GazePoint GP3 eyetracker and psychopy

[Martin Guest]

Hi,

I recently attended the psychopy for neuroscience course, and heard a talk on the eyetracking libraries from Michael Macaskill. He really sold the idea to me, to use the psychopy libraries as it has open standards. We recently purchased a GazePoint eyetracker, the website is here: http://www.gazept.com/ and although it's sampling rate is 60 hz it seems to work pretty well, particularly as its cost is quite cheap. Also getting data from it is pretty straight forward too, it uses something called "OPEN GAZE API", don't know how common this is, but there is pdf outlining how to interface with it. Essentially it just uses XML type commands, the pdf found here http://gazept.com/Publications/Gazepoint_API_v2.0.pdf. 

We already use it within psychopy, Jon Pierce helped me developed a gaze contingency script, using the aperture libraries, while I was at nottingham that we now use with the gazepoint gp3 eyetracker, but I believe integrating into psychopy eyetracking libraries would give use extra functionality.I am willing to do the hard work myself but would be most grateful for any advice on how to start\get going with this. Michael mentioned a YAML file that the psychopy interface uses, there was also mention of a specific eyetracking mailing list/group being setup. If there is I would be most keen to join in.

Thanks in Advance

Martin 

[Sol Simpson]

Hi Martin,

Adding support for GP3 tracker would be great!

Let me look over the information you supplied and I'll email you back with  suggestions on what might be the easiest way to create the common eye tracker interface for the gp3. 

In general,  to add support for an eye tracker to iohub, you would 

1) Add a directory off of psychopy.iohub.devices.eyetracker.hw package / folder for the new eye tracker type; you can look at the other existing implementations to get a sense of what the folder and file hierarchy is. I would probably suggest :

eyetracker.hw.gazepoint.gp3

as the folder to add, so the gp3 EyeTracker class implementation would be:

eyetracker.hw.gazepoint.gp3.EyeTracker

2) In the new gp3 folder, create a class called EyeTracker which has EyeTrackerDevice as its parent. Fill in the guts of the eyetracker class  stub so that the necessary eye tracker methods work with the VP3 and that it is generating appropriate iohub eye tracker events.

3) Create a default_eyetracker.yaml and supported_config_settings.yaml file in the gp3 folder, that should have any GP3 specific configuration setting / options that  need to be configurable by users of the device interface. These files would all go in the GP3 specific eyetracker module mentioned in 1).

Looking at the existing eye tracker implementations, and the eye tracker device documentation should help clarify all this a bit I think. It is a lot easier to do than it sounds as long as the gp3 has a client interface /communication protocol that is python friendly. ;)

Should be able to email you back mid next week with more input. Feel free to email me if you have any questions before then.

Thanks again,

Sol

[Martin Guest]

Hi Sol,

Would be great to work with you on this. The API does seem to be python friendly, there is an application called "gazepoint control" that needs to run in the background, without this  the eyetracker doesn't power on, i think it creates the connection too. But this is not an big issue.

I very simple program that comes with the application is here:

*******************************************

# Gazepoint Client Example

import socket

# Host machine IP

HOST = '127.0.0.1'

# Gazepoint Port

PORT = 4242

ADDRESS = (HOST, PORT)

s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)

s.connect(ADDRESS)

s.send(str.encode('<SET ID="ENABLE_SEND_CURSOR" STATE="1" />\r\n'))

s.send(str.encode('<SET ID="ENABLE_SEND_POG_FIX" STATE="1" />\r\n'))

s.send(str.encode('<SET ID="ENABLE_SEND_DATA" STATE="1" />\r\n'))

while 1:

    rxdat = s.recv(1024)    

    print(bytes.decode(rxdat))

s.close()

*******************************************

The USB device is treated as a TCP/IP socket. When you read the data in from the port, each line it is in the following format, 

<REC FPOGX="0.26148" FPOGY="0.67862" FPOGS="68.02086" FPOGD="0.14784" FPOGID="166" FPOGV="1" CX="0.34609" CY="0.07227" CS="0" />

REC at the start of the string means a recording, ACK at the start would mean an acknowledge to a command you have sent:

you can add more fields too, like pupil size etc.

to access the individual fields I use the split command say something like this:

data = rxdat.split('"')

then I use the appropriate index to get to the field i want, like x_coord = data[3] ... might be a better way of doing this btw, so open to any suggestions.

