3 pages of drama

11/14/2005 8:24:59 PM

Quote :

"lots of high school classes already do a simple version of this (tracking checkers moves) with a legos robot camera. "

Link? I found some college people that did it in a very hacked way, no high school kids. Thanks for trying.

11/14/2005 9:28:13 PM

we did it at c.e. jordan high school in durham, it was part of educational kit so we cant be the only ones
i dont know how to link that but heres what it was http://mindstorms.lego.com/eng/products/vc/index.asp

11/14/2005 9:33:52 PM

Funny how everyone has these grand ideas but you ask them to deliver real links with real examples and they come up limp dick.

I don't need a link to the damn camera, I just told you I found a page that uses it. But they were college kids and it was a rather complex way to do it, and this is just checkers mind you, not chess.

11/14/2005 9:36:59 PM

well im sorry the teacher never made a webpage

11/14/2005 9:40:06 PM

i'm starting to think that anyone who doesn't do it my way is probably trying to be overly complex.

11/14/2005 10:27:54 PM

mine is very simple

and i'm not sure why state409c needs a link - i just explained how it works.... maybe if i make a separate webpage with the same explanation and link to it, he'll buy it?

eh whatever.

shadowrunner, your solution is pretty good.

JonHGuth, your solution of having a top-down camera is fucking stupid. stop advocating it.

but having a camera that looks up at the pieces from below is actually feasible and is a worthwhile solution.


[Edited on November 14, 2005 at 10:56 PM. Reason : s]

11/14/2005 10:55:11 PM

i haven't read any of this thread

but i would go with jonhguth's ideas

he's usually spot on

11/14/2005 10:59:18 PM

Quote :

"JonHGuth, your solution of having a top-down camera is fucking stupid. stop advocating it."

bottom up, not top down
colors distinguish peices



[Edited on November 14, 2005 at 11:08 PM. Reason : like i said from the beginning]

11/14/2005 11:07:36 PM

real links with real examples:

http://www.johncage.info/workscage/reunion.html
http://www.toutfait.com/duchamp.jsp?postid=1457&keyword=
http://members.chello.nl/j.seegers1/bib_duchamp/music_duchamp.html

Checkers playing robot:

http://www.comp.nus.edu.sg/~g0502275/Robot.htm

[Edited on November 14, 2005 at 11:32 PM. Reason : -]

11/14/2005 11:29:50 PM

aha

like i said, i didn't really read the thread before

but you're trying to ape john cage

?!

gtfo

11/14/2005 11:33:38 PM

it would be innovating on his original concept (and Lowell Cross' design)

our chess board would generate its own music

Cage's simply processed a live performance

11/15/2005 12:01:51 AM

11/15/2005 12:32:50 AM

i will eat my own shit in the middle of the brickyard if you follow through with this 100%

11/15/2005 12:38:06 AM

http://www.priorartdatabase.com/IPCOM/000128875/

Method for using a webcam to automatically read moves played on a chess board.
Disclosed by IBM.

you best be worried

11/15/2005 1:04:42 AM

thats more like what our checker thing did, we had to manually take a picture after every move

11/15/2005 1:07:37 AM

Chess Autonomous Robot
http://www.seas.upenn.edu/~timothee/papers/ChessAutonomousRobot.pdf

TINA (an open source environment developed to accelerate the process of image analysis research)
http://www.tina-vision.net/

Computer Chess Programming Theory
http://www.frayn.net/beowulf/theory.html

Chess Recognition and Tracking
http://visl.technion.ac.il/projects/2004w29/

Visual Chess (very helpful)
http://homepages.inf.ed.ac.uk/s0346435/projects/chess_project/visual_chess_project.htm

Software and Data Sets for “2D Object Detection and Recognition”
(includes Chess)
http://galton.uchicago.edu/~amit/book/

11/15/2005 1:22:48 AM

Quote :

"A. Kumar, Santanu Chaudhury and Subhashis Banerjee.. Understanding Chess Board Positions through Visual Search: A Case study in Active and Purposive Vision. Proc. Indian Conference on Pattern Recognition, Image Processing and Computer Vision, IIT Kharagpur. pp. 275-280, December, 1995"

my ideas are definitely not plausible

seriously, you guys are dicks

11/15/2005 1:25:23 AM

no one said your ideas weren't plausible

i just don't think YOU have what it takes pull something like this off

you're comparing the abilities of you and your stoner flunkies to that of a man who attended the "Indian Conference on Pattern Recognition, Image Processing and Computer Vision"

give me a break

this is a fucking joke

oh wait

the post before that lists a bunch of shit done by others

so like i said before, your plan is just to copy the work already done by someone else

WOW YOU ARE A FUCKING WIZARD

this shit won't even get off the ground

stop fooling yourself

i don't know you personally, but aren't you an english major at unca

did you flunk out of state

don't you have one of this sparks every few weeks

have any of them actually gone anywhere

give it up man

give it up

11/15/2005 1:50:31 AM

I'll just state that I've tried to be nothing but helpful in this thread, Nick, so hopefully I'm not getting lumped in with the haters. I think your idea is very plausible, but unless one of the guys on your team is already intimately familiar with implementing one of the visual processors via camera, I think my idea would be a whole lot easier.

