Using Active Appearance Model to detect driver fatigue has been proposed in past research. This paper proposes a rapid method of driver fatigue detection by applying Active Appearance Model with simplified landmarks only around eyes, and also proposes to convert the eye contour aspect ratio into PERCLOS to judge the degree of the driver fatigue. Compared to the original method, this method reduces redundant facial texture information, and accelerates the speed of eye state detection while improving detection accuracy rate of driver fatigue. The experiments results show that this method is fast and effective that can be used in real-time detection.