\n    plt.legend(['Train', 'Val'], loc='upper left')\n    plt.show()\n    \nacc = history.history['accuracy']\nval_acc = history.history['val_accuracy']\nloss = history.history['loss']\nval_loss = history.history['val_loss']\nepochs = range(1,len(acc)+1)","metadata":{"trusted":true},"execution_count":24,"outputs":[]},{"cell_type":"code","source":"# Accuracy curve\nplotgraph(epochs, acc, val_acc)","metadata":{"trusted":true},"execution_count":25,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 432x288 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":"# loss curve\nplotgraph(epochs, loss, val_loss)","metadata":{"trusted":true},"execution_count":26,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 432x288 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"markdown","source":"### Test model","metadata":{}},{"cell_type":"code","source":"# Load best model\nmodel = load_model('/kaggle/working/my_model.h5')\nmodel.summary()","metadata":{"trusted":true},"execution_count":27,"outputs":[{"name":"stdout","text":"Model: \"sequential\"\n_________________________________________________________________\nLayer (type)                 Output Shape              Param #   \n=================================================================\nconv2d (Conv2D)              (None, 28, 28, 32)        832       \n_________________________________________________________________\nmax_pooling2d (MaxPooling2D) (None, 14, 14, 32)        0         \n_________________________________________________________________\nconv2d_1 (Conv2D)            (None, 10, 10, 48)        38448     \n_________________________________________________________________\nmax_pooling2d_1 (MaxPooling2 (None, 5, 5, 48)          0         \n_________________________________________________________________\nflatten (Flatten)            (None, 1200)              0         \n_________________________________________________________________\ndense (Dense)                (None, 256)               307456    \n_________________________________________________________________\ndense_1 (Dense)              (None, 84)                21588     \n_________________________________________________________________\ndense_2 (Dense)              (None, 47)                3995      \n=================================================================\nTotal params: 372,319\nTrainable params: 372,319\nNon-trainable params: 0\n_________________________________________________________________\n","output_type":"stream"}]},{"cell_type":"code","source":"y_pred = model.predict(X_val)\ny_pred.shape","metadata":{"trusted":true},"execution_count":28,"outputs":[{"execution_count":28,"output_type":"execute_result","data":{"text/plain":"(11280, 47)"},"metadata":{}}]},{"cell_type":"code","source":"for i in range(10, 16):\n    plt.subplot(380 + (i%10+1))\n    plt.imshow(X_val[i].reshape(28, 28), cmap=plt.cm.gray)\n    plt.title(label_dictionary[y_pred[i].argmax()])","metadata":{"trusted":true},"execution_count":29,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 432x288 with 6 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"code","source":"for i in range(42, 48):\n    plt.subplot(380 + (i%10+1))\n    plt.imshow(X_val[i].reshape(28, 28), cmap=plt.cm.gray)\n    plt.title(label_dictionary[y_pred[i].argmax()])","metadata":{"trusted":true},"execution_count":30,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 432x288 with 6 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"cell_type":"markdown","source":"#### Test accuracy","metadata":{}},{"cell_type":"code","source":"model.evaluate(X_val, y_val)","metadata":{"trusted":true},"execution_count":31,"outputs":[{"name":"stdout","text":"353/353 [==============================] - 1s 2ms/step - loss: 0.3640 - accuracy: 0.8715\n","output_type":"stream"},{"execution_count":31,"output_type":"execute_result","data":{"text/plain":"[0.363991379737854, 0.8714538812637329]"},"metadata":{}}]},{"cell_type":"markdown","source":"### Load test dataset & preprocess image like how we did to train dataset","metadata":{}},{"cell_type":"code","source":"test_df = pd.read_csv('/kaggle/input/emnist/emnist-balanced-test.csv', header=None)\ntest_df.shape","metadata":{"trusted":true},"execution_count":32,"outputs":[{"execution_count":32,"output_type":"execute_result","data":{"text/plain":"(18800, 785)"},"metadata":{}}]},{"cell_type":"code","source":"test_df.describe()","metadata":{"trusted":true},"execution_count":33,"outputs":[{"execution_count":33,"output_type":"execute_result","data":{"text/plain":"                0        1        2        3        4             5    \\\ncount  18800.000000  18800.0  18800.0  18800.0  18800.0  18800.000000   \nmean      23.000000      0.0      0.0      0.0      0.0      0.006064   \nstd       13.565021      0.0      0.0      0.0      0.0      0.528828   \nmin        0.000000      0.0      0.0      0.0      0.0      0.000000   \n25%       11.000000      0.0      0.0      0.0      0.0      0.000000   \n50%       23.000000      0.0      0.0      0.0      0.0      0.000000   \n75%       35.000000      0.0      0.0      0.0      0.0      0.000000   \nmax       46.000000      0.0      0.0      0.0      0.0     63.000000   \n\n                6             7             8             9    ...      775  \\\ncount  18800.000000  18800.000000  18800.000000  18800.000000  ...  18800.0   \nmean       0.019734      0.011277      0.009362      0.011277  ...      