RasLearnSignCNN.py

# -*- coding: utf-8 -*-
from timeit import repeat
from keras.models import Sequential
from keras.layers import Dense, Dropout, Activation, Flatten, Activation
from keras.layers import Conv2D, MaxPooling2D
from keras.optimizers import SGD, Adam, RMSprop
from keras.utils import np_utils
from sklearn import model_selection
import numpy as np
import time
import os
import random
imgW, imgH = 100, 100
nClass = 86436

def buildModelCNN():
  # CNNのモデルを構築
  model = Sequential()
  model.add(Conv2D(32, (3, 3), input_shape=(imgW, imgH, 3), activation='relu'))
  model.add(Conv2D(32, (3, 3), activation='relu'))
  model.add(MaxPooling2D(pool_size=(2, 2)))
  model.add(Conv2D(64, (3, 3), activation='relu'))
  model.add(MaxPooling2D(pool_size=(2, 2)))
  model.add(Dropout(0.25))
  model.add(Flatten())
  model.add(Dense(128, activation='relu'))
  model.add(Dropout(0.25))
  model.add(Dense(nClass, activation='softmax'))
  model.compile(loss='categorical_crossentropy',
    optimizer=Adam(),
    metrics=['accuracy'])
  return model

if __name__ == '__main__':
  import matplotlib.pyplot as plt
  signFileName = "./assets/dataset/X100/RasSignImg_X100.npz"
  if not os.path.exists(signFileName):
    print("Sign File does not exist")
    exit()
  saveModelFile = "./assets/model/RasSignModelCNN_3.h5"
  sTime = time.time()
  # フォント画像のデータを読む
  # rng = np.random.default_rng()
  xy = np.load(signFileName)
  # xy = rng.shuffle(xy,axis=0)
  X = xy["x"]
  Y = xy["y"]
  print(X.shape)
  print(Y.shape)
  X = X[:3000,:]
  Y = Y[:3000]   
  print(X)
  print(Y)
  print(X.shape)
  print(Y.shape)
  # データを正規化
  X = X.reshape(X.shape[0], imgW, imgH, 3).astype('float32')
  X /= 255
  Y = np_utils.to_categorical(Y, nClass)
  # 訓練データとテストデータに分割
  X_train, X_test, y_train, y_test = model_selection.train_test_split(X, Y,test_size=0.25)
  # モデルを構築
  model = buildModelCNN()
  history = model.fit(X_train, y_train,
                                  batch_size=1024, epochs=5, verbose=1,
                                  validation_data=(X_test, y_test))
  # モデルを保存
  model.save(saveModelFile)
  model.summary()
  # モデルを評価
  score = model.evaluate(X_test, y_test, verbose=0)
  print("Test Loss = {},  Test Accuracy = {}".format(score[0], score[1]))
  print('Conputation Time = ', time.time() - sTime)
  # 学習過程をグラフ化
  plt.figure(1, figsize = (12, 3))
  plt.subplots_adjust(wspace=0.5)
  plt.subplot(1,2,1)
  plt.plot(history.history['loss'], 'black', label='training')
  plt.plot(history.history['val_loss'], 'red', label='test')
  plt.legend()
  plt.subplot(1,2,2)
  plt.plot(history.history['accuracy'], 'black', label='training')
  plt.plot(history.history['val_accuracy'], 'red', label='test')
  plt.legend()
  plt.show()