Lua-gestures: effortless recognition of trained single-stroke gestures

[josip]

Hi there. I published this project today:

https://github.com/jmiskovic/lua-gestures

It is a library for training and recognizing single-stroke gestures, based on very efficient "$1 Unistroke Recognizer" algorithm. The library comes with basic set of gestures, and you can dynamically add your own ones.



This could come handy for those games where you conjure magic by drawing runes, or for mobile-based text entry, or for quick mouse/touch shortcuts inside any design tool.

I'm not author of original algorithm or of Lua port. I merely fixed some bugs, adapted the API to fit better with Love and made the simple demo. I hope somebody finds a use for it.

[darkfrei]

I like it! Can you attach a .love file?

[Gunroar:Cannon()]

Nice, this seems really useful!

[monolifed]

link to the original code for reference:
https://depts.washington.edu/acelab/pro ... index.html

[josip]

I've made some more improvements over last days.

Recognizer can now use oriented gestures. This makes it possible to distinguish between larger number of trained gestures, for example '7' and 'L' will be correctly identified.

Another option added to recognizer is ability to work with very thin gestures like 'I' and '-'.

LOVE demo app has now grown into a basic gesture studio. It can browse defined gestures, add new ones and export the collection to console.

The library documentation is improved.

[ReFreezed]

Nice! I'm having visions of new game projects.

[togFox]

I can see this being a lot of fun with tablets and mobile devices.

[monolifed]

* made it use flat arrays for points
* moved functions around (a lot)
* tried to reduce the creation of new (temp) tables
* noticed a typo in recognize score before return ("/" instead of "-")
* probably some other minor optimizations

https://github.com/monolifed/lua-module ... stures.lua