Real Time Human Computer Interaction Using Facial Gestures
Rohit Lal, Shital S. Chiddarwar · 2019
The main focus of this paper is to advance the study of controlling a personal computer using various face gestures and movements captured using RGB webcam in real time. This is highly useful for physically challenged people. Recent state-of-the-art face landmark detection algorithm is used to localize facial features which helped in estimating various parameters for computer control with high precision. Position of nose tip is tracked for cursor control and smile detection is carried out for scrolling of pages. A unique mathematical expression is introduced which is used to detect the state of the eye (blinking or not). The algorithm developed was found to be robust against different head orientation and lighting condition. This algorithm also outperforms many other related works both in terms of accuracy and computation time. The relevant pipeline architecture and results are demonstrated in this paper.