Author |
Topic: ISCAN Tracker (Read 10655 times) |
|
Nathan Mundhenk
YaBB Full Member
HA!
  
 
Gender: 
Posts: 140
|
 |
ISCAN Tracker
« on: 10/14/02 at 20:59:39 » |
 Quote  Modify
|
Issue 1: WE recently recieved a new ISCAN tracker at Ilab. The camera output monitor shows 8 identical images (tiled) and the tilt does not seem to work on the camera. Has anyone (I guess anyone being Rob) had these issues? If not then I'll just call ISCAN tech suport or something. Laurent, maybe you could also take a look at it. Maybe I missed something. I have the Docs with me tonight, but i'll bring them back tomorrow. Other then that, the bugger is set up.
Issue 2: Should the stim. display run of the ISCAN computer or should we set up a seperate computer to control the second stim. monitor? Rob have you had NE experiance that might suggest one over the other.
Issue 3: How should we network the new lab location. Should we run the triple bond there? I had this weird idea that we could make a triple bond wireless access point and just get three wireless PCI cards for NE computer we put in that office. It would cost $$$, but might save time.
Thanks 
|
|
 Logged |
T. Nathan Mundhenk
http://www.nerd-cam.com
http://www.cool-ai.com
|
|
|
Laurent Itti
YaBB Moderator
YaBB God
iLab rocks!
Gender:
Posts: 550
|
|
Re: ISCAN Tracker
« Reply #1 on: 10/14/02 at 21:45:35 » |
Quote Modify
|
sounds good!
For those who did not know: this machine actually is not exactly ours, but is a shared piece of equipment that a group of about 15 neuroscience faculty just received. We are helping out with the setup, though. Come by room 13B (one of Prof. Elaine Andersen's psychophysics labs donated to the cause) and check it out!
I'll have a look at your various points tomorrow. Sounds promising overall. Now regarding the triple network... well, the point was to be fast (300Mbps) and wireless is slooooooow  A single wired connection will probably give us better throughput.
thanks for setting it up!
-- laurent
|
|
Logged |
|
|
|
Rob Peters
YaBB Senior Member
Gender:
Posts: 398
|
|
Re: ISCAN Tracker
« Reply #2 on: 10/15/02 at 08:33:19 » |
Quote Modify
|
on 10/14/02 at 20:59:39, Nathan Mundhenk wrote:| Issue 1: WE recently recieved a new ISCAN tracker at Ilab. The camera output monitor shows 8 identical images (tiled) and the tilt does not seem to work on the camera. Has anyone (I guess anyone being Rob) had these issues? |
|
Yes, we also get the tiled output on the eye camera monitor. They tell me at ISCAN that this is because they have to multiplex information onto a 60Hz video signal in order to get the faster 120Hz/240Hz temporal frequencies of their high-end systems. It would seem to me that there must be some sacrifice of spatial resolution then in this scenario... Or, maybe the system internally is doing everything at 120/240Hz at full spatial resolution, and they just tile the images for display purposes. In that case, I'm not really sure why they don't display it at 60Hz and just drop the extra frames for display purposes. in any case, the rest of their hardware/software knows how to automagically extract the proper spatial information out of the tiled video signal.
Quote:| Issue 2: Should the stim. display run of the ISCAN computer or should we set up a seperate computer to control the second stim. monitor? Rob have you had NE experiance that might suggest one over the other. |
|
I would say you definitely want a separate computer for stimulus display. ISCAN's DAQ software for running the eye tracker is a bit fascist about having total control of the computer that it's running on (e.g. it runs in DOS mode and doesn't like to be minimzed down onto the Windows task bar...) For communicating between stimulus display and eyetracker, it's not too hard to rig up a serial cable between the two machines to handle communication (just make sure it's a crossover or null modem cable, and not a straight serial cable). With this approach you can send a single byte from the stim computer to the ISCAN program to tell it when to start+stop recordering. This gives you good temporal synchronization between your stim program and the data sequence. I think you can also have the DAQ program send eye position information in real time to your stim program, in case you want to adjust the display based on the subject's eye position (but, see below about calibration...)
Quote:| Issue 3: How should we network the new lab location. Should we run the triple bond there? I had this weird idea that we could make a triple bond wireless access point and just get three wireless PCI cards for NE computer we put in that office. |
|
Unfortunately, we've had some problems using both networking and the DAQ program on the eyetracking compter. I think their docs say that the program may not work when there's a network connection running. In practice, we haven't had that specific problem. But, we have had triple interaction problems between ethernet/serial-port/DAQ. That is, if you have a network connection up, then the serial port doesn't work in the DAQ program, or something like that. It's a bit of a pain... 
Last point: you may want to do your own calibration rather than using the calibration built in to DAQ. That's because at best you can do a 9-point calibration with the built-in approach, and I found that to give less than satisfactory accuracy (i.e. errors of several degrees of visual angle). The problem is exacerbated as you use a larger field of view for the stimulus display. That's because the points that you're tracking on the eye are rotating on a roughly spherical surface (the eyeball), but you're mapping those points to a flat surface via the calibration. So a coarse sampling with only 9 points doesn't do very well...
A better approach is to generate your own calibration grid, with e.g. 25 or 50 points. Have subjects fixate each point in turn. Then you can do a post-hoc calibration by taking the raw P-CR-H and P-CR-V (that's the vector difference pupil minus corneal reflection, horizontal and vertical components) and using e.g. Matlab's griddata function to link those to the screen coordinates at which you displayed the fixation crosses. With this approach I was able to achieve average accuracy of < 0.25 degrees. You can still use DAQ's built-in calibration to get rough real-time feedback for the experimenter about the subject's eye position as the experiment is in progress.
Cheers,
Rob
|
|
Logged |
|
|
|
Nathan Mundhenk
YaBB Full Member
HA!
Gender:
Posts: 140
|
|
Re: ISCAN Tracker
« Reply #3 on: 10/15/02 at 12:48:24 » |
Quote Modify
|
Hey, I had an idea. Maybe since we are both working on the same gadget we could set up a CVS repository to store any widgets we develop for the ISCAN. That way we can share and what not. Sound reasonable?
|
|
Logged |
T. Nathan Mundhenk
http://www.nerd-cam.com
http://www.cool-ai.com
|
|
|
okosoft
YaBB Newbie
I love YaBB 1G - SP1!
Posts: 2
|
|
Re: ISCAN Tracker
« Reply #5 on: 02/28/10 at 02:17:46 » |
Quote Modify
|
Hi, guys!
I have a iScan Eye Tracking system, but I have no software for it.
Could you please tell me where can I download software?
Thank you very much.
|
|
Logged |
|
|
|
okosoft
YaBB Newbie
I love YaBB 1G - SP1!
Posts: 2
|
|
Re: ISCAN Tracker
« Reply #6 on: 04/21/11 at 01:58:11 » |
Quote Modify
|
Hi, guys.
I would like to notify you that I developed a software that connects to Iscan 400 and record video with point of eye view.
If you are interested in, you can contact me by email - okosoft@gmail.com
or from website http://www.eyetracking.ws
I use different types of eye trackers - head mounted and desktop mounted eye trackers and develope software for them.
|
| « Last Edit: 04/21/11 at 01:58:59 by okosoft » |
Logged |
|
|
|
|
Home
Help
Search
Members
Login
Register
Reply
Notify of replies
Send Topic
Print