Digit recognition on my own handwritten dataset using trained model from MNIST

Question

I am trying to see if I can train my own CNN model on MNIST digits to classify my own handwritten digits. The model achieves a 98% train and test accuracy on MN