Project Aria:一种用于自我中心多模态 AI 研究的新工具
人机交互
2023-10-03 v3 计算机视觉与模式识别
摘要
未来增强现实(AR)设备上可用的自我中心多模态数据给机器感知带来了独特挑战与机遇。这些未来设备需要以社会可接受的形态因子全天候可穿戴,以支持始终可用、上下文感知和个性化的 AI 应用。我们在 Meta Reality Labs Research 的团队构建了 Aria 设备,一种自我中心多模态数据记录与流传输设备,旨在促进并加速该领域的研究。在本文中,我们描述了 Aria 设备硬件,包括其传感器配置以及支持此类数据记录与处理的相应软件工具。
引用
@article{arxiv.2308.13561,
title = {Project Aria: A New Tool for Egocentric Multi-Modal AI Research},
author = {Jakob Engel and Kiran Somasundaram and Michael Goesele and Albert Sun and Alexander Gamino and Andrew Turner and Arjang Talattof and Arnie Yuan and Bilal Souti and Brighid Meredith and Cheng Peng and Chris Sweeney and Cole Wilson and Dan Barnes and Daniel DeTone and David Caruso and Derek Valleroy and Dinesh Ginjupalli and Duncan Frost and Edward Miller and Elias Mueggler and Evgeniy Oleinik and Fan Zhang and Guruprasad Somasundaram and Gustavo Solaira and Harry Lanaras and Henry Howard-Jenkins and Huixuan Tang and Hyo Jin Kim and Jaime Rivera and Ji Luo and Jing Dong and Julian Straub and Kevin Bailey and Kevin Eckenhoff and Lingni Ma and Luis Pesqueira and Mark Schwesinger and Maurizio Monge and Nan Yang and Nick Charron and Nikhil Raina and Omkar Parkhi and Peter Borschowa and Pierre Moulon and Prince Gupta and Raul Mur-Artal and Robbie Pennington and Sachin Kulkarni and Sagar Miglani and Santosh Gondi and Saransh Solanki and Sean Diener and Shangyi Cheng and Simon Green and Steve Saarinen and Suvam Patra and Tassos Mourikis and Thomas Whelan and Tripti Singh and Vasileios Balntas and Vijay Baiyya and Wilson Dreewes and Xiaqing Pan and Yang Lou and Yipu Zhao and Yusuf Mansour and Yuyang Zou and Zhaoyang Lv and Zijian Wang and Mingfei Yan and Carl Ren and Renzo De Nardi and Richard Newcombe},
journal= {arXiv preprint arXiv:2308.13561},
year = {2023}
}