Since the application system can only accept files of limited types, I've upload my project to Github. You can retrieve it at ryanxiauic/HandwritingRecognition · GitHub

The project is a preliminary demo of our OCR system. The aim of the project is to identify handwritten letters, numbers and grading marks. Serveral CNN models have been trained to recognize and distinguish the handwrittings.

To utilize the project, first please prepare your python environment accoding to the Pipfile. Then, you can launch flask api server in main.py. Several tests can be found in the jupyter notebook. The model is based on convolutional neural networks. I remember that the model stucture was partly refers to an example in Professor Andrew Ng's course but with extesion of recognization of grading marks, ABCD and some special case.