Programs I've been writing

[Jesin]

This is a thread for me to talk about some programs I've been writing.

To start off, I've been working on edge detection in AI class. I started with this image: http://academics.tjhsst.edu/compsci/ai/vision/italy/italy.png

I converted it to grayscale, and then I ran it through a 3x3 median filter (ignoring the outer rim because I'm lazy like that) to reduce noise. Then I ran my own variation of Canny edge detection on it. The result is attached.

 

[Artifice Orisit]

I haven't done much regarding AI to real world interactions, but still I don't envy anyone trying to teach a computer to interpret fuzzy data in an accurate and dynamic way. Perhaps some sort of adaptive algorithm that filters the input based upon attributes such as image density & clarity.

I've being thinking about the applications of chat-bots in character-to-player interactions for interactive games. Namely the use of two chat-bots interacting within the virtual world where one of them is being influenced/directed by the player. The idea being to enable players to participate (if indirectly) in dynamic conversations with NPCs without enabling them to introduce topics the AI cannot cope with, e.g. current events in the real world.

 

[FusionKnight]

Visual recognition is one of the most complex problems in AI/robotics. I've done a little with this as well (combining sonar with stereoscopic images and dead-reckoning to navigate an obstacle course). It's amazing to think that our brains can complete this task almost from birth, without any intentional training of any kind...

 

[Oblivious]

Yeah it only really seems amazing when you start trying to program motor drivers in assembly with a central processor taking in information from other integrated sensors...

Then you look at a tiny ant crawling happily across your desk, avoiding obstacles with perfect motor coordination before a profound sense of indignation overtakes you.

 

[FusionKnight]

Yeah it only really seems amazing when you start trying to program motor drivers in assembly with a central processor taking in information from other integrated sensors...

Then you look at a tiny ant crawling happily across your desk, avoiding obstacles with perfect motor coordination before a profound sense of indignation overtakes you.

Exactly!!

 

[Jesin]

I found and fixed a major flaw in my program. It works much better now.

Also, I'm using a Gaussian filter instead of a median filter for noise reduction now.

I've attached the new results.

 

[Da Blob]

I found and fixed a major flaw in my program. It works much better now.

Also, I'm using a Gaussian filter instead of a median filter for noise reduction now.

I've attached the new results.

Crud, I'm coming at this issue from a different direction "Signal Theory" so I do not know if this old article, is relevant or of much use (just trying to be helpful)

http://www.apple.com/acg/pdf/g4wavelet.pdf

 

[Da Blob]

http://eprints.ecs.soton.ac.uk/16708/

 

[Jesin]

I asked Fukyo, and she suggested I try it on this image:

Here are the results. (I did not use the Gaussian filter this time, by the way.)