GSoC 2014 - Practice handwriting app

[Katarzyna Baraniak]

Hello,

My name is Katarzyna Baraniak from Poland. I am currently studying for my master’s degree in Information Technology at Polish-Japanese Institute of Information Technology in Warsaw. My specialization on masters is Intelligent data processing systems. I graduate engineering studies in IT with specialization Network of mobile devices.

I am interested in your project ‘Practice handwriting’ application. In the past I have made Windows Phone application based on handwriting:

http://www.windowsphone.com/pl-pl/store/app/kanji/6bc5c7e5-c6da-4c93-ae7d-a3248dab1289

This app helps to recognize Japanese kanji. I implement simple algorithm which takes four points from drawn line of kanji, scales it and calculate the distance to points from saved samples with the same numbers of lines. If the distance is enough small to one of the samples it shows its meaning. I also implement apps on Android OS on my own and as my engineering project which was multiple access game in augmented reality. I give a link to it in my proposal. I worked on android CRM applications during my summer internship. Also I attended an internship in Samsung R&D as an engineer of artificial intelligence where i worked in project connected with handwriting recognition.

 
I considered two OS for his project Android and Windows.  I choose first one because there is more possibilities on it and more tools to extend apps. My first idea to implement this project is simple. What I want to do is an application helping kids to learn handwriting in easy way and accurate to their age, by drawing characters on screen. Firstly I want to focus on User Interface. It’s important to carefully prepare what app should look like to meet the expectation of kids. It should be intuition as much as possible for kids who cannot read and colorful to make them attracted. Then I would implement the rest. I wonder if it’s better to implement algorithms by my own or use ready library? For example LipTk which supports Android and contains algorithms recognizing different shapes like gestures and characters and i think it's suitable for this project. App should have different levels of difficult and possibility to save progress measure by correct drawings or percent of covering correct shape.

I came up with idea of second version of this app. I could make it in augmentedreality.  What I mean is the shape of characters can be displayed on the screen when camera is focused on paper with marker. Kids write character strictly on paper and app will check if it's correct. It's challenging task but i have experience with Artoolkit and Vuforia, tools for augmented reality. Do you think it is good idea? Your response will help me to decide which way to choose and I can finished my proposal with detailed work plan.

Best regards,
Katarzyna Baraniak

[Pankaj Nathani]

Hi Katarzyna

Thanks for your interest. 

 

I am interested in your project ‘Practice handwriting’ application. In the past I have made Windows Phone application based on handwriting:

http://www.windowsphone.com/pl-pl/store/app/kanji/6bc5c7e5-c6da-4c93-ae7d-a3248dab1289

This app helps to recognize Japanese kanji. I implement simple algorithm which takes four points from drawn line of kanji, scales it and calculate the distance to points from saved samples with the same numbers of lines. If the distance is enough small to one of the samples it shows its meaning.

It's an interesting (and simple) logic. I don't know Japanese, however my guess is that it works! What's the accuracy of recognising the letter? 

 I wonder if it’s better to implement algorithms by my own or use ready library? For example LipTk which supports Android and contains algorithms recognizing different shapes like gestures and characters and i think it's suitable for this project. App should have different levels of difficult and possibility to save progress measure by correct drawings or percent of covering correct shape.

We are fine with either. If you use a freely available library or write your own implementation - upshot is that it should work. So feel free to include libraries in your proposal.

I came up with idea of second version of this app. I could make it in augmentedreality.  What I mean is the shape of characters can be displayed on the screen when camera is focused on paper with marker. Kids write character strictly on paper and app will check if it's correct. It's challenging task but i have experience with Artoolkit and Vuforia, tools for augmented reality. Do you think it is good idea?

I think it's a good idea as well. It adds a fun factor for the kids. At the same time, it does add one more step - i.e. to take a photo. May be this is something we can try if there's time left after implementing the original idea.

 

Your response will help me to decide which way to choose and I can finished my proposal with detailed work plan.

Hope that helps to clarify your queries :-)