GSoC-Hand Writing Application Ideas

[Hüseyin Zengin]

Hi everyone, these are my ideas for Hand Writing Application just about consept. Please don't keep your comments for yourself, share it :)

‘Practice handwriting’ application

This application aims to create a funny environment for practicing handwriting for students who is trying to learn how to write. Practising on smartphones or tablets is more inviting for kids than using their pencils.

At least this must be an application that shows a guideline of characters,digits etc. to kids and wants to follow them. This is the least feature at starting but it will be a little boring when kid has improved his/her handwriting skill. Then, i think that there must be some additional features for this application.

Note that these are just about the end-user things to discuss, i will give technical details after some discussion on these. But i have all algorithm solutions for ideas given below.

1. First show only a character guideline, show anything “cute”(a football, a ladybug etc.) that follows that guideline. And ask kid to do the same. When trace is wrong, warn the kid and restart the exercise.

2. After some exercising with single big character, show guidelines of more than one random characters and want to trace them just like first exercise. But if kid fails in one of the characters even the last after completing others successfully, restart the whole exercise from beginning.

3.  Show a random English(just for example) word to practice. This feature looks like same with previous one but, showing some “real” words will help kid for getting familiar with words.

4. Show guidelines for character(s) as same as before but for this time not forever. after a interval guidelines will disappear, kid still must trace the guidelines but there must be some tolerant of course. normalizing the line drawn by kid a little will deal with little mistakes. There will also be a ranking from 1 to 5 stars based on speed, accuracy of tracing.

5. Application shows a character,word or something like these, but there will only borders for them, not like full guidelines just horizontal and vertical borders of word. Kid will try to write the word in these borders, we are not waiting for perfect words so there will be some backround algorithms for comparing the given input with asked output to rank kid. This will be the hardest practice for kid to succeed.

PS: if there is anything hard to imagine please just say, i can provide some drafting :)

[Hüseyin Zengin]

Hi again, no comments on my ideas? Can i assume they are perfect(just kidding). Nothing to say?

[Pankaj Nathani]

Hi Hüseyin,

Seems I missed out replying to you.. thanks for bumping the thread.

Firstly, who are you? It would be great if you could give a brief introduction about yourself and past work, any experience on mobile app development you have?

 

1. First show only a character guideline, show anything “cute”(a football, a ladybug etc.) that follows that guideline. And ask kid to do the same. When trace is wrong, warn the kid and restart the exercise.

So you mean having the kids trace cute objects before the alphabets to gauge their interest and show them how this works? Yes, that's a good idea.

 

2. After some exercising with single big character, show guidelines of more than one random characters and want to trace them just like first exercise. But if kid fails in one of the characters even the last after completing others successfully, restart the whole exercise from beginning.

Yes, also you may need to record if the trace is continuous. To begin with you may allow the kid to practice with a non continuos trace, but finally he needs to trace the right strokes. Also, what I explained depends on the character (and language) he is tracing for. Sure, when he goes outside or fails to trace successfully, let his try again i.e. restart.

 

3.  Show a random English(just for example) word to practice. This feature looks like same with previous one but, showing some “real” words will help kid for getting familiar with words.

Why random and not alphabetically? I would suggest first let the kid finish the whole sequence alphabetically and then let the random test begin.

 

4. Show guidelines for character(s) as same as before but for this time not forever. after a interval guidelines will disappear, kid still must trace the guidelines but there must be some tolerant of course. normalizing the line drawn by kid a little will deal with little mistakes. There will also be a ranking from 1 to 5 stars based on speed, accuracy of tracing.

Good idea, showing trace guidelines are important. 

5. Application shows a character,word or something like these, but there will only borders for them, not like full guidelines just horizontal and vertical borders of word. Kid will try to write the word in these borders, we are not waiting for perfect words so there will be some backround algorithms for comparing the given input with asked output to rank kid. This will be the hardest practice for kid to succeed.

