The problem Safe and effective commute solves

There are thousands of accidents on road every year due to drowsiness of the driver. Each year, 52% of the road accidents occur only becuase of sleepy driver, causing thousands of deaths. Mitigating those deaths is a big problem the world is facing. We here have found a solution which is both accurate and feasible. It does not need any special hardware -- all it needs is a mobile phone with a camera and internet connection. Our deep learning model detects when the driver is feeling drowsy or alert with almost 100% accuracy. It can be deployed on a web server so that anyone can access it. It also provides a live stream of the driver to a person designated by the driver and also mails and calls him with live location in case of an emergency.

While commuting on a cab-hailing service, many times we face a communication barrier with the driver due to different languages that we speak. We have added a feature in the project so that both the commuter and driver can easily communicate with each other through sound waves, even if they speak different languages, translating the messages back and forth.

Challenges we ran into

There are many models for detecting driver drowsiness but none of them give accurate or 100% accuracy as they detect all the eyes in the car even the passenger's eyes also. It was a challenge to get only the driver's eyes to be modeled. We solved it by getting the largest eyes in the camera.