And Excoriator, I understand your idea and think it's a good one as well, I've just never liked the idea of having to program in tolerances and error margins if there's another approach that wouldn't need them.

11/15/2005 3:19:59 AM

Quote :

"bottom up, not top down colors distinguish peices"

sorry - seemed like you were talking about top down. yours is a good idea.

basically we have three good ideas, here, snewf. they each have their own benefits and drawbacks. I'll summarize:

JonHGuth
pros: very clean and simple to implement
cons: very sophisticated programming required (at least for a hobbyist type project)

Excoriator
pros: very easy to program and debug
cons: harder to implement mechanically
cons: large number of discrete components

Shadowrunner
pros: originality
cons: hard to implement mechanically
cons: hard to design electrically
cons: large number of discrete components


these are just my own impressions and please don't assume that the quantity of pros/cons correlates to the idea's merit because in my opinion, the difficulty of programming something to read a webcam is harder than all the cons of the other two ideas combined. but if you're a CSC major you'll probably have a much different perspective.

note: i didn't list "move checking" as a pro under shadowrunner's idea because any of the ideas can implement that in the main CPU because they will all, by design, know the state of the board.

I know i didn't list any good pros for your idea, but that's because i couldn't think of any. feel free to add your own.

[Edited on November 15, 2005 at 8:33 AM. Reason : s]

11/15/2005 8:31:59 AM

so i thought about this during my commute and I am now a full convert to JonHGuth's idea. its clean, simple to implement, and now that i've thought about it, the programming isn't nearly as difficult as I originally thought it would be. All you'd need is the jpeg spec and you're set.

My one concern though would be ensuring that the camera and PC connection can update images at 1Hz or better - and your program would also have to be pretty fast (not really difficult to do).

11/15/2005 9:59:46 AM

Only it isn't simple to implement.

11/15/2005 10:39:26 AM

snewf nobody is denying that someone else couldn't complete this project - we all agree it can be done relatively easily - we are saying that because it's your idea it will never see the light of day

11/15/2005 10:51:34 AM

Quote :

"Only it isn't simple to implement."

why not.

11/15/2005 10:58:08 AM

For starters, remind me again how a camera sees through the black?

Unless we are talking about some sort of hack to the board such that the whole square isn't colored in black.

11/15/2005 11:13:29 AM

i was assuming you could cheat and not color the entire square black
if you need black squares i would try tinting them some and seeing if that allowed enough light through

11/15/2005 11:18:17 AM

^^ that is a trivial problem. if you're going to dismiss his idea out of hand for being too complicated to implement, then you've gotta back it up with real concerns.

11/15/2005 11:26:32 AM

Its a real concern.

I am being specifically vague about my concerns as you guys are being specifically vague about how easy you think this solution would be.

I'll address my very valid concerns after lunch. Maybe.

11/15/2005 11:34:27 AM

if worse came to worse, you could create little indentations with holes that the pieces fit into - that would greatly reduce light leakage and ensure a blank canvas that your program would compare onto

11/15/2005 11:36:06 AM

most of the implementation I've seen (by people with Masters degrees and Ph.Ds) used a camera suspended above the board

and open source software has been written for much more complicated visual recognition problems

I don't think tracking the game is the most difficult problem anymore

converting the data we collect into music that is pleasing to the ear remains the most challenging hurdle

11/15/2005 12:52:14 PM

^ you're in for a world of hurt if you try to do top-down image recognition. existing open-source programs be damned.

11/15/2005 12:57:53 PM

Attn:

Snewf, you are divorced from this debate.

11/15/2005 1:04:41 PM

Excoriator may be a fag, but he's right

[Edited on November 15, 2005 at 1:05 PM. Reason : about visual recognition being a bitch]

11/15/2005 1:05:13 PM

My guess is that the implementations you've seen done with a top-down camera were done that way because the projects were implemented for a masters' or PhD project on visual recognition, NOT because that is the simplest way to accomplish the task at hand. I assume chess was simply a fun application of the technology being worked on in their project, and was also convenient for a student-level project, given that the visual recognition being used is relatively basic (ie. two discrete and contrasting colors on two discrete and contrasting backgrounds, with standardized and easily distinguishable features on each kind of piece).

I would then submit that the fact you've found more implementations by masters' and PhD students using a top-down camera approach, probably means that this is indeed the most difficult way to do it. If it was the simplest way to approach recognition of chess pieces, it would be way too simple to be worthy of a PhD dissertation.


Regarding JonHGuth and Excoriator's designs, I think State409c does have a valid concern. The board would likely need to be altered in some way so that the camera identifies an individual piece correctly under both a "white" and "black" square (or red and black, whatever the board colors are).

