The HRI30 database for industrial action recognition from videos contains 30 categories of industrial-like actions and 2940 manually annotated clips. The classes are of three types: Human-Object Perception, Body-Motion Only, and Human-Robot Collaboration.
A thorough analysis of the existing human action recognition datasets demonstrates that only a few HRI datasets are available that target real-world applications, all of which are adapted to home settings. Given the shortage of datasets in industrial tasks, HRI30 aims to provide the community with a dataset created in a laboratory setting that includes actions commonly performed within manufacturing and service industries. In addition, the dataset meets the requirements of deep learning algorithms for the development of intelligent learning models for action recognition and imitation in HRI applications.
@inproceedings{9956300,
title = {HRI30: An Action Recognition Dataset for Industrial Human-Robot Interaction},
author = {Iodice, Francesco and De Momi, Elena and Ajoudani, Arash},
booktitle = {2022 26th International Conference on Pattern Recognition (ICPR)},
year = {2022},
doi = {10.1109/ICPR56361.2022.9956300}
}
