search

simonarvin / eyeloop

EyeLoop is a Python 3-based eye-tracker tailored specifically to dynamic, closed-loop experiments on consumer-grade hardware.
GNU General Public License v3.0
479 stars 68 forks source link

Hardware recommendations #35

Closed iandol closed 3 years ago

iandol commented 3 years ago

Hi, do you have recommended camera hardware, or a list of options. Your readme says >1000Hz on non-specialised hardware, but most standard camera do not approach this frame rate, so I assume you mean the processing rate if a camera could feed in images fast enough. openEyeTrack use a Genie Nano M640 NIR and IR illuminator that can go up to 800fps, I assume this could be used?

How does your system handle calibration (for human and/or non-human primate subjects)?

simonarvin commented 3 years ago

Hi @iandol, Your assumption is correct: The 1000 Hz is assuming that frames are fed to EyeLoop fast enough, e.g., from an offline recording or fast camera feed. Unfortunately, I have no recommendations for hardware. The Genie camera should be compatible with EyeLoop, if you can retrieve the camera frames through Python, e.g., using OpenCV.

EyeLoop does not provide calibration. It could be implemented through an Extractor module

Best, Simon

iandol commented 3 years ago

Thanks Simon. Do you plan to add calibration in the future, as this is fairly core feature for eye tracking?

simonarvin commented 3 years ago

This is not yet in our plans for the near future, however I can look into implementing it. The calibration step converts the eye tracking data to monitor pixel coordinates, correct? Such to reveal where the eye is looking