Fatigue and drowsiness of drivers are among the primary reasons of road crashes, as these symptoms often lead to micro sleep episodes where drivers briefly fall asleep without realizing it. Such lapses in attention can result in severe accidents, causing injuries or fatalities. Researchers from the automobile industry are actively exploring various solutions to tackle this problem by leveraging advancements in technology. This paper aims to detect driver fatigue to lower the number of road crashes occurred by drowsiness, thereby improving overall road safety. To achieve this, driver features such as eye blinking and yawning are monitored to assess the driver's state of alertness. This research proposes a methodology of drowsiness detection according to changes in facial geometric features, specifically focusing on eye blinks and yawning. Facial landmarks are detected from the input stream of a camera and are subsequently given as input to convolutional neural network (CNN) to predict drowsiness. By utilizing these advanced techniques, the system can provide real-time alerts, enhancing driver safety and preventing potential accidents caused by fatigue.