preview

Essay On Security Detection

Decent Essays

There are some previous researches that are used to be referred to develop a real-time system on drowsiness detection. These are researches and journals that are related to this project either directly or indirectly:- Alioua et al. [17] proposed a robust and novel algorithm which does not require any training data or any special cameras for eye state analysis by iris detection using Circular Hough Transform (CHT) technique. The experimental results were represented as statistical measures such as Confusion Matrix, Correct Classification Rate and kappa Statistic. The experiment was performed on 7 real video sequences and it provided 99% correct classification rate and 88% kappa statistic value. For future work, they focused on inserting …show more content…

In PERCLOS estimation, for each frame, linear support vector machine (LSVM) method was used to classify edge detected eye images. In gaze estimation, exponentially smoothed vertical and horizontal movements of the pupils were taken into consideration. This methodology achieved an overall accuracy of 70.99% for offline tests of eye state detection while the real time eye state detection provided a precision of 78.57% and recall of 18.33% and 83.64% overall accuracy. The gaze estimation methodology achieves 80.5% accuracy. For future work, they focused on performing parallel optimizations in order to achieve high performance, fine tune the system’s fatigue detection function based on machine learning. Dahiphale and Sathyanarayana [14] designed and tested driver vigilance level monitoring system on raspberry pi. The system contained two main parts including driver’s face and eye detection part using Viola Jones face detection with AdaBoost (Adaptive- Boosting) method and Circular Hough Transform technique and driver’s face tracking part using CAMSHIFT (Continuously Adaptive Mean Shift). They focused on vehicle states, weather conditions, signal changes for person in different conditions such as illness in future works. Hichri et al. [15] proposed a robust system for drowsiness detection of drivers based on eye state in real-time. The system contained two main parts

Get Access