Universitas Dian Nuswantoro (UDINUS), through the participation of Alzami, Farrikh; Naufal, Muhammad; Al Azies, Harun; Winarno, Sri; and Soeleman, Moch Arief, conducted a study on eye-region drowsiness detection in low light conditions. The activity addressed driver drowsiness as a critical factor in road safety and examined an approach integrating Auto-CLAHE with Time Distributed MobileNetV2 to enhance drowsiness detection accuracy.
The study used the ULg Multimodality Drowsiness Database (DROZY) for facial expression analysis, focusing on the eye region. Its methodology segmented videos into 10-second intervals, extracted 20 images per segment, and applied the Haar Cascade method for eye-region detection. Auto-CLAHE was developed to dynamically adjust contrast enhancement parameters according to image characteristics and was integrated with Time Distributed MobileNetV2 for drowsiness classification.
The analysis yielded a classification accuracy of 93.62%. This result exceeded the reported results for Greyscale, AHE, and CLAHE, which were 92.55%, 92.91%, and 91.13%, respectively. The study also obtained a precision of 93.71% in detecting drowsiness, a recall of 93.62%, and an F1 score of 93.59%. Statistical analysis using ANOVA and Tukey HSD tests confirmed the significance of the study results.
The key innovation was the implementation of Auto-CLAHE to improve image contrast adaptation. The approach surpassed AHE and basic CLAHE in drowsiness detection performance and demonstrated robustness across diverse lighting conditions and facial expressions. This UDINUS research activity provides evidence of technology-based driver drowsiness analysis relevant to road safety and the stated sustainable development goals.