---
license: bsd-3-clause
tags:
- 3d-instance-segmentation
- point-cloud
- scannet
- s3dis
- stpls3d
---

# ISBNet: a 3D Point Cloud Instance Segmentation Network with Instance-aware Sampling and Box-aware Dynamic Convolution

Pretrained weights for **ISBNet** (CVPR 2023).

- Paper: [arXiv:2303.00246](https://arxiv.org/abs/2303.00246)
- Code: [github.com/VinAIResearch/ISBNet](https://github.com/VinAIResearch/ISBNet)

## Checkpoints

Full models (use with `tools/test.py`):

| Dataset | AP | AP_50 | Config | File |
|:-:|:-:|:-:|:-:|:-:|
| ScanNetV2 val | 56.8 | 73.3 | [config](https://github.com/VinAIResearch/ISBNet/blob/master/configs/scannetv2/isbnet_scannetv2.yaml) | [`scannetv2/head_scannetv2_val.pth`](https://huggingface.co/TuanNgo/ISBNet/resolve/main/scannetv2/head_scannetv2_val.pth) |
| ScanNetV2 val (lightweight) | 50.1 | 68.9 | [config](https://github.com/VinAIResearch/ISBNet/blob/master/configs/scannetv2/isbnet_lightweight_scannetv2.yaml) | [`scannetv2/head_small_scannetv2_val.pth`](https://huggingface.co/TuanNgo/ISBNet/resolve/main/scannetv2/head_small_scannetv2_val.pth) |
| ScanNet200 val | 24.5 | 32.7 | [config](https://github.com/VinAIResearch/ISBNet/blob/master/configs/scannet200/isbnet_scannet200.yaml) | [`scannet200/head_scannetv2_200_val.pth`](https://huggingface.co/TuanNgo/ISBNet/resolve/main/scannet200/head_scannetv2_200_val.pth) |
| S3DIS Area 5 | 56.3 | 67.5 | [config](https://github.com/VinAIResearch/ISBNet/blob/master/configs/s3dis/isbnet_s3dis_area5.yaml) | [`s3dis/head_s3dis_area5.pth`](https://huggingface.co/TuanNgo/ISBNet/resolve/main/s3dis/head_s3dis_area5.pth) |
| STPLS3D val | 51.2 | 66.7 | [config](https://github.com/VinAIResearch/ISBNet/blob/master/configs/stpls3d/isbnet_stpls3d.yaml) | [`stpls3d/head_stpls3d_val.pth`](https://huggingface.co/TuanNgo/ISBNet/resolve/main/stpls3d/head_stpls3d_val.pth) |

Pretrained 3D U-Net backbones (the `pretrain:` field of each config):

| Dataset | File |
|:-:|:-:|
| ScanNetV2 val | [`scannetv2/pretrain_scannetv2_val.pth`](https://huggingface.co/TuanNgo/ISBNet/resolve/main/scannetv2/pretrain_scannetv2_val.pth) |
| ScanNetV2 train+val (benchmark) | [`scannetv2/pretrain_scannetv2_trainval.pth`](https://huggingface.co/TuanNgo/ISBNet/resolve/main/scannetv2/pretrain_scannetv2_trainval.pth) |
| ScanNet200 val | [`scannet200/pretrain_scannetv2_200_val.pth`](https://huggingface.co/TuanNgo/ISBNet/resolve/main/scannet200/pretrain_scannetv2_200_val.pth) |
| S3DIS Area 5 | [`s3dis/pretrain_s3dis_area5.pth`](https://huggingface.co/TuanNgo/ISBNet/resolve/main/s3dis/pretrain_s3dis_area5.pth) |
| STPLS3D val | [`stpls3d/pretrain_stpls3d_val.pth`](https://huggingface.co/TuanNgo/ISBNet/resolve/main/stpls3d/pretrain_stpls3d_val.pth) |

Also included: `scannetv2/head_3dyco_scannetv2_val.pth`.

## Usage

```bash
# Download everything into ./pretrains (matches the default `pretrain:` paths in the configs)
hf download TuanNgo/ISBNet --local-dir pretrains

# Evaluate
python3 tools/test.py configs/scannetv2/isbnet_scannetv2.yaml pretrains/scannetv2/head_scannetv2_val.pth
```

## Citation

```bibtex
@inproceedings{ngo2023isbnet,
 author={Tuan Duc Ngo, Binh-Son Hua, Khoi Nguyen},
 booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
 title={ISBNet: a 3D Point Cloud Instance Segmentation Network with Instance-aware Sampling and Box-aware Dynamic Convolution},
 year= {2023}
}
```