ENIGMA-51 is a new dataset composed of 51 egocentric videos acquired in an industrial environment which simulates a real industrial laboratory. The dataset was acquired by 19 subjects who wore a Microsoft HoloLens 2 headset and followed audio and AR instructions provided by the device to complete repairing procedures on electrical boards. The subjects interact with industrial tools such as an electric screwdriver and pliers, as well as with electronic instruments such as a power supply and an oscilloscope while executing the steps to complete a specific procedure. ENIGMA-51 has been annotated with a rich set of annotations which allows to study large variety of tasks, especially tasks related to human-object interactions.
We labelled the ENIGMA-51 dataset with a rich set of fine-grained annotations that can be used and combined to study different aspects of human behavior.
F. Ragusa, R. Leonardi, M. Mazzamuto, C. Bonanno, R. Scavo, A. Furnari, G. M. Farinella. ENIGMA-51: Towards a Fine-Grained Understanding of Human-Object Interactions in Industrial Scenarios. 2023. Cite our paper: ArXiv.
@article{ragusa2023enigma51,
title={ENIGMA-51: Towards a Fine-Grained Understanding of Human-Object Interactions in Industrial Scenarios},
author={Francesco Ragusa and Rosario Leonardi and Michele Mazzamuto and Claudia Bonanno and Rosario Scavo and Antonino Furnari and Giovanni Maria Farinella},
journal = {IEEE Winter Conference on Application of Computer Vision (WACV)},
year = {2024}
}
Visit our page dedicated to First Person Vision Research for other related publications.
More details on the dataset, the annotation phase and the baselines can be found in the supplementary material associated to the publication.
