The offical implementaion of "Open-set Hierarchical Semantic Segmentation for 3D Scene".
conda env create -f enviroment.yaml
pip install git+https://github.com/NVlabs/nvdiffrast
conda activate tree_seg
cd semantic_sam/ops
bash ./make.sh
cd tree_segmentation/extension/_C
python setup.py build_ext --inplaceDownload sam_vit_h_4b8939.pth from Segment Anything in floder weights
Download swinl_only_sam_many2many.pth and swint_only_sam_many2many.pth from Sematic-SAM in floder weights
@InProceedings{icme24_ots,
title = {Open-set Hierarchical Semantic Segmentation for 3D Scene},
author = {Wan, Diwen and Tang, Jiaxiang and Wang, Jingbo and Chen, Xiaokang and Gan, Lingyun and Zeng, Gang},
booktitle = {IEEE Conference on Multimedia Expo 2024},
year = {2024},
}