Skip to content

jhan15/driver_monitoring

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

17 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

driver_monitoring_system

This is a project aimed to monitor a driver's status and actions, such as yawn, phonecall, etc.

Architecture

The driver monitoring system consists of two parts.

  • Facial tracking: an API based on Mediapipe to track facial status, which predicts the eye status (open, close), if open, the gazing direction (left, right, center), and yawn.
  • Action detection: a deep learning model (MobileNet) to predcit driver's behavior (phonecall, texting). Yolov5 is further used to detect phones to enhance performance.

Requirements

python=3.8
tensorflow=2.8.0
torch=1.11.0
opencv-python=4.5.5
mediapipe=0.8.9.1
matplotlib=3.5.1
numpy=1.22.3
scikit-learn=1.0.2

Usage

$ git clone https://github.com/jhan15/driver_monitoring.git
$ cd driver_monitoring

# driver monitorting system
$ python3 dms.py --checkpoint models/model_split.h5 --video <path_to_video> 
                                                    --webcam <cam_id> # or

# play with only facial tracking
$ python3 facial.py

Dataset

The dataset used to train action detection model is DMD.

Demo

Releases

No releases published

Packages

No packages published

Languages