Society of Robots - Robot Forum
Electronics => Electronics => Topic started by: tuttut89 on December 30, 2006, 11:10:44 PM
-
I have looked on the official site and searched many forums so please don't burn me on this question :)
(and I did read the object detection question posted earlier)
I have a limited budget and would like to do some object recognition with a cmucam but I don't know which version I would need. I would like to run through a basic algorithm to detect a certain shape and then look at the center of the object. I read that the CMUCam2 accesses pixel data from the frame buffer, making it faster, but what does this mean? Will there be lower fps? If I were to do all of this calculating from a microcontroller, how fast would that microcontroller need to be? Last of all, would it be best to just save my money for a cmucam3?
I will appreciate any help, and I hope you have a Happy New Year! ;D
-
I would look at the AVRcam, which does more or less the same thing, but is open source, and costs somewhat less.
http://www.jrobot.net/Projects/AVRcam.html
- Jon
-
I have decided to purchase the CMUCam2, since it has faster frame dumps and I read somewhere that the CMUCam1 took forever to dump a frame.
-
As someone who has used the CMUcam1 on multiple robots, I would highly discourage anyone from using it.
It has serious firmware bugs, poor mounting holes, a misaligned camera lens from poor manufacture, and very little available source code. :(
I havnt used CMUcam2, but curious if it has the same problems . . . let us know how it goes.
-
hi, do anyone have object recognition program code for CMUcam3? or the step?
-
It doesnt have object recognition built in. Do you mean uploading video data to a PC that then does object recognition?
this page has source code that you might want:
http://www.cmucam.org/wiki/Software
You might also want to try roborealm:
http://www.roborealm.com/
and a tutorial on computer vision:
http://www.societyofrobots.com/programming_computer_vision_tutorial.shtml