simple_cnn_model.py

model = Sequential([
  # feature scaling
  layers.Rescaling(1./255, input_shape=(img_height, img_width, 3)),

  # feature extraction
  layers.Conv2D(160, kernel_size = 7, padding = 'valid', strides = (2, 2), activation = 'relu', input_shape = (img_height, img_width, 3)),
  layers.MaxPooling2D(pool_size = (3, 3)),
  layers.Conv2D(240, kernel_size = 5, padding = 'valid', strides = (1, 1), activation = 'relu'),
  layers.MaxPooling2D(),
  layers.Conv2D(320, kernel_size = 3, padding = 'valid', strides = (1, 1), activation = 'relu'),
  layers.MaxPooling2D(),  
  layers.Conv2D(480, kernel_size = 3, padding = 'valid', strides = (1, 1), activation = 'relu'),
  layers.MaxPooling2D(),  

  # classification
  layers.Flatten(),  
  layers.Dense(128, activation='relu'),
  layers.Dense(num_classes)
])