The gaze contingency program is below, the coords that the gazepoint uses is from 0 to 1, with top left 0,0 and bottom right 1,1.

*******************************************************

from psychopy import visual, event, core

import socket

# Host machine IP

HOST = '127.0.0.1'

# Gazepoint Port

PORT = 4242

ADDRESS = (HOST, PORT)

s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)

s.connect(ADDRESS)

w = 1280

h=1024

s.send(str.encode('<SET ID="ENABLE_SEND_COUNTER" STATE="1" />\r\n'))

s.send(str.encode('<SET ID="ENABLE_SEND_POG_FIX" STATE="1" />\r\n'))

s.send(str.encode('<SET ID="ENABLE_SEND_DATA" STATE="0" />\r\n'))

s.send(str.encode('<SET ID="ENABLE_SEND_USER_DATA" STATE="1" />\r\n'))

def calibration():

    s.send(str.encode('<SET ID="CALIBRATE_SHOW" STATE="1" />\r\n'))

    s.send(str.encode('<SET ID="CALIBRATE_START" STATE="1" />\r\n'))

flag='REC'

win = visual.Window([1280,1024], units='pix',allowStencil=True)

foreGround = visual.ImageStim(win, image='ForeGround.jpg')

backGround = visual.ImageStim(win, image='BackGround.jpg', size=[1280,1024])

aperture = visual.Aperture(win, size=[300,300], shape='square')

s.send(str.encode('<SET ID="ENABLE_SEND_DATA" STATE="1" />\r\n'))

calibration()

for frameN in range(600):

    data=s.recv(1024)

    if data[1:4] == flag: # make sure it’s a recording not an acknowledgement

        tempList = data.split('"')

        xRaw = float(tempList[3])

        yRaw=float(tempList[5])

        x = (xRaw-0.5)*w

        y=-1*(yRaw-0.5)*h

        print x, y

        aperture.disable()

        foreGround.draw()

        aperture.enable()

        aperture.setPos([x,y])

        backGround.draw()

        win.flip()

s.send(str.encode('<SET ID="ENABLE_SEND_DATA" STATE="0" />\r\n'))

*******************************************************

Cheers

Martin

[Martin Guest]

Hi Sol,

I downloaded the ioHub onto my computer, and started to look at the other folders, in eyetracker.hw. The eyetracker.py file for eyelink looks quite a big file. Which would be the best/simplest folder to look at? Also what would be the most important methods to fill in, with regards to the eyetracker class?

Thanks

Martin

On Thursday, June 12, 2014 10:56:58 PM UTC+1, Sol Simpson wrote:

[Sol Simpson]

will email you later tonight, sorry for the delay.

Sol

[Michał Kuniecki]

Hello Martin,

I noticed your entry on this group while checking the web for gazepoint-PsychoPy intergration. Apparently it seems that you have already managed to successfully combine gazepoint tracker with  PsychoPy. Therefore I wonder if you would be willing to share the knowledge on how to control gazepoint tracker from within PsychoPy software. This know-how would be of great value for our small team as we lack professional programming resources.  We tried to design some experiments using Ogama, which features support for this particular tracker,  but the experiment building capabilities of this software are very limited, not to mention that it is also quite unreliable.

Best regards

Michal Kuniecki PhD
Jagiellonian University
Institute of Psychology
Psychophysiology Lab
Al. Mickiewicza 3
31-120 Krakow, POLAND

[Jibo He]

Hi, Martin, Sol,

How is the quality of the GP3 eye tracker? I am considering purchasing an eye-tracker.

