FAVOR: Full-Body AR-Driven Virtual Object Rearrangement Guided by Instruction Text

Authors: Kailin Li, Lixin Yang, Zenan Lin, Jian Xu, Xinyu Zhan, Yifei Zhao, Pengxiang Zhu, Wenxiong Kang, Kejian Wu, Cewu Lu

AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results, both qualitative and quantitative, suggest that this dataset and pipeline deliver high-quality motion sequences.
Researcher Affiliation Collaboration 1Shanghai Jiao Tong University 2XREAL 3South China University of Technology
Pseudocode No The paper describes its methods and algorithms in text and diagrams (e.g., Figure 3 workflow, descriptions of KNET and INET), but it does not include any explicitly labeled "Pseudocode" or "Algorithm" blocks.
Open Source Code Yes Our dataset, code, and appendix are available at https://kailinli.github.io/FAVOR.
Open Datasets Yes Our dataset, code, and appendix are available at https://kailinli.github.io/FAVOR.
Dataset Splits Yes The dataset is split into training, validation, and test sets in an 8:1:1 ratio, based on motion sequences.
Hardware Specification Yes The infrared motion capture system includes 12 temporally synchronized Optitrack Prime 13W infrared cameras used for tracking reflective markers (Fig. 2 I.). We utilize the XREAL X AR glasses for scene rendering.
Software Dependencies No The paper mentions several software components and models such as GPT-4, Owl-ViT, SMPL-X, Vposer, and HuMoR, but it does not provide specific version numbers for any of these or for any programming languages or libraries used.
Experiment Setup No The paper states: 'Training details are in the Appx.', indicating that specific experimental setup details such as hyperparameters are not provided in the main text.