A Comparative Analysis of Methods for Hand Pose Detection in 3D Environments Academic Article in Scopus uri icon

abstract

  • The ability to discern the pose and gesture of the human hand is of big importance in the field of humancomputer interaction, particularly in the context of sign language interpretation, gesture-based control and augmented reality applications. Some models employ different methodologies to estimate the position of the hand. However, few have provided a comprehensive and objective comparison, resulting in a limited understanding of the approaches among researchers. The present study assesses the efficacy of three-dimensional (3D) hand pose estimation techniques, with a particular focus on those that derive the hand pose directly from depth maps or stereo images. The evaluation of the models considers endpoint pixel error as a principal metric for comparison between methods, with the aim of identifying the most effective approach. The objective is to identify a method that is suitable for virtual reality training considering memory usage, speed, accuracy, adaptability, and robustness. Furthermore, this study can help other researchers understand the construction of such models and develop their own models. © 2024 by SCITEPRESS-Science and Technology Publications, Lda.

publication date

  • January 1, 2024