Thanks!

GP3 Desktop Eye Tracker

Complete eye tracking system. Top-of-the-line performance at a consumer friendly price point!

• Accuracy 0.5 – 1 degree of visual angle
• 60Hz sampling rate
• 5 point or 9 point calibration
• Easy to use and open standard API
• 25cm x 11cm  (horizontal x vertical) movement
• ±15 cm range of depth movement
• USB2.0
• Works with 22″ displays or smaller

Purchase includes: GP3 Eye tracker, Gazepoint Analysis Standard software, SDK, Fruit Ninja demo, sample Visual Studio code, USB cables and tripod. 1 year warranty & technical support. System Requirements: Intel Core i5 or faster, 4 GB RAM, Windows 7 or 8; Mac and Linux not currently supported Price: $795 $495 USD online purchase special

Quantity:

Gazepoint Analysis Software (UX Edition)

Gazepoint Analysis UX Edition eye tracking software. Has all the features of Gazepoint Analysis Professional plus Thinkaloud Voice and Webcam recording functionality. It’s perfect for UX studies or cognitive research eye tracking studies. UX Edition software includes:

• Heat Map
• Gaze Fixation Path
• Screen Capture / Image / Video / Web Multiple User Data Aggregation
• Dynamic Areas of Interest (AOIs)
• Image, Video and Statistics Export
• Thinkaloud Voice & Webcam Recording
• 1 year software updates and technical support

System Requirements: Intel Core i5 or faster, 4 GB RAM, Windows 7 or 8; Mac and Linux not currently supported Price: $1495 USD

Quantity:

 

Gazepoint Analysis Software (Professional Edition)

Gazepoint Analysis Professional Edition eye tracking software. Analysis Pro is an incredibly powerful yet easy to use software package that simplifies developing, deploying, and analyzing the results of eye tracking studies. Professional Edition software includes:

• Heat Map
• Gaze Fixation Path
• Screen Capture / Image / Video / Web Multiple User Data Aggregation
• Dynamic Areas of Interest (AOIs)
• Image, Video and Statistics Export
• 1 year software updates and technical support

System Requirements: Intel Core i5 or faster, 4 GB RAM, Windows 7 or 8; Mac and Linux not currently supported Price: $995 USD

Quantity:

Gazepoint Analysis UX Edition Product Bundles

Gazepoint GP3 Ultimate Bundle

Includes all the software and accessory add-ons we have available. Discounted when purchased as a bundle! If you need to perform any kind of analysis of gaze information we highly recommend the UX Edition software package. Analysis UX follows the design philosophy of the GP3, providing high end, research grade, performance with a simple and easy to use interface. Bundle purchase includes:

• GP3 Desktop Eye tracker
• Gazepoint Analysis UX Edition software
• Remote Viewer Add-On software
• VESA Screen Mount
• Laptop Mount
• 3 year warranty, software updates & technical support

Price: $2700 USD bundle price

Quantity:

GP3 Desktop Eye Tracker with Gazepoint Analysis (UX Edition) Bundle

GP3 Eye tracking system and Analysis UX Edition eye tracking software. Discounted when purchased as a bundle! If you need to perform any kind of analysis of gaze information we highly recommend the UX Edition software package. Analysis UX follows the design philosophy of the GP3, providing high end, research grade, performance with a simple and easy to use interface. Bundle purchase includes:

• GP3 Desktop Eye tracker
• Gazepoint Analysis UX Edition software
• 1 year warranty, software updates & technical support

Price: $1895 USD bundle price

Quantity:

GP3 Desktop Eye Tracker with Gazepoint Analysis (UX Edition) & GP3 VESA Screen Mount Bundle

