Abstract: Drowsiness is considered as a significant risk factor that contributes to large number of accidents. This study, focuses on methodologies developed for counteracting its effects with very accurate classification techniques categorizing the different drowsy states and alerting the person at definite instants. An optimal Bootstrap technique is applied to features extracted by Daubechies Wavelet Transform (DWT) and the drowsy states are classified using Neural Network (NN) classifier. The Receiver Operating Characteristics (ROC) curve shows the classification accuracy and the computation time is also calculated. In order to improve the efficiency of the proposed method, Fractional Fourier Transform (FrFT) based feature extraction is implemented with ABC (Artificial Bee Colony) for optimization and classification done using NN and Sparse classifiers. The three methods exhibit high efficiency in improving the systems performance in terms of accuracy F1 score and computation time. A comparative study of the three methods is also done with the latter showing better results.
Reena Daphne and A. Albert Raj, 2016. An Innovative Optimization Technique for Drowsiness Detection Based on Feature Extraction Capitalizing Neural Network and Sparse Classifiers. Asian Journal of Information Technology, 15: 2399-2410.