Table 3 Key challenges in drowsiness detection datasets.

From: Real-time driver drowsiness detection using transformer architectures: a novel deep learning approach

Dataset

Key challenges

MRL

Poor lighting conditions in night mode

Presence of eyeglasses

Close-up eye images only (no full face)

Grayscale images only

NTHU-DDD

Varying facial angles and head pose

Different lighting conditions (day/night)

Slow blink with nodding

Yawning

Sleepy combinations

Eye occlusion due to glasses

CEW

Varying head orientation

Wide diversity in facial expressions

Presence of eyeglasses and sunglasses

Occlusions and real-world variability (e.g., hair, hand)

Back to article page