Deformable Face Fitting Based Drowsiness Detection in Real Time System

Drowsiness is the state where a person is not able to perform any task at his/her optimum efficiency. Due to negative impacts of drowsiness on daily activities, drowsiness detection is important to prevent consequences. Non-intrusive computer vision techniques are the most suitable method to detect drowsiness. In this method, one camera is required to analyze facial features. Among various facial features, features around mouth and eye region are most reliable parameters. In this research, eye closure and yawning data are used for drowsiness detection. Once face is detected by cascade classifier with Haar features, then eyes and mouth are detected and tracked by deformable face fitting. From eye and mouth area, their state can be estimated which is used to detect drowsiness. This system is practically feasible because it is non-intrusive; also it is well suited in real time due to its adequate speed and accuracy.