GP3 Eye tracking system, Analysis UX Edition eye tracking software & GP3 VESA Screen Mount. Discounted when purchased as a bundle! In addition to the GP3 and Analysis UX, we have included the VESA Screen Mount in this bundle. The screen mount simplifies setup by clearing desk space, and locking the eye tracker to the display preventing accidental bumps requiring re-calibration. The VESA Screen Mount pivoting mechanism also allows for easily adjusting the eye tracker angle, avoiding the need to raise or lower subject participants on a chair. Bundle purchase includes:

• GP3 Desktop Eye tracker
• Gazepoint Analysis UX Edition software
• GP3 VESA Screen Mount
• 1 year warranty, software updates & technical support

Price: $1995 USD bundle price

Quantity:

GP3 Desktop Eye Tracker with Gazepoint Analysis (UX Edition) & GP3 Laptop Mount Bundle

GP3 Eye tracking system, Analysis UX Edition eye tracking software & GP3 Laptop Mount. Discounted when purchased as a bundle! A similar bundle as above, but for studies on the go this bundle includes a Laptop Mount. Bundle purchase includes:

• GP3 Desktop Eye tracker
• Gazepoint Analysis UX Edition software
• GP3 Laptop Mount
• 1 year warranty, software updates & technical support

Price: $1995 USD bundle special

Quantity:

Gazepoint Analysis Professional Edition Product Bundles

GP3 Desktop Eye Tracker with Gazepoint Analysis Professional Edition(Professional Edition) Bundle

GP3 Eye tracking system and Analysis Professional Edition eye tracking software. Discounted when purchased as a bundle! If you need to perform any kind of analysis of gaze information we highly recommend the Gazepoint Analysis Professional software package. Analysis Pro follows the design philosophy of the GP3, providing high end, research grade, performance with a simple and easy to use interface. Bundle purchase includes:

• GP3 Desktop Eye tracker
• Gazepoint Analysis Professional software
• 1 year warranty, software updates & technical support

Price: $1490 $1395 USD bundle price (save $95)

Quantity:

GP3 Desktop Eye Tracker with Gazepoint Analysis (Professional Edition) & GP3 VESA Screen Mount Bundle

GP3 Eye tracking system, Analysis Professional Edition eye tracking software & GP3 VESA Screen Mount. Discounted when purchased as a bundle! In addition to the GP3 and Analysis Pro, we have included the VESA Screen Mount in this bundle. The screen mount simplifies setup by clearing desk space, and locking the eye tracker to the display preventing accidental bumps requiring re-calibration. The VESA Screen Mount pivoting mechanism also allows for easily adjusting the eye tracker angle, avoiding the need to raise or lower subject participants on a chair. Bundle purchase includes:

• GP3 Desktop Eye tracker
• Gazepoint Analysis Professional software
• GP3 VESA Screen Mount
• 1 year warranty, software updates & technical support

Price: $1610 $1495 USD bundle price (save $115)

Quantity:

GP3 Desktop Eye Tracker with Gazepoint Analysis (Professional Edition) & GP3 Laptop Mount Bundle

GP3 Eye tracking system, Analysis Professional Edition eye tracking software & GP3 Laptop Mount. Discounted when purchased as a bundle! A similar bundle as above, but for studies on the go this bundle includes a Laptop Mount. Bundle purchase includes:

• GP3 Desktop Eye tracker
• Gazepoint Analysis Professional software
• GP3 Laptop Mount
• 1 year warranty, software updates & technical support

Price: $1610 $1495 USD bundle special (save $115)

Quantity:

Product Accessories / Add-Ons

Remote Viewer Software Add-On

Remote Viewer software allows for remote monitoring of Gazepoint data collection and experiments. Requires Gazepoint Professional or UX Edition. Price: $395 USD

Quantity:

3 Yr Extended Warranty & Support

Extend your warranty and technical support from 1 yr to 3 yrs.

• Includes software updates

Price: $195 USD

Quantity:

 

GP3 VESA Screen Mount

VESA monitor bracket to mount the GP3 directly to your computer screen.

