[CVPR 2026 (Highlight)] Uni3R: Unified 3D Reconstruction and Semantic Understanding via Generalizable Gaussian Splatting from Unposed Multi-View Images

Xiangyu Sun*, Haoyi Jiang*, Liu Liu, Seungtae Nam, Gyeongjin Kang, Xinjie Wang,
Wei Sui, Zhizhong Su, Wenyu Liu, Xinggang Wang, Eunbyung Park

Download

The root config.yaml lists all checkpoint variants, their file sizes and SHA256 digests. Download it with the checkpoint you need. For example:

from huggingface_hub import snapshot_download

model_dir = snapshot_download(
    "HorizonRobotics/Uni3R",
    allow_patterns=["config.yaml", "arbitrary_views.pth"],
    local_dir="Uni3R",
)

For 2views, 8views or 16views, replace arbitrary_views.pth in allow_patterns with the corresponding pattern, such as 2views.pth.part_*. Concatenate every part for that variant in the order listed in config.yaml before loading it. For example, from the downloaded directory:

cat 2views.pth.part_aa 2views.pth.part_ab 2views.pth.part_ac > 2views.pth

Use the Uni3R code to load the assembled checkpoint or the single-file arbitrary_views.pth checkpoint.

Citation

@misc{sun2026uni3runified3dreconstruction,
      title={Uni3R: Unified 3D Reconstruction and Semantic Understanding via Generalizable Gaussian Splatting from Unposed Multi-View Images}, 
      author={Xiangyu Sun and Haoyi Jiang and Liu Liu and Seungtae Nam and Gyeongjin Kang and Xinjie Wang and Wei Sui and Zhizhong Su and Wenyu Liu and Xinggang Wang and Eunbyung Park},
      year={2026},
      eprint={2508.03643},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2508.03643}, 
}