Let's start with alphabets first, once we get the trace engine right for alphabets.. we can try words. For this GSoC project, we want the trace engine for alphabets done right if not anything else.

Regards,

Pankaj. 

[Hüseyin Zengin]

Hi Pankaj,

Thanks for helpful comments i will consider them while preparing my proposal :)

Here is my personal information related to this project,

I am a computer engineering student in Turkey, i was an amateur programmer while my high school education. Now i am working as a freelance devoleper while studiying. 

In 2010 me and two friends of mine developed a student agenda for Windows Mobile with C++/Qt, and participated to Infomatrix(which is an International Informatics Project Competition) at Romaina and won Silver Medal.

I participated to Google Code-in 2012, I contributed code to KDE but i was have to gave up becase of some private reasons before program ends. I have complated 12 tasks and get a 400$ paycheck BTW :)

These days, -as a freelancer- I developed an Android application which is downloaded and used nearly 50k times, it is mobile application of a service for buying bus and airplane tickets online. I can't send it here because of my bussiness agreement but i can send it privately if necessary, i can even provide some code examples. 

The company liked it and ask me to develop a Windows Phone version of it, i am developing it now, which is not 

obstacle for GSoC because i will deliver it in 2 weeks.

I am also developing a new project with one of my friend to help people for learning sign language interactively using Leap Motion(which is a narrow area motion detector for hand detecting, check: www.leapmotion.com). This project is supported by The Scientific and Technological Research Council of Turkey(url: http://www.tubitak.gov.tr/en) PS: this is a very long term project so it doesn't a obstacle for GSoC too.

I can say that I have quite experience with Java with Android, C#.Net(both desktop and mobile) and C++/Qt

I also have some experience with Machine Learning and signal processing, i think, this is a advantage for this project. For example when guidelines disappear kids can make some ignorable mistakes, filtering trace data is required for ignoring. 

[Pankaj Nathani]

Thanks for your introduction Hüseyin.. it's great to know that you took part in Google Code-in earlier. Feel free to ask any specific questions you may have or start working on a draft proposal, we have some guidelines here: https://github.com/BuildmLearn/BuildmLearn-Toolkit/wiki/Guidance-for-Students

[Hüseyin Zengin]

Hi again, sorry for replying late, Code-in was a great experience for me too. I have a spesific question I think that the trace engine that project needs, can read the boundaries of letters and words from open source font formats. This will give a great power to engine, for being scalable to other languages. What do you think about this? Should my proposal include that? I think this will require some hard work to create a library to read fonts etc. Do you think that does it worth to study?

[Pankaj Nathani]

Hi Hüseyin

 

Hi again, sorry for replying late, Code-in was a great experience for me too. I have a spesific question I think that the trace engine that project needs, can read the boundaries of letters and words from open source font formats. This will give a great power to engine, for being scalable to other languages.

Yes, that's a good idea. We are open to the trace engine accepting fonts (supporting english at least) or even a specific image formats that can be generated. But fonts would be preferred over generated images.

 

What do you think about this? Should my proposal include that? I think this will require some hard work to create a library to read fonts etc. Do you think that does it worth to study? 

Absolutely. Your proposal should include your thought process and approach towards development. Once you have a decent proposal draft ready, you can request a review if you want. See details here: https://groups.google.com/d/msg/buildmlearn/droaQTxy7aU/L1aQ_aJYxSUJ

[Hüseyin Zengin]

I will finish my draft tomorrow but i need to get one more specific opinion, should the trace engine directly deal with font(this needs much more work and cause some performance issue) or should it take a special processed data ? Thanks for comments :)

[Pankaj Nathani]

Hi Hüseyin
 

I will finish my draft tomorrow but i need to get one more specific opinion, should the trace engine directly deal with font(this needs much more work and cause some performance issue) or should it take a special processed data ?

We are ok with either as long as the approach is scalable for other popular languages except english. 

[Hüseyin Zengin]

Hi Pankaj,

Thanks again, i will be working on my draft, i will also try to give depth details about my implemation plans. See you soon