• Mount for 22″ display sizes or smaller
• Designed to fit both 100 mm and 75 mm VESA hole spacing
• Reduce clutter on your desktop!
• Locks eye tracker to screen to minimize need for recalibration

Price: $120 USD

Quantity:

 

GP3 Laptop Mount

Compact mounting stand for use with GP3 and laptops or notebooks and mobile use.

• Mount for laptops and notebooks
• Compact design
• Stable platform
• Optimal GP3 height

Price: $120 USD

Quantity:

--
You received this message because you are subscribed to the Google Groups "psychopy-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to psychopy-user...@googlegroups.com.
To post to this group, send email to psychop...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/psychopy-users/bdda1dcd-c7a3-4319-a3aa-737e9d7ff667%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Claudine LeBosquain]

Hello Sol, 

I would like to use the GP3 tracker in combination with PyGaze (which incorporates PsychoPy), with regards to setting up a brain-machine interface that incorporates information about eye-movements. Do you think it would be possible for you to explain how you went about incorporating the GP3 tracker with PsychoPy? I believe that if I can get it working with PsychoPy, the GP3 tracker might work with PyGaze. If not, the implementation should be similar. Would you be willing to share code or assist me in trying to set up the GP3 tracker with PsychoPy, as you seem to have done with Martin? 

Thank you,

Claudine LeBosquain

[Martin Guest]

Hi Michal,

Apologies for delay in replying, but I have been on holiday the past few weeks. I haven't managed to integrate GP3 into the psychopy eyetracking libraries written by sol yet, plan on doing this fairly soon. I am able to extract data from the GP3 directly from psychopy, then dump the data into an excel spreadsheet, however the libraries would be better to use as they have extra functionality. i found however, some issues with integrating the calibration application, that comes with the GP3 eyetracker and psychopy.

Thanks Martin

[Martin Guest]

Hi Jibo,

For £300 its not a bad eyetracker, however you will not get the same accuracy as you would with more faster/expensive system, and it is only 60hz. However its fairly easy to setup, and connecting and extracting data seems fairly straight forward.

Thanks

Martin

[Claudine LeBosquain]

Hello Martin, 

This seems to be of use to me too... I was wondering if you would mind keeping me updated with the details of your implementation? I am also trying to integrate the GP3 eyetracker with psychopy but have relatively little programming experience and would really appreciate some help with the integration. 

Please let me know,

Claudine

[Ali Nazjoo]

Hi Martin, 

How successful is your GP3 data extraction from python? We tried embedding the socket code mentioned in their API documents directly into our psychopy python code and found instability when trying to insert markers at the USER_DATA column prior to and after stimuli presentation. I've posted the command we used below:

s.send(str.encode('<SET ID="USER_DATA" VALUE="1">\r\n'))

Would love to hear your thoughts on this! Also, let us know if you need any additional help for the collaboration you are working on with Sol on GP3/iohub integration.

Thanks,

Ali

On Monday, August 4, 2014 4:19:58 AM UTC-4, Martin Guest wrote:

[Martin Guest]

Hi Claudine,

No problem, I will keep you posted

:)

Martin

[Martin Guest]

Hi Ali,

Interesting, we are still piloting the gp3, but haven't had any real issues with the data field as such, and at first glance the line of code you sent seems fine. The only problem is the sampling rate, which will always have a 16ms delay, due to it being 60hz. Also will be starting off the project this week (always planned august, because of other commitments), and yes any help would be much appreciated.

Thanks

Martin

[Sean]

Hi Martin,

You mentioned there was some issue with the calibration process during integration. Can you elaborate more? Thanks!

Sean

[Martin Guest]

Hi Sean,

the calibration program, by itself is fine. However when you initiate it from within psychopy, its sends the calibration into the background, that is the windows start menu appears. This makes it tough to get back into psychopy without manually using the mouse. I asked the company about this and they said they were aware of the issue, and were working on it. They do seem to release updates quite frequently, so hopefully they will fix it.

