Computer Sciences and Information Technology
Title : | Development of a prototype of disabled-friendly automatic virtual text-entry keyboard interface system under practical environmental conditions |
Area of research : | Computer Sciences and Information Technology |
Focus area : | Development of virtual keyboard |
Principal Investigator : | Dr Rabul Hussain Laskar, Associate Professor, National Institute of Technology (NIT) Silchar |
Timeline Start Year : | 2019 |
Contact info : | rabul2u4u@gmail.com; rabul18@yahoo.com |
Details
Executive Summary : | The proposed system is a vision-based 2D hand gesture keyboard system to replace traditional text entry interface systems. The virtual keyboard will comprise printable characters of the traditional keyboard that can be used by disabled and elderly people to control home appliances such as T.V., A.C., password-based systems, virtual data entry interface etc., without any physical contact. The most recent hand gesture system with printable keyboard characters are limited to 58 characters (includes alphabets, numbers, arithmetic operators, and some ASCII characters) using color markers [1], and limited to 40 characters for bare-hand systems [2]. Literature survey showed that most of the existing gesture systems are developed for static gestures or short-duration gestures, under non-practical (simple background) conditions. Proposed gesture system will be developed for both color marker as well as bare-hand under practical environment conditions. As observed, though the naturalness of bare hand system cannot be replaced by color-markers, however, it is observed that due to easy detection and tracking of the color-markers, the calculated gesture trajectory points are less noisy leading to higher accuracies for the same set of features. Therefore, the actual significance of the features are more prominent in color-marker systems. Based on the reference accuracy of color marker system, bare-hand system can be optimized. In the existing 58 character system, it is observed that with the increase of gesture characters in the system, the misclassification within the classes increases which affects the overall accuracy. Therefore, more unique geometrical features of the characters need to be explored for maximum accuracy. Difficulty in detection of bare-hand under complex and dynamic background is another major issue which is not addressed in any of the previous studies. This project will address different issues in all the vital phases to develop a robust virtual keyboard |
Co-PI: | Dr Taimoor Khan, Assistant Professor, National Institute of Technology (NIT) Silchar |
Total Budget (INR): | 89,45,420 |
Publications : | 6 |
Organizations involved