Towards a Real-time System based on Regression Model to Evaluate Driver's Attention

Autor(es)

Lago, Thiago K. ; Gonzalez, Ernesto Rodriguez ; Campista, Miguel Elias M. ; IEEE

• Tipo de documento: Artigo Científico
• Fonte: 2021 IEEE INTERNATIONAL SMART CITIES CONFERENCE (ISC2); v. N/A, p. 7-pg., 2021-01-01.

Resumo

The use of mobile devices while driving is one of the major causes of accidents. Thus, this paper aims to create a real-time system to alert drivers of inattention moments, and thus reduce the number of traffic accidents. For this, we use computer vision algorithms to determine the driver's gaze direction which, together with vehicle data such as speed and acceleration, infer whether the driver is distracted. The key idea is to calibrate the drive's head rotation thresholds as a function of vehicle speed and acceleration. If the vehicle is at low speed or low acceleration, the head rotation can be more pronounced. In addition to head rotation, we also use blind eye detection as a criterion for determining distraction. The system uses a temporal sliding window procedure to prevent oscillations between inattention and attention states. The implementation is based on hardware and software architectures composed of cost-effective and open source libraries. Experimental results in controlled and real environments show accurate results and quick detection time. (AU)

• Processo FAPESP: 15/24494-8 - Comunicação e processamento de big data em nuvens e névoas computacionais
• Beneficiário: Nelson Luis Saldanha da Fonseca
• Modalidade de apoio: Auxílio à Pesquisa - Temático