Cite this model
Rapa 2024 online PPG drowsiness TCN (steering-wheel PPG, ULP SoC). (2024).
@misc{rapappgtcndrowsi2024rapa,
title = {{Rapa 2024 online PPG drowsiness TCN (steering-wheel PPG, ULP SoC)}},
year = {2024}
}Activity counts measure attention, not model quality. Missing counts are unknown. Publication status follows the linked source; a listed venue alone does not confirm peer review.
Rapa et al. detect driver drowsiness with a temporal convolutional network on PPG recorded from steering-wheel sensors. On 16 subjects with leave-one-subject-out validation, mean accuracy was 77.03%, with a false-positive rate of 8.21% and a false-negative rate of 13.92%. A quantized model on a commercial ultra-low-power SoC ran in 4.8 ms and 117 µJ per inference (IEEE AICAS 2024).
Temporal convolutional network, quantised for an ultra-low-power SoC

Rapa 2024 online PPG drowsiness TCN (steering-wheel PPG, ULP SoC). (2024).
@misc{rapappgtcndrowsi2024rapa,
title = {{Rapa 2024 online PPG drowsiness TCN (steering-wheel PPG, ULP SoC)}},
year = {2024}
}Added to the ChatPPG research hub