Hand Palm Motions (HPM) dataset
收藏资源简介:
本研究介绍了KU Leuven研究机构创建的Hand Palm Motions (HPM)数据集,该数据集包含手部掌部运动的3D时空轨迹,旨在用于稳健的手部掌部运动手势识别。数据集由35个基本手势记录组成,通过人工变换和扰动扩展到420个试验,以模拟不同的上下文变化。这些手势运动易于执行,且设计时不依赖特定的坐标参考框架或方向,使得数据集在不变性描述符方法研究中具有重要价值。
This study presents the Hand Palm Motions (HPM) dataset developed by the KU Leuven research institute. This dataset contains 3D spatiotemporal trajectories of hand palm motions, and is intended for robust hand palm motion gesture recognition. The dataset consists of 35 basic gesture recordings, which are expanded to 420 trials via artificial transformations and perturbations to simulate diverse contextual variations. These gesture motions are easy to perform and designed without relying on specific coordinate reference frames or orientations, making the dataset highly valuable for research on invariance descriptor-based methods.




