What it does

Kanroooo is a complete point-to-point software, which uses AI and automation to make online classes for early learners more fun and interactive. As for the teachers it lightens their load as the AI helps in assessment so that they can distribute appropriate attention.

Kanroooo has the following features:

  • A simple UI, making it extremely easy for the kids to understand. The UI consists of one or two canvas, and 2 buttons (recognize and clear).
  • AI system-The recognize buttons activate one of 2 AI systems present on the software, one can recognize numbers and the other English alphabets. The characters or doodles on the canvas are judged by the AI model and evaluated as right or wrong, along with capturing a photo, for the log file.
  • The clear button, clears the canvas for the next recognition.
  • Log Creation- Once after every recognition a screenshot is clicked of the AI evaluated material. This screenshot is used to make a pdf log, this pdf is mailed to the teacher for his/her evaluation. This entire process is automated, the child does not need to do anything.

How we built it

The software built using Python, and Deep Learning is at the heart of its code.

It uses a CNN model that is trained specifically to deal with human handwriting for both numerical digits and English alphabets. The canvas feature of the software enables the user to write directly using a mouse/stylus or a trackpad. Later the writing on the canvas is used to recognize the writing, to determine the correct answer is written or not.

The entire submission system is automated through an e-mail system to ensure submission of the entire thing.

Challenges we ran into

This project had a lot of advanced concepts to deal with, the most difficult of all was to create a usable CNN model. CNN models are often tricky to deal with, for that reason, it takes a lot of time to finally create a system that could be used in the project.

Another problem that we ran into, which took a lot of time to fix was to make sure only the content in the canvas was evaluated by the CNN model. A lot of the side code or button text was also evaluated.

Accomplishments that we're proud of

Building the entire system itself makes us as a group very proud. We want this software to bring in smiles on the faces of little kids, who should enjoy school even though they are forced inside their houses. Additionally, we want to help the poor teachers who are struggling to control these kinds on the normal online meeting platform.

From the technical side, getting a successfully trained and fully usable CNN model made us proud of that achievement. The entire software as a whole tries to ease the burden of early education.

What we learned

It was our first time dealing with OCR and Deep Learning, we are extremely pleased that we were able to deal with all this in such a small amount of time. It was a great learning experience, to learn a technology that could change the world.

What's next for Kanroooo

Kanroooo is an advanced system that uses Deep Learning at its core. The complete system is at a prototype stage as of now and has a long way before completely developing into full-fledged software. The good thing is that the prototype has complete and all functionality of the main software, symbolizing it will not be difficult to convert it into complete software.

Even at its infancy, Kanroooo is capable of analyzing a large set of handwriting in numerics and English alphabets. The future implementation will host multiple language scripts depending on the region it is used in and added features to recognize basic drawing expanding its functionality.

Kanroooo will be completely open-source and free software as it will be used for the good of the future generation of planet earth.

Stay tuned for Kanroooo2.0. :).

Built With

  • deeplearning
  • figma
  • primere-pro
  • python
  • smtp
Share this project: