---
license: apache-2.0
datasets:
- vikki23/LAS
---
# Model Card for LandSegmenter

<!-- Provide a quick summary of what the model is/does. -->

This repo provides the weights for LandSegmenter, a task-specific Foundation Model (FM) for Land Use and Land Cover (LULC) mapping, characterized by high flexibility in both inputs (multi-band, multi-resolution imagery) and outputs (customizable category definitions), supporting both zero-shot inference and fine-tuning.

## Model Details

### Model Description

<!-- Provide a longer summary of what this model is. -->
LandSegmenter is the first LULC FM trained with LAS, designed with high flexibility to handle diverse input modalities and customizable category settings. It builds on [SAM2](https://github.com/facebookresearch/sam2)'s backbone for robust multi-scale spatial representation, further enhanced by multispectral features from [DOFA](https://github.com/zhu-xlab/DOFA) and high-frequency components for refined structural details. A text-based prompter derived from [GeoRSCLIP](https://github.com/om-ai-lab/RS5M), which takes class names as inputs, further strengthens its semantic understanding, enabling concept-aware and adaptable segmentation across heterogeneous data sources


- **Developed by:** Chenying Liu, Wei Huang, Xiao Xiang Zhu
- **Funded by:** Munich Center for Machine Learning (MCML) 

### Model Sources

<!-- Provide the basic links for the model. -->

- **Repository:** [GitHub](https://github.com/zhu-xlab/LandSegmenter)
- **Paper [optional]:** [arXiv](https://arxiv.org/abs/2511.08156)

## Uses

Please refer to the GitHub repository for implementation details.

## Model Card Authors

Chenying Liu

## Model Card Contact

chenying.liu@tum.de; chenying.liu023@gmail.com