Distraction Detection to Predict Vehicle Crashes: A Deep Learning Approach

Reda Bekka, Samia Kherbouche, Houda El Bouhissi


The Road safety is a major issue, both in terms of the number of road casualties andthe economic cost of these accidents at the global, regional and national levels.Combating road insecurity is a priority concern for every country, as travel con-tinues to increase and, despite the measures taken in many countries to improveroad safety, much remains to be done to reduce the number of deaths. In thispaper, we review applied approaches related to the detection of distractions indriver assistance, building on existing approaches with key improvements to in-crease the rate of detection of various drivers public transport distractions, andpresent a new approach to preventing road crashes in the context of intelligenttransport. Preliminary experiences indicate that the proposed approach providehigh eciency and accuracy.


CNN, distraction, detection, deep learning, drowsiness, intelligent transport, opencv, transfer learning, VGG16

Full Text: PDF