Cheers

Martin

[Martin Guest]

FYI

version 1.81 of psychopy has support for gp3. please find a file called gccursor.7z attached. calibration is still done outside of psychopy.

first make sure you are running version 1.81

download and unzip gccursor.7z

start the gazepoint control app and calibrate

then in the gccursor folder start the run.py program

btw, if you are going to use the iohub for connecting to the gp3 then you will need at least a computer with a i5 processor, anything less and the program will eventually slow down and become unresponsive. essentially you need something that is properly multi-cored.

thanks

[Martin Guest]

Hi,

Thought I would update this thread, if you run the program from a command line with the command:

C:\Program Files (x86)\PsychoPy2\python.exe" run.py

this solves the problem of the program becoming unresponsive. Hence you won't need a high spec computer to run iohub scripts on the gp3. The problem lay with the psychopy IDE  not liking  lots of runtime prints. Running it from the command line solves this

:)

[Geoff Carre]

Hi Folks.Have put in a grant application to purchase the Gazepoint GP3 in part because of this thread. I have used Psychopy in the past to give tasks such as the Stroop and Backward Digit Span Tasks. I would like to be able to do eye-tracking and pupil dilation. Has anyone tried pupil size with this setup?

Thanks in Advance
Geoff

[Martin Guest]

Hi Geoff,

The GP3 has a field that gives the pupil diameter in pixels, and I think its in relation to the camera image, however if you contact gazepoint they can give you more details. I have found they very helpful in the past.

Thanks

Martin

[Geoff Carre]

Thanks Martin! For others looking for cheap pupillometry, here was their reply:

"The system operates at 60Hz and does record pupil diameter change data but the resolution is low so may not get biometric data - pupil response to stimulus."

and

"When I said resolution is low, I meant in terms of the image resolution of pupil size itself. Fine, minute changes may not be detected with the lower resolution.

 We are working on a hi res offering for pupilometry targeted as being released 6 months from now. It'll be priced around $5k with a hi-res camera that will give a really good clear image of the pupil."

I am going to try the Eyetribe, which is very inexpensive. Here is a paper that did lab comparisons between it and the EyeLink 1000 and the eyetribe compared well for pupillometry.

https://peerj.com/preprints/585.pdf

cheers/geoff

[Sean T Ma]

Hi Geoff,

Thanks for the useful information on GP3's limitation for pupilometry.

Seems like I should start reviewing Sol's Eyetribe's control module in Psychopy.

Happy holidays!


Sean

 

Best, 

Sean T. Ma, Ph.D.
Postdoc Fellow in Human Brain Imaging

Dept of Psychiatry / Marketing 
University of Michigan Health System

VA Ann Arbor Healthcare System

tehs...@gmail.com
+1-734-363-8848 (cell)

LinkedIn.com/in/seantma

@_seantma (twitter)

--
You received this message because you are subscribed to a topic in the Google Groups "psychopy-users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/psychopy-users/SxcYlj9KwsQ/unsubscribe.
To unsubscribe from this group and all its topics, send an email to psychopy-user...@googlegroups.com.

To post to this group, send email to psychop...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/psychopy-users/715955b8-4dfb-4858-bc4d-d8cd0f9506e3%40googlegroups.com.

[Sol Simpson]

FYI: The eyetribe interface, contributed by Zaheer Ahmed with coding input from myself,  is not complete right now. It provides the eye sample stream from theeyetribe system, but calibration must be done externally using the software provided by theeyetribe. As well, the sample timestamp conversion from device_time to iohub time should be considered incomplete / inaccurate and you may want to rely on using the device_time field itself right now. I have no idea if / when time will become available to polish the interface off to the level of some of the other supported eye tracker makers, like lc technologies,  smi, sr research,  tobii (listed in alphabetical order).

Thanks and happy holidays.