Skip to content

In this repository we keep the code for the implementation of the eye-tracking experiment for the COST action MultiplEYE. The eye-tracking-while-reading experiment is implemented using Python.

Notifications You must be signed in to change notification settings

MultiplEYE-COST/wg1-experiment-implementation

Repository files navigation

MultiplEYE WG1: Experiment Implementation

This repository contains the code for a MultiplEYE eye-tracking-while-reading experiment for multiple languages. After you have read this README make sure to read the relevant files in the guidelines folder. There exists a MARKDOWN and a HTML version of the guidelines. Both are exactly the same. You can read whatever format you prefer. Also for this README, there exist two versions.

If you would like to contribute, please read the following guidelines: CONTRIBUTING.md and contact multipleye@cl.uzh.ch.

Contents

Installation

In order to run the MultiplEYE experiment you will need to complete the following steps

  1. Please read the official MultiplEYE guidelines linked on this page: MultiplEYE contribute
  2. Download the experiment code from this repository. Either clone via git or download the zip file (see download as zip).
  3. Following the guidelines, prepare the stimulus files which includes the creation of the images
  4. Copy the stimulus files to the correct location in the experiment data folder: experiment_implementation/data/[your stimulus folder]
  5. Prepare the environment for the experiment following the guidelines in CONDA_ENVIRONMENT.md
  6. Install the necessary packages for your eye-tracker. For EyeLink eye-trackers INSTALL_PYLINK.md. For Tobii eye-trackers, please see develop for Tobii further below.
  7. Run the dummy experiment to check if everything is working correctly
  8. Run the experiment with an eye-tracker

Download the experiment code as zip file

If you do not have git installed, you can download the code as a zip file from this website. Click on the green button Code and then click on Download ZIP. Unzip the folder in our preferred location.

Note: The repository contains a folder with toy stimuli to test the experiment. Sometimes you cannot unzip these files because the path gets too long. In this case, you can either move the folder to a different location closer to the root or unzip and ignore the files in the toy stimuli folder (i.e. skip the error message that pops up during unzipping). To use the toy stimuli, please contact multipleye@cl.uzh.ch and we can provide them and you can copy them to the correct location.

Download

Run the dummy experiment

The experiment can be run in dummy mode which means that can be run without an actual eye-tracker. If you'd like to run it, make sure you have completed the above steps, and you have your conda environment activated.

To be sure that you are running the dummy version, activate the environment:

conda activate multipleye3.9

Then you can navigate to the root folder of your local clone of the repository (your path should now end with wg1-experiment-implementation. Run the following command to run the dummy experiment:

python experiment_implementation/start_multipleye_session.py

In the GUI that will pop up you can tick the box Dummy version to run the experiment in dummy mode.

The result files

The experiment will write log and data files to a newly created results folder for your language and country in the data folder (data/eye_tracking_data...). In there it will create a folder depending on the experiment type. For example, if you run the test session, it will create a folder called test_dataset. Within those folders it will simply create a new folder for each participant. The folder name is the participant ID (three-digit number). If you run the script for the core dataset, it will prevent you from running the experiment twice for the same participant. Note that if you run a test session, it will not warn you if you enter the same participant ID more than once. It will just write the files to the same folder.

The naming scheme of the log files is a follows: [log_file_type]_[session_id]_[participant_id]_[date]_[timestamp].txt.

All logfiles are csv files. Note that the timestamps are relative to the start of the experiment. The experiment starts at timestamp 0.

Run the experiment with an eye-tracker

In order to run the experiment with an actual eye-tracker you can tick the respective box in the GUI in the lab settings section. The box Dummy version should be un-ticked.

Depending on what eye-tracker you intend to use you need to install the software that comes with the eye-tracker. For EyeLink and tobii you can download the software online for free.

Develop and run experiments for EyeLink

The code has been tested with EyeLink eye-trackers and mostly on a Windows presentation PC. However, MacOS should work as well, but it has not been tested as thoroughly. You will need to install pylink a package provided by SR Research if you use EyeLink eye-trackers. Note that using pip to install pylink installs a different package although the names are the same! You have to follow these step-by-step instructions of how to install pylink can be found here: INSTALL_PYLINK.md

Develop and run experiments for Tobii

An early version of the experiment has been tested with tobii eye-trackers. However, the code is not yet fully developed and tested as the experiment as been developed further since then. If you want to use a tobii eye-tracker, you definitively need to follow the following steps. Once those are completed, it will still be necessary to adjust the code in order to work with the tobii eye-trackers:

  1. Install Tobii Pro SDK: link
  2. Download the SDK
  3. Unzip the folder and copy all files from either the '32' or '64' folder to the experiment_implementation folder
  4. Then you can pip install the package: pip install tobii_research

Develop and run experiments for other eye-trackers

Depending on what is needed we can add more eye-trackers. There is also a team that is trying to set up experiment using a webcam. Please contact multipleye@cl.uzh.ch for more information.

About

In this repository we keep the code for the implementation of the eye-tracking experiment for the COST action MultiplEYE. The eye-tracking-while-reading experiment is implemented using Python.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 4

  •  
  •  
  •  
  •