Metadata-Version: 2.4
Name: earth-ai-remote-sensing-models
Version: 0.1.1
Summary: Earth AI Remote Sensing Models
Keywords: 
Author-email: Earth AI Remote Sensing Models authors <earth-ai-remote-sensing-external@google.com>
Requires-Python: >=3.11
Description-Content-Type: text/markdown
Classifier: License :: OSI Approved :: Apache Software License
Classifier: Intended Audience :: Science/Research
Classifier: Programming Language :: Python :: 3
Classifier: Operating System :: OS Independent
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
License-File: LICENSE
Requires-Dist: etils[epath]~=1.14.0
Requires-Dist: ml-collections
Requires-Dist: numpy
Requires-Dist: opencv-python
Requires-Dist: overrides
Requires-Dist: pandas
Requires-Dist: scikit-learn
Requires-Dist: scipy
Requires-Dist: matplotlib
Requires-Dist: torch
Requires-Dist: torchvision
Requires-Dist: transformers~=5.5.0
Requires-Dist: typing-extensions
Requires-Dist: pytest ; extra == "dev"
Requires-Dist: pytest-xdist ; extra == "dev"
Requires-Dist: pylint>=2.6.0 ; extra == "dev"
Requires-Dist: pyink ; extra == "dev"
Project-URL: changelog, https://github.com/google-research/remote-sensing/blob/main/CHANGELOG.md
Project-URL: homepage, https://github.com/google-research/remote-sensing
Project-URL: repository, https://github.com/google-research/remote-sensing
Provides-Extra: dev

# 🛰️ Earth AI Remote Sensing Models

*🤝 This is not an officially supported Google product.*

This project is not eligible for the [Google Open Source Software Vulnerability
Rewards Program](https://bughunters.google.com/open-source-security).

Visit the [Earth AI website](https://ai.google/earth-ai/) and
check the [technical report](https://arxiv.org/abs/2510.18318) for additional
information about Earth AI and the different model families.

## 📘 Research Papers

* FM-ChangeNet: Learning Change through Pathwise Feature Transport ([abstract](https://arxiv.org/abs/2607.04750), [pdf](https://arxiv.org/pdf/2607.04750))
* RSRCC: A Remote Sensing Regional Change Comprehension Benchmark Constructed via Retrieval-Augmented Best-of-N Ranking ([abstract](https://arxiv.org/abs/2604.20623), [pdf](https://arxiv.org/pdf/2604.20623))
* GroundSet: A Cadastral-Grounded Dataset for Spatial Understanding with Vector Data (joint work with Roger Ferrod and Sylvain Lobry from Université Paris Cité), ECCV 2026 ([abstract](https://arxiv.org/abs/2603.14609), [pdf](https://arxiv.org/pdf/2603.14609))
* RS-OVC: Open-Vocabulary Counting for Remote-Sensing Data - [pdf](https://link.springer.com/content/pdf/10.1007/978-3-032-31663-9_38), [ICPR 2026 slides](https://icpr2026orgteam.github.io/OralPres/446.pdf), [code](https://github.com/tamirshor7/RS-OVC). ArXiv version: ([abstract](https://arxiv.org/abs/2604.08704), [pdf](https://arxiv.org/pdf/2604.08704))
* Type-Aware Ranking of Urban Similarity from Aerial Imagery ([abstract](https://research.google/pubs/type-aware-ranking-of-urban-similarity-from-aerial-imagery/), [pdf](https://openaccess.thecvf.com/content/WACV2026W/GeoCV/papers/Kligvasser_Type-Aware_Ranking_of_Urban_Similarity_from_Aerial_Imagery_WACVW_2026_paper.pdf))
* A Recipe for Improving Remote Sensing VLM Zero Shot Generalization ([abstract](https://arxiv.org/abs/2503.08722), [pdf](https://arxiv.org/pdf/2503.08722))
* On-the-Fly OVD Adaptation with FLAME: Few-shot Localization via Active Marginal-Samples Exploration ([abstract](https://openaccess.thecvf.com/content/WACV2026W/GeoCV/html/Refael_On-the-Fly_OVD_Adaptation_with_FLAME_Few-shot_Localization_via_Active_Marginal-Samples_WACVW_2026_paper.html), [pdf](https://openaccess.thecvf.com/content/WACV2026W/GeoCV/papers/Refael_On-the-Fly_OVD_Adaptation_with_FLAME_Few-shot_Localization_via_Active_Marginal-Samples_WACVW_2026_paper.pdf))
* Zero-Shot Multi-Spectral Learning: Reimagining a Generalist Multimodal Gemini 2.5 Model for Remote Sensing Applications ([abstract](https://arxiv.org/abs/2509.19087), [pdf](https://arxiv.org/pdf/2509.19087), [blog](https://developers.googleblog.com/unlocking-multi-spectral-data-with-gemini/), [colab](https://github.com/google-gemini/cookbook/blob/main/examples/multi_spectral_remote_sensing.ipynb))
* Enhancing Remote Sensing Representations Through Mixed-Modality Masked Autoencoding ([abstract](https://ieeexplore.ieee.org/document/10972577), [pdf](https://openaccess.thecvf.com/content/WACV2025W/GeoCV/papers/Linial_Enhancing_Remote_Sensing_Representations_Through_Mixed-Modality_Masked_Autoencoding_WACVW_2025_paper.pdf))

## 📊 Datasets

* [RSRCC](https://huggingface.co/datasets/google/RSRCC) - The dataset is designed
  for semantic change understanding in remote sensing, pairing multi-temporal
  image evidence with natural language questions and answers.
* [GroundSet](https://huggingface.co/datasets/RogerFerrod/GroundSet) - This dataset
  by Roger Ferrod and Sylvain Lobry from Université Paris Cité can be used to
  train and evaluate VLMs on a variety of Remote Sensing tasks.

## 💻 Code repository

[Implementation of the Few-shot algorithm](remote_sensing/fewshot/) from the [On-the-Fly OVD Adaptation with FLAME: Few-shot Localization via Active Marginal-Samples Exploration](https://openaccess.thecvf.com/content/WACV2026W/GeoCV/papers/Refael_On-the-Fly_OVD_Adaptation_with_FLAME_Few-shot_Localization_via_Active_Marginal-Samples_WACVW_2026_paper.pdf) paper.

## Remote Sensing Models on Vertex AI Model Garden

[Vertex AI Model Garden](https://console.cloud.google.com/vertex-ai/model-garden/google/earth-ai/)
hosts the Classification and Retrieval Model for Remote Sensing
(the [MaMMUT embeddings model](https://arxiv.org/abs/2503.08722))
and the Object Detection for Remote Sensing (based on the OWL-ViT architecture).
Deployment instructions will be provided when access is granted.

### ⚠️ Additional useful notebooks:

* [Model deployment notebook](remote_sensing/vertex_ai/notebooks/model_garden_remote_sensing_deployment.ipynb)
* [VLM Batch prediction](remote_sensing/vertex_ai/notebooks/vlm_batch_prediction_example.ipynb)


