Hand Keypoints as Feature in Hand Gesture Detection Using Concatenation Keypoint Transpose

Sign languages have been ubiquitous in the past decade. The sign languages have been very instrumental for human-to-human communication for many years particularly to facilitate communication with hearing impaired peoples. Apart from being utilized by the hearing impaired, sign language is gaining popularity due to increased public awareness of the community. Nevertheless, sign language is rather difficult to learn and may take a prolonged period for some individuals to learn. These application potentials have made sign language an interesting re-search topic especially in the field of computer science and modelling. This re-search aims to address one of the prominent problems in the sign language recognition process, namely the detection of hand shape and pose. Based on experimentation, the proposed Concatenation Keypoint Transform (CKT) method achieved 92% average accuracy for hand pose or gesture detection. In addition, the method achieved a high average accuracy during the recognition process. The result of this research is valuable for the subsequent step of converting sign language gestures into text-based language. It is expected the proposed method can contribute to alleviating the learning process for people who are interested in learning sign languages.