---
license: other
license_name: sam-license
license_link: https://huggingface.co/facebook/sam3/blob/main/LICENSE
base_model: facebook/sam3
tags:
  - lora
  - peft
  - segmentation
  - image-segmentation
  - geospatial
  - sam3
---

# GeoPlanAgent — fine-tuned weights

The trained weights behind
[*Plan2Map: A Multimodal Benchmark for Document-Grounded Geospatial Boundary
Reconstruction from Planning Records*](https://arxiv.org/abs/2606.02747)
([code](https://github.com/fabiandegen/GeoPlanAgent) ·
[project page](https://odeb1.github.io/Plan2Map_Project_Page/)).

Both models are trained as 5-fold cross-validation over a shared case→fold
split: each fold's copy has a fifth of the benchmark cases held out of its
training data, and at inference every case is served by the copy that never
saw it during training.

| Path | What it is | Performance |
|---|---|---|
| `sam3_lora/fold_{0..4}/` | PEFT LoRA adapters for [facebook/sam3](https://huggingface.co/facebook/sam3), fine-tuned to segment drawn planning boundaries on scanned UK planning maps | 0.912 mean pixel IoU |
| `rotation_classifier_kfold/fold_{0..4}/best.pt` | ResNet50 (ImageNet-pretrained) fine-tuned to classify scanned-map orientation (0°/90°/180°/270°) | 0.981 accuracy (with test-time augmentation) |

## Usage

These weights are consumed by the
[GeoPlanAgent pipeline](https://github.com/fabiandegen/GeoPlanAgent), which
handles fold routing, adapter loading, and inference. From the root of a
clone of that repository, download them straight into `models/`:

```bash
hf download fabiandegen/GeoPlanAgent --include "sam3_lora/*" "rotation_classifier_kfold/*" --local-dir models
```

The SAM3 base weights are not included — they download from
[facebook/sam3](https://huggingface.co/facebook/sam3) (gated; accept Meta's
SAM License there) on the pipeline's first run.

## Licence

- `sam3_lora/` — the adapters are fine-tuned from Meta's SAM 3 and are
  therefore distributed under the [SAM License](https://huggingface.co/fabiandegen/GeoPlanAgent/blob/main/SAM_LICENSE).
- `rotation_classifier_kfold/` — fine-tuned from torchvision's
  ImageNet-pretrained ResNet50 (BSD-3-Clause); no additional restrictions.

## Citation

```bibtex
@misc{Plan2Map2026,
      title={Plan2Map: A Multimodal Benchmark for Document-Grounded Geospatial Boundary Reconstruction from Planning Records}, 
      author={Fabian Degen and Oishi Deb and Jindong Gu and Junchi Yu and Samuele Marro and Philip Torr and Jialin Yu},
      year={2026},
      eprint={2606.02747},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2606.02747},
}
```