We talked about the naive bayes algorithm in my CS class and about how it can be used to classify images. I wanted to do something more complicated but I figured it would be easiest to work with something that was just in black and white and would be easy to find large samples for training/testing.

What it does

It was supposed to train a model based on the Naive Bayes algorithm for classifying images (in ascii text format) of handwritten letters (english: A-Z, a-z) and digits (0-9). However, I can't tell if it works on letters because I couldn't find a dataset to use (see challenges) but I have a version of it that works on just numbers.

How I built it

I built it using C++ and CLion

Challenges I ran into

I couldn't find a dataset of handwritten characters that I could use. There were some available but, even with the help of mentors, I couldn't find a way to extract the files that were available into a usable format.

Accomplishments that I'm proud of

Getting this to work with numbers!

What's next for Handwritten Character Recognizer

I'm hoping to actually get a dataset of images of characters to work with my project, which might mean adjusting the way my algorithm works (I could possibly create my own). After I get that to work, it would be cool to try and get it to work for recognizing strings of letters and words, not just isolated characters, so I could use it to turn pictures of handwritten words into text, or train it to recognize people's messy handwriting so it could turn it into something readable

Built With

Share this project: