Although 98% of the binocular and monocular data gets used during calibration, some times 30% of the pupil data gets rejected due to being below the 80% confidence threshold. This might be resolved by increasing the resolution or other there are other issues that need to be at least investigated and benchmarked for gaze accuracy and robustness.
validation_log.txt
Although 98% of the binocular and monocular data gets used during calibration, some times 30% of the pupil data gets rejected due to being below the
80% confidence threshold. This might be resolved by increasing the resolution or other there are other issues that need to be at least investigated and benchmarked for gaze accuracy and robustness. validation_log.txt