Скачать с ютуб [CVPR2024] HOI-M3: Capture Multiple Humans and Objects Interaction within Contextual Environment в хорошем качестве

[CVPR2024] HOI-M3: Capture Multiple Humans and Objects Interaction within Contextual Environment

Project Page: https://juzezhang.github.io/HOIM3_ProjectP... Arxiv: https://arxiv.org/pdf/2404.00299 Humans naturally interact with both others and the surrounding multiple objects, engaging in various social activities. However, due to fundamental data scarcity, recent advances in modeling human-object interactions mostly focus on perceiving isolated individuals and objects. In this paper, we introduce HOI-M3, a novel large-scale dataset for modeling the interactions of Multiple huMans and Multiple objects. It provides accurate 3D tracking for humans and objects from dense RGB and object-mounted IMU inputs, covering 199 sequences and 181M frames of diverse humans and objects under rich activities. With the unique HOI-M$^3$ dataset, we introduce two novel data-driven tasks with companion strong baselines: monocular capture and unstructured generation of multiple human-object interactions. Extensive experiments demonstrate that our dataset is challenging and worthy of further research about multiple human-object interactions and behavior analysis. Juze Zhang*, Jingyan Zhang 1*, Zining Song, Zhanhe Shi, Chengfeng Zhao, Ye Shi, Jingyi Yu, Lan Xu, Jingya Wang, HOI-M3: Capture Multiple Humans and Objects Interaction within Contextual Environment