Skip to content

Latest commit

 

History

History
82 lines (39 loc) · 2.13 KB

README.md

File metadata and controls

82 lines (39 loc) · 2.13 KB

Driver-monitering-system

Driver behaviour detecting app that warns if the behaviour is detrimental.

How it works:

  • The app starts of with the camera running and it looks for the eyes in the video.
  • It then looks for any distracting behaviour shown by the driver.
  • If the behaviour is detected to be bad it warns the driver.
  • The alert runs till the driver touches the screen and if it takes too long an sos is sent to his emergency contact number

Note:

The repo is divides into 3 files:

  • The alert folder contains the app that triggers the alarm by flashing the screen red and also makes the flashlight of the camera to blink.
  • The tfmobile folder consists of the android app that detects around 9 classes of distractions such as:
    • safe driving
    • texting - right
    • talking on the phone - right
    • texting - left
    • talking on the phone - left
    • operating the radio
    • drinking
    • reaching behind
    • hair and makeup
    • talking to passenger
    • Here is the accuracy and loss of the cnn model used to build this application:
  • The vigilate folder contains the app that detects if the driver is sleeping or not.
  • example: Image 1 image 2

Installation:

  • First install openCv manager for android by using this link.
  • Next open the folder of the app you wish to use via android studio and you are good to go.

Important note:

Th three apps are now seperate apps, if you would like to contribute plases help by combining the three into one.

Credits:

Bharath Nair,
Steve Paul

References:

tensorflow for poets
OpenCV for android
The ML model for the tfmobile app