Abstract
The manual transcription process of Sign Language is a workintensive step which requires considerable effort to create Signs. Even, often the result of this step misses the natural aspect of motion to be conform to the natural human interpretation. In other words, the lack of the sign language annotated corpora is closely related to the difficulty of the sign creation task. In this paper, we propose a novel tool Sign(motion) for creating an annotated sign language corpus based on Natural Human Gestures: By overlaying a real signer motion onto an articulated 3D skeleton using Microsoft Kinect and Leap motion sensors. Signmotion is created to support the natural 3D facial expression, the natural 3D body posture and gives the possibility to annotate and analyze each sign and motion in the recorded animation. The resulting data and structure are precise enough to create and to store signs to be used for Sign Language data analysis or Machine Translation using virtual signer.