Microsleep refers to very brief, unintentional episodes of sleep that can occur when a person is awake and trying to stay alert. During microsleep, the individual experiences a sudden and temporary loss of awareness and consciousness. The main goal is to categorise characteristics of human fill to microsleep and to develop a Microsleep Detector System a warning system to notify drivers when microsleep occurs. This project used Eye Aspect Ratio (EAR), Percentage of Eye Closure (PERCLOS), Mouth Aspect Ratio (YAR), and Tilt Head/Nodding Detection, which functioned with Python and OpenCV connected to a webcam to detect the microsleep of the driver. The detection uses the eye (EAR and PERCLOS), mouth (YAR), and head (Tilt Head Angle). The alarm of sound warning will be triggered if the value of EAR, MAR, and PERCLOS exceeds or is less than the required threshold. The experiment was conducted for 30 min using a driving simulator and a driving environment that had been created using Unity 3D. There were 20 participants involved, consisting of males and females with a specific range of age. The data were analysed using the Modified Karolinska Sleepiness Scale and IBM SPSS ANOVA. The expected result is to get the driving behaviour parameter according to the Modified Karolinska Sleepiness Scale that leads to microsleep episodes such as having a conversation while driving, driving with one hand, yawning one time while driving, rubbing eyes and body incline while driving. The findings suggest that by utilizing this device, accidents related to microsleep can be reduced, contributing to improved road safety.