MAGIC: A Dataset Capturing Mid-Air Gesture Performance for Interaction and Feature Analysis

You can download the full dataset from our SFTP-Server: sftp://dareftp.uol.de/ACBGC8 The MAGIC is a multimodal dataset designed to support research on mid-air gesture interaction, gesture recognition, human-computer interaction. The dataset contains gesture performances collected from 42 participants, covering 18 commonly used uni-manual mid-air gestures derived from prior gesture elicitation and interaction design studies.

Each gesture was performed across multiple repetitions under controlled experimental conditions. MAGIC integrates three synchronized sensing modalities: RGB video captured from three Intel RealSense D455 cameras Depth video captured from the same three viewpoints Electromagnetic motion tracking data recorded using a Polhemus VIPER tracking system Dataset Contents gesture recordings of RGB videos, Depth videos, and electromagnetic motion tracking data Funding This project is funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under grant agreement 425868555 and is part of Priority Program SPP2199 on Scalable Interaction Paradigms for Pervasive Computing Environments.