How it Works

Inspiration

It is difficult for teachers to gage the classes’ true understanding of what they’re teaching in large classrooms. A single teacher can’t check the entire classes’ steps and solution to a math problem. And school districts are struggling to just fill teaching positions, so they can’t hire teaching assistants to aid in checking work

So, we want to make a platform that can provide a teacher with detailed, real-time insight on the classes' understanding of the content. We envision the teacher assigning a problem and having the students digitally write out their solution and steps. Then, we will use an optical character recognition model to convert the image to a sequence of strings, which will be analyzed by our mistake detection engine. This engine will output a summary of the classes' understanding of certain concepts involved in solving the problem.

Our ultimate goal is to provide equal opportunities to learn despite a students' socioeconomic status, location, or background.

An example output would look like: Understanding Metric=Percent of Students Correct Solution = 40% Factoring Mistake = 46% Arithmetic Mistake = 25% Applied Arithmetic Operation to Only one Side of Equation = 60%

What it does

So far, we have made a basic website with a canvas that can successfully convert digital handwriting to a string of text. This string of text then gets put through our mistake analysis engine which is capable of detecting factoring mistakes.

How we built it

We built this project using Express, Node JS, MathPix, HTML, CSS, and JavaScript.

Challenges we ran into

For the first 10 hours of hacking, we ran into many roadblocks because certain web frameworks didn't allow us to do what needed to be done. We explored using Swift, Django, and React Native. Eventually, we ended up implement the tools listed above.

Accomplishments that we're proud of

We can successfully detect factoring mistakes!

What we learned

We learned to be less ambitious with our goals and more realistic. Also, certain team members learned more about web development from the other members.

What's next for NUMERO

We have lots planned for NUMERO. We plan to detect more math mistakes, apply our solutions to calculus, and make our own optical character recognition model.