A visual approach for driver inattention detection

摘要: Monitoring driver fatigue, inattention, and lack of sleep is very important in preventing motor vehicles accidents. A visual system for automatic vigilance has to address two fundamental problems. First all, it analyze the sequence images detect if his eyes open or closed, then evaluate temporal occurrence estimate driver's attention level. In this paper we propose a approach that solves both neural classifier applied recognize image, selecting candidate regions might contain by using iris geometrical information symmetry. The novelty work algorithm works on complex without constraints background, skin color segmentation so on. Several experiments were carried out subjects with different eye colors, some them wearing glasses, light conditions. Tests show robustness respect situations such as partially occluded, head rotation particular, when where people have closed proposed correctly reveals absence eyes. Next, analysis image sequences probabilistic model anomalous behaviors inattention sleepiness. Image acquired laboratory while driving car used test behavior demonstrate effectiveness whole approach.