@inproceedings{miyanishi_2024_3DV,
title={Cross3DVG: Cross-Dataset 3D Visual Grounding on Different RGB-D Scans},
author={Miyanishi, Taiki and Azuma, Daichi and Kurita, Shuhei and Kawanabe, Motoaki},
booktitle={The 10th International Conference on 3D Vision},
year={2024}
}
Using crowdsourcing services, we collected 63,602 descriptions for approximately 249 unique objects across 1,380 scans as a RIORefer dataset.
Paper Code