0.0   \nstd        1.761298      1.054441      0.892318      1.517275  ...      0.0   \nmin        0.000000      0.000000      0.000000      0.000000  ...      0.0   \n25%        0.000000      0.000000      0.000000      0.000000  ...      0.0   \n50%        0.000000      0.000000      0.000000      0.000000  ...      0.0   \n75%        0.000000      0.000000      0.000000      0.000000  ...      0.0   \nmax      213.000000    108.000000     95.000000    208.000000  ...      0.0   \n\n           776      777      778      779      780      781      782      783  \\\ncount  18800.0  18800.0  18800.0  18800.0  18800.0  18800.0  18800.0  18800.0   \nmean       0.0      0.0      0.0      0.0      0.0      0.0      0.0      0.0   \nstd        0.0      0.0      0.0      0.0      0.0      0.0      0.0      0.0   \nmin        0.0      0.0      0.0      0.0      0.0      0.0      0.0      0.0   \n25%        0.0      0.0      0.0      0.0      0.0      0.0      0.0      0.0   \n50%        0.0      0.0      0.0      0.0      0.0      0.0      0.0      0.0   \n75%        0.0      0.0      0.0      0.0      0.0      0.0      0.0      0.0   \nmax        0.0      0.0      0.0      0.0      0.0      0.0      0.0      0.0   \n\n           784  \ncount  18800.0  \nmean       0.0  \nstd        0.0  \nmin        0.0  \n25%        0.0  \n50%        0.0  \n75%        0.0  \nmax        0.0  \n\n[8 rows x 785 columns]","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>0</th>\n      <th>1</th>\n      <th>2</th>\n      <th>3</th>\n      <th>4</th>\n      <th>5</th>\n      <th>6</th>\n      <th>7</th>\n      <th>8</th>\n      <th>9</th>\n      <th>...</th>\n      <th>775</th>\n      <th>776</th>\n      <th>777</th>\n      <th>778</th>\n      <th>779</th>\n      <th>780</th>\n      <th>781</th>\n      <th>782</th>\n      <th>783</th>\n      <th>784</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>count</th>\n      <td>18800.000000</td>\n      <td>18800.0</td>\n      <td>18800.0</td>\n      <td>18800.0</td>\n      <td>18800.0</td>\n      <td>18800.000000</td>\n      <td>18800.000000</td>\n      <td>18800.000000</td>\n      <td>18800.000000</td>\n      <td>18800.000000</td>\n      <td>...</td>\n      <td>18800.0</td>\n      <td>18800.0</td>\n      <td>18800.0</td>\n      <td>18800.0</td>\n      <td>18800.0</td>\n      <td>18800.0</td>\n      <td>18800.0</td>\n      <td>18800.0</td>\n      <td>18800.0</td>\n      <td>18800.0</td>\n    </tr>\n    <tr>\n      <th>mean</th>\n      <td>23.000000</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.006064</td>\n      <td>0.019734</td>\n      <td>0.011277</td>\n      <td>0.009362</td>\n      <td>0.011277</td>\n      <td>...</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>std</th>\n      <td>13.565021</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.528828</td>\n      <td>1.761298</td>\n      <td>1.054441</td>\n      <td>0.892318</td>\n      <td>1.517275</td>\n      <td>...</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>min</th>\n      <td>0.000000</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>...</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>25%</th>\n      <td>11.000000</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>...</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>50%</th>\n      <td>23.000000</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>...</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>75%</th>\n      <td>35.000000</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>...</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>max</th>\n      <td>46.000000</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>63.000000</td>\n      <td>213.000000</td>\n      <td>108.000000</td>\n      <td>95.000000</td>\n      <td>208.000000</td>\n      <td>...</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0.0</td>\n    </tr>\n  </tbody>\n</table>\n<p>8 rows × 785 columns</p>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"X_test = test_df.loc[:, 1:]\ny_test = test_df.loc[:, 0]\n\nX_test.shape, y_test.shape","metadata":{"trusted":true},"execution_count":34,"outputs":[{"execution_count":34,"output_type":"execute_result","data":{"text/plain":"((18800, 784), (18800,))"},"metadata":{}}]},{"cell_type":"code","source":"X_test = np.apply_along_axis(reshape_and_rotate, 1, X_test.values)\ny_test = np_utils.to_categorical(y_test, number_of_classes)\n\nX_test.shape, y_test.shape","metadata":{"trusted":true},"execution_count":35,"outputs":[{"execution_count":35,"output_type":"execute_result","data":{"text/plain":"((18800, 28, 28), (18800, 47))"},"metadata":{}}]},{"cell_type":"code","source":"X_test = X_test.astype('float32') / 255","metadata":{"trusted":true},"execution_count":36,"outputs":[]},{"cell_type":"code","source":"X_test = X_test.reshape(-1, W, H, 1)\nX_test.shape","metadata":{"trusted":true},"execution_count":37,"outputs":[{"execution_count":37,"output_type":"execute_result","data":{"text/plain":"(18800, 28, 28, 1)"},"metadata":{}}]},{"cell_type":"code","source":"model.evaluate(X_test, y_test)","metadata":{"trusted":true},"execution_count":38,"outputs":[{"name":"stdout","text":"588/588 [==============================] - 1s 2ms/step - loss: 0.3747 - accuracy: 0.8701\n","output_type":"stream"},{"execution_count":38,"output_type":"execute_result","data":{"text/plain":"[0.3747466206550598, 0.8700531721115112]"},"metadata":{}}]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]}]}