BiliSakura / TerraMind-1.0-Tokenizer-LULC

TerraMind-1.0-Tokenizer-LULC

This repository provides the LULC tokenizer checkpoint from TerraMind 1.0.

Model details

• Modality: LULC
• Input classes: 10
• Image size: 256x256
• Tokenizer backbone: vit_b_enc
• Quantization: FSQ ( codebook_size=7-5-5-5-5 , latent_dim=5 )

Quick use (diffusers-style)

from huggingface_hub import snapshot_download
import torch
import sys

# Download model repository
model_dir = snapshot_download("BiliSakura/TerraMind-1.0-Tokenizer-LULC")

# Expose local custom module, then load with from_pretrained(...)
sys.path.insert(0, model_dir)
from terramind_tokenizer import TerraMindTokenizer

tokenizer = TerraMindTokenizer.from_pretrained(
    model_dir,
    torch_dtype=torch.float32,
    device="cpu",
)

# LULC input: class map [B, H, W] with labels in [0, 9]
x = torch.randint(0, 10, (1, 256, 256), dtype=torch.long)
tokens = tokenizer.tokenize(x)
print(tokens.shape)  # e.g. [1, 16, 16]

Notes

• The loading API follows the diffusers-style pattern: from_pretrained(...) .
• This package is tokenizer-focused ( tokenize / encode ).
• Please follow TerraMind and TerraTorch licenses/usage terms from the upstream project.

Credits

• Original TerraMind project: https://github.com/IBM/terramind
• Original TerraMind model code (TerraTorch): https://github.com/terrastackai/terratorch/tree/main/terratorch/models/backbones/terramind
• This repository adapts tokenizer checkpoints for convenient Hugging Face usage.

Citation

If you use TerraMind in your research, please cite:

@article{jakubik2025terramind,
  title={TerraMind: Large-Scale Generative Multimodality for Earth Observation},
  author={Jakubik, Johannes and Yang, Felix and Blumenstiel, Benedikt and Scheurer, Erik and Sedona, Rocco and Maurogiovanni, Stefano and Bosmans, Jente and Dionelis, Nikolaos and Marsocci, Valerio and Kopp, Niklas and others},
  journal={IEEE/CVF International Conference on Computer Vision (ICCV)},
  year={2025}
}