In the last few years, more accidents are happened mainly due to the drowsiness of the driver. Various accident prevention technologies are developed but still accidents are happening. This is due to that, the technologies which are available in present are all detecting the drowsiness of the driver at the time accidents. So, there is a possibility for happening of accidents. If the drowsiness of the driver, can be predicted before driving, it will be very useful to prevent the accidents. In this method, there is a solution to detect the drowsiness of the driver, before the driver starts to drive. There is system in this method from which the driver got approval for driving. The system has two level verification process for the drivers to detect the drowsiness. The First level of verification process is captcha process using python libraries or audio listening process using gTTS library. The second level verification process is based on detecting the facial expression of the drivers using the haar cascade classifier with OpenCV library in python. This process has the accuracy level of 95% in pre-driving drowsiness detection. The above levels are the two levels which are used detect the drowsiness of the drivers before start to drive.