HCMUS-Handwritting-Recognition

1/ Authors:

This is our final project for Introduction to Information Technology.

Our team:

1. Phuc, Song Dong Gia

2. Loi, Nguyen Minh

3. Thang, Nguyen Quang

2/ Environment

2.1. Python 3.7

Download at https://docs.conda.io/en/latest/miniconda.html

2.2. Visual Studio Code (VS Code):

Download at https://code.visualstudio.com/Download/

Install extension "Python" after installing VS Code.

2.3. Numpy Library:

pip install numpy

2.4. Matplotlib Library:

pip install matplotlib

2.5. cv2 Library

pip install opencv-python

3/ Prepare MNIST dataset

Download MNIST dataset at: http://yann.lecun.com/exdb/mnist/ and DO NOT UNZIP FILES.

The MNIST dataset contains 60,000 images used to recognise input numbers called train, and 10,000 images used to check if the algorithm is good or bad, called test. Every image has its label, respective to the number written in the image.

4/ Organise project

4 zips of MNIST dataset is in data subfolder.

5/ Before we start...

Run test_MNIST.py file to make sure MNIST dataset is successfully installed and set up.

6/ How do we recognise the numbers?

• Step 1: Vectorize all the images of train dataset and the input img.

• Step 2: Find the distance between input img and each img in train.

• Step 3: Sort all the distances in increasing order.

• Step 4: Choose k smallest value, called k nearest neighbours (KNN). k can be 50, 100, 500, etc. You can choose any value for it.

• Step 5: Count and find in k labels which label has the largest frequency. That is the number this algorithm guess.

7/ Run code

Run file main.py.

Run by this cmd: python main.py

8/ Optimize Speed

Use C++ code to increase speed.

8.1/ Prepare

You must have C++ compiler to compile C++ code.

Get the lib.hpp and lib.cpp files.

Run these command (I use GNU-GCC):

g++ -c -fPIC lib.cpp -o lib.o

g++ -shared lib.o -o lib.so

Or compile them by Visual Studio.

You now have a lib.so file. Keep this file and main_optimze.py file in same directory.

If you don't want to edit the library or you don't have a compiler, use mine instead of building by yourself.

8.2/ Let's Rock!

Run main_optimize.py file instead of main.py file.

The only difference between these files is main.py runs guess() function in Python, but in main_optimize.py, the guess() function calls the guess_optimize() function written by C++ in lib.so.