ActiveDrive Driver Drowsiness Detection System

Project Title

ActiveDrive Driver Drowsiness Detection System

Project Area of Specialization Artificial IntelligenceProject Summary

ActiveDrive is a driver drowsiness & fatigue detection system that has to be installed in the vehicle dashboard. It uses Image Processing & Deep Learning algorithms (that run on an Nvidia Jetson Nano development board) to accurately detect driver's level of fatigue by capturing his facial features (through an IR camera) and then alert the driver and recommends him to stop his car and rest for a while before driving again.

Project Objectives

In 2017, according to NHTSA reported that 91000 crashes due to drowsy drivers that 50,000 injured and 800 dead. ActiveDrive aims to stop or atleast lower the high amount of fatalities that are caused due to road accidents due to sleepy or tired drivers in a not only cost efficient but also the most reliable and accurate way possible. One of the key objectives to detect signs of drowsiness and fatigue before the driver falls asleep behind the wheel, this is done through Machine Learning algorithms running on a small portable Jetson computer installted in the vehicle and also a camera installed on the driver dashboard that will observe the driver and warn if any signs of drowsiness are detected through a small alarm and warning sign without distracting or frightening the already drowsy driver.

Project Implementation Method

To accurately detect early signs of tiredness, sleepiness or fatigue, a camera (IR camera) installed on the dashboard directly infront of the driver will capture the driver's face and feed the video into an Nvidia JETSON NANO development board where Image Processing algorithms (running in Python) will extract the driver's facial features and landmarks (Eyes & Mouth location and size ratios) and will feed the data into a pre-trained Neural Network model that will accurately predict whether the person is feeling drowsy or not (desired accuracy is 90%) and will alert the driver by a gentle  alarm tone. This will be all done in realtime. Also the Jetson Board and camera are trained to detect whether some other intruder has entered into the car or not and will notify the owner of the vehicle through a phone application.

Benefits of the Project

A study by AAA Foundation For Traffic Safety estimated that 3,28,000 traffic accidents due to drowsy drivers occur internationally, that's more than 3 times the police reported number, the same study found that 109,000 injuries occured due to drowsy driver crashes and also resulted in 6400 fatalities. We hope to reduce these large amounts of unnecessary deaths by implementing a device that can accurately detect the driver's level of fatigue and issue a warning. Every vehicle owner can benefit from this project especially the ones who have to go on long-night drives. Moreover the device will also provide a security feature by detecting any intruder or unknown driver and notify the driver via Android Application.

Technical Details of Final Deliverable

Our final project will be an Nvidia JETSON NANO board enclosed in a casing that will be fitted in the car's dashboard and will be connected to an IR camera that itself will be fitted behind the steering wheel infront of the driver to capture his facial features and then these features (such as Eye and Mouth location, size and aspect ratios) will be processed from OPENCV Image Processing algorithms and then all this data will be fed into a trained Neural Network model which will then predict the levels of drowsiness accurately.

Final Deliverable of the Project HW/SW integrated systemCore Industry ITOther IndustriesCore Technology Artificial Intelligence(AI)Other Technologies Internet of Things (IoT)Sustainable Development Goals Good Health and Well-Being for PeopleRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	29000
Nvidia Jetson Nano	Equipment	1	25000	25000
Dashboard camera	Equipment	1	4000	4000