A Comprehensive Hand Gesture Dataset

Currently, a bunch of research is going in the field of human computer interaction, such as hand gesture recognition, estimation, and hand tracking, one of the important and popular research directions. However existing hand gesture datasets have their limitation such as shortage of number of data, lack of various background and enough volunteers. In order to deal with these problems, this paper proposes a novel and comprehensive dataset of hand gesture. This dataset consists of two parts, Dataset1 and Dataset2. Both datasets include various backgrounds and abundant involved persons. In this paper, a detailed comparison of existing popular datasets is listed. Dataset1, has ten thousand images in total for images classification and Dataset2 contains seven thousand in total for hand tracking. Details about how to build these datasets are represented. Both of datasets are shared online for usage in imaged-based hand gesture recognition research.