EV-Action

EV-Action数据集是由东北大学开发的一个大规模多模态动作数据集，包含RGB、深度、肌电图(EMG)和两种骨骼模态。该数据集通过光学跟踪系统Vicon获取高质量的骨骼模态，并首次引入EMG模态，为动作分析提供更多维度的信息。数据集包含7000个动作片段，由70名参与者执行20种常见动作，旨在探索不同模态间的潜在关联，并提升动作分析任务的性能。EV-Action数据集的应用领域广泛，包括人类运动分析、计算机视觉、机器学习、生物力学等跨学科领域。

The EV-Action dataset is a large-scale multimodal action dataset developed by Northeastern University (China). It encompasses RGB, depth, electromyography (EMG), and two skeletal modalities. The high-quality skeletal data is collected via the optical tracking system Vicon, and the EMG modality is introduced for the first time in this dataset, providing multi-dimensional information for action analysis. Comprising 7,000 action clips performed by 70 participants executing 20 common actions, this dataset aims to explore the underlying correlations among different modalities and enhance the performance of action analysis tasks. The EV-Action dataset has broad application prospects across interdisciplinary domains including human motion analysis, computer vision, machine learning, and biomechanics.