Driver Inattention Detection System: A PSO-Based Multiview Classification Approach

This paper presents driver simulation results for multi-sensory platform aimed at providing data that are used for driver states classification. This work explores PCA and S-PCA, for dimensionality reduction, and Random Forest, for classification. Finally, PSO-based multi-view classification is used for final fusion of the individual classifiers. The results suggest S-PCA, Random Forest and PSO-based multi-view classification as the best combination reaching an accuracy of 91.46% for the limited available input data.