Gesture Recognition on Low Resolution American Sign Language (ASL) Images
Altaful Amin
aa547@duke.edu
| Paper PDF |
|
American Sign Language is widely used among the Deaf communities in the US and Canada. A machine learning system that can take an image of the sign being made and interpret it for other non-users would be very helpful to bridge the gap between ASL users and non-users without an interpreter. In this project, I developed a CNN model that will train on images of the ASL letters of the alphabet. I also simulated lower resolution images to see how well the model performs on different sizes of image, keeping in mind an end goal of developing a model that can be used by a wide range of image capturing devices. Results showed that the model performed at 89% accuracy and higher on the image datasets.
|
| Paper: |
Data:
|