A Machine Learning Based System To Detect Driver Drowsiness

Now a days, many of the accident are happening due to the drunk-driving or due to the sleep while driving and feeling drowsy, what happens if the driver won't be allowed to drive the vehicle if he/she isn't conscious and not able to drive, many of the accidents can be prevented by it. This idea can be implemented by detecting the conscious state of a person. Here an integrated approach is presented with the help of MQ-3 sensor, Pi-Camera, and machine learning algorithm to detect driver drowsiness. The face and mouth landmarks are used to detect the driver drowsiness condition using Long-Short Term Memory (LSTM) machine learning approach. Alternatively with the help of MQ-3 sensor, the alcohol level of a driver can be calculated and if the person is found to be drunk within threshold limit and is conscious, car ignition will be permissible else ignition won't start and an alarm will be generated.