Loading...

Probabilistic Modelling of Fatigue Detection with Facial Features

Gholamipour Fard, Rahil | 2016

614 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 48745 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): karbalaie Aghajan, Hamid
  7. Abstract:
  8. Today, everyone is looking for ways to achieve comfort and safety in the workplace and, in so doing, appeals to various sciences. One of these sciences is "ergonomics", which examines the human relationship with the work. One of the areas that could be of great interest is the driving ergonomics. After all nowadays, many people use personal vehicles. The increasing use of personal vehicles has increased the number of accidents and deaths. In recent decades, many researches have been done in the field of driver fatigue detection. To avoid road accidents, researchers have focused on monitoring driver and vehicle behavior and tried to analyse status of the driver. Using computer vision, we can extract and process facial features to detect fatigue and other factors that may cause harm. Monitoring facial features of driver, such as blinking rate and yawning, can be effective in detecting fatigue and drowsiness. Other application of fatigue detection is monitoring computer users. Here, calculating user’s average work, break periods, and distance to computer monitor, we can detect fatigue and recommend users to take a rest. In this thesis, using computer vision, we extracted features from driver’s eyes. Then, using facial parameters, we modeled person behavior in order to warn user in abnormal conditions .To increase confidence factor in making decision, in addition to momentary data, we used context data. The achieved model can provide the statistical comparison between different people
  9. Keywords:
  10. Hidden Markov Model ; Computer Vision ; Fatigue ; Abnormal Event Detection ; Constitutive Modeling ; Drowsiness ; Facial Features

 Digital Object List

 Bookmark

No TOC