[Edited on November 15, 2005 at 2:18 PM. Reason : ]

11/15/2005 2:15:32 PM

the board is going to be altered under EVERY one of the three working solutions that have been proposed.

second,

Quote :

"if worse came to worse, you could create little indentations with holes that the pieces fit into - that would greatly reduce light leakage and ensure a blank canvas that your program would compare onto"

11/15/2005 2:56:19 PM

I realize that... mine would probably require a completely custom-made board with the circuitry in it. I'm just saying that his point was pretty valid, and it's not as trivial a concern as you're making it out to be.

11/15/2005 3:04:36 PM

it IS trivial from the perspective of engineering.... now we're getting into "shop class" type discussions. The very fact that I have to explain this is evidence of how simple the engineering is behind this solution. you would use a table-saw to score the squares and a drill press with large bit to indent the sockets for the pieces, then a smaller bit to create a hole for the camera to see up into the piece.

wtf is so difficult - if you're gonna say "oh god this isn't simple" then BACK IT UP - what do you think is going to be difficult?

11/15/2005 3:09:23 PM

I dunno what battles you guys think you're winning arguing about something this simple for 3 pages, maybe you're too close to the problem to realize this, but: all of you look like idiots for it

11/15/2005 3:12:09 PM

i'm not arguing, but i am frustrated that you guys are saying that a simple solution is not simple - without mentioning what part seems difficult to you

11/15/2005 3:13:33 PM

Quote :

"then a smaller bit to create a hole for the camera to see up into the piece. wtf is so difficult - if you're gonna say "oh god this isn't simple" then BACK IT UP - what do you think is going to be difficult? "

It isn't simple.

11/15/2005 3:20:42 PM

eh ok. i guess it wouldn't be simple for you, then. i won't argue that.

11/15/2005 3:21:42 PM

Haha, its funny you want people to address whey they don't think it is easy, yet you think drilling a little hole is enough detail to fully address the solution and it doesn't even come close.

What kind of camera are we talking here? Web camera?

11/15/2005 3:22:53 PM

if you're wanting to straight up drill holes in the board, then yeah, it's trivial. i was just thinking of doing something more aesthetically similar to a normal chess board; there you would have problems even if you used a glass board with translucent and frosted squares. when you first mentioned an indentation, i thought you just meant drilling out part of the board's thickness on each square so there was less board for the light to shine through, but yeah, if you're completely making a small hole in the middle of each indentation, that should be fine. misunderstanding of what you were trying to say.

and seriously, incognegro, we might sound silly discussing this, but these kind of threads are really what i think Tech Talk should be about. do you have any idea how many times you've come across sounding like a smartass with no social skills whining about how smart you are and why can't anyone see how much better your ideas are and omg you're such a dumb n00b for even thinking that could possibly work? these threads are infinitely better than another OMFG SPYWARE HELP PLZ or HOW I BURN DVD?!? thread.

[Edited on November 15, 2005 at 3:27 PM. Reason : ]

11/15/2005 3:24:54 PM

The real irony is he becomes part of the "battle" that he is trying to laugh and call people out for.

Maybe he realizes this, he probably doesn't.

Next Tech Talk project : Automated pizza delivery route optimization.

Next next Tech Talk project : Bandpass boxes and why they are useless

11/15/2005 3:34:03 PM

Quote :

"i thought you just meant drilling out part of the board's thickness on each square so there was less board for the light to shine through, but yeah, if you're completely making a small hole in the middle of each indentation, that should be fine. misunderstanding of what you were trying to say."

word.

Quote :

"yet you think drilling a little hole is enough detail to fully address the solution and it doesn't even come close."

if you'd like to ask me how to implement a specific piece of the solution, i'd be happy to explain. however, i'm not gonna post a huge description of the entire process because you're just as likely to be trolling me and will merely respond "its not that simple".

so, please, ask me a specific question and i'll give you a specific response. until then, i'll just asume you to be either trolling or unable to comprehend this particular approach. which is fine - some people are more visual and need to have things shown to them... unfortunately this medium doesn't really lend itself to that sort of teaching.

[Edited on November 15, 2005 at 3:45 PM. Reason : s]

11/15/2005 3:36:44 PM

So wait. You don't mind coming off as this pedantic asshole with the perfect simple solution, so long as it doesn't require work. But then when it is inquired, you use trolling as a defense for being lazy?

I'm not trolling at all here. You are dead confident on the simplicity of this solution, it should be a trivial few minutes post to discuss some of the finer details and potential pitfalls for a camera based solution. WHY IS THIS SO HARD? WHAT TYPE OF CAMERA would be a starting point as this is the main constraint to begin with. I can think of several problems right now depending on what type of camera is used and it isn't even my idea. Why can't you?

11/15/2005 4:01:23 PM

it's not really the rigor you're treating this simple problem with that strikes me, it's the fact that at least two of you are behaving like children in the debate

u cite ur sources first

NUH-HUH I'M BEING INTENTIONALLY VAGUE ILL GIVE U TILL LUNCH 2 COME UP W/ SOMETHING BETTAR LOL

u butt face ur an insult 2 the proud tradition of ingineering

11/15/2005 4:31:38 PM

wtg guys on keeping the thread on topic

11/15/2005 4:32:53 PM

EXCORIATOR STARTED IT!!!!!

11/15/2005 4:36:42 PM