pupil-t265

pupil-t265 provides plugins for the Pupil Core mobile eye tracking software for calibrating, recording and exporting tracking data from the Intel RealSense T265 tracking camera. This way, the T265 can be used for simultaneous head tracking when rigidly mounted with respect to the Pupil Core world camera.

Citation

If you use pupil-t265 in your research, please cite our paper Positional head-eye tracking outside the lab: an open-source solution:

@inproceedings{10.1145/3379156.3391365,
    author = {Hausamann, Peter and Sinnott, Christian and MacNeilage, Paul R.},
    title = {Positional Head-Eye Tracking Outside the Lab: An Open-Source Solution},
    year = {2020},
    isbn = {9781450371346},
    publisher = {Association for Computing Machinery},
    address = {New York, NY, USA},
    url = {https://doi.org/10.1145/3379156.3391365},
    doi = {10.1145/3379156.3391365},
    booktitle = {Symposium on Eye Tracking Research and Applications},
    articleno = {14},
    numpages = {5},
    keywords = {gaze estimation, locomotion, open source, Eye tracking, head tracking, 
                mobile, simultaneous localization and mapping},
    location = {Stuttgart, Germany},
    series = {ETRA ’20}
}

Installation

The plugins require a recent version of the Pupil Core software (recommended 1.21). Copy t265_capture.py to pupil_capture_settings/plugins and t265_player.py to pupil_player_settings/plugins.

Usage

Capture

Use the T265 Calibration plugin to calculate the camera extrinsics, i.e. the position and orientation of the left T265 camera with respect to the world camera.

The T265 Recorder will display live position, orientation and velocities of the T265 device. During a recording, these will also be saved to disk.

Player

Use the T265 Exporter plugin to export the recorded data to csv. The export contains two files: odometry.csv and t265_left_extrinsics.csv.

Example analysis

The examples folder contains two recorded and exported datasets along with analysis code to reproduce the results of the paper.

To run the example code, clone or download the repository code first. The recommended way of installing the dependencies is through Anaconda/Miniconda by creating a dedicated conda environment:

conda env create -f examples/example_env.yml

In your base environment, install nb_conda:

conda install nb_conda

Afterwards, start a Jupyter Notebook server from your base environment, open the etra2020.ipynb Jupyter Notebook, go to Kernel > Change kernel and select Python [conda env:pupil-t265].

Alternatively, you can install the dependencies via pip:

pip install xarray scipy numpy-quaternion numba fast-histogram matplotlib colorcet

T265 mount

The repository contains .stl files for 3D printing a mount for the T265 that can be attached to the Pupil Core world camera in the mount folder. t265_housing.stl is a housing for the tracker and t265_clip.stl is a clip that is printed twice and glued to the housing in order to attach it to the world camera (see mount_example.jpg).