I am using Keras to create model :

def my_model():
  model = Sequential()
  model.add(Conv2D(60, (5,5), input_shape=(32,32,1), activation='relu'))
  model.add(Conv2D(60, (5,5), activation='relu'))
  model.add(MaxPooling2D(pool_size=(2,2)))

  model.add(Conv2D(30, (3,3), activation='relu'))
  model.add(Conv2D(30, (3,3), activation='relu'))
  model.add(MaxPooling2D(pool_size=(2,2)))
  #model.add(Dropout(0.5))

  model.add(Flatten())
  model.add(Dense(500, activation='relu'))
  model.add(Dropout(0.5))
  model.add(Dense(num_classes, activation='softmax'))
  ## Compile the model
  model.compile(Adam(learning_rate = 0.001), loss = 'categorical_crossentropy', metrics=['accuracy'])
  return model

Now save the model and train it :

model = my_model()
print(model.summary())
model.save('my_model.h5')
history = model.fit(X_train, y_train, epochs = 10, validation_data=(X_val, y_val), batch_size = 400, verbose = 1, shuffle = 1)

#### Now predict some image
pred = model.predict(img)
prediction = np.argmax(pred,axis=1)
print("predicted sign: "+ str(prediction))

‘model.fit()’ function is the most CPU intensive job. In a less-powerful platform it takes lot of time to train. Is it possible to save the trained model and just use it for the purpose of prediction ?

I know there is way to load the saved model :

from tensorflow.keras.models import load_model 
model = load_model('my_model.h5') 

However still I have to run ‘model.fit()’ which is CPU intensive. Is there any way to avoid model.fit() and load the model and use for prediction ?