Geospatial AI's Reference Repository

We’re actively looking for ambitious researchers to join our mission. Come join us.

Consider this a repository of Geospatial AI papers and works.

Earth Observation Models / GFMs / Remote Sensing

• Neural Plasticity-Inspired Multimodal Foundation Model for Earth ObservationPaper
• Clay-CNN Hybrids: Leveraging Geospatial Foundation Models as Auxiliary Context for Landslide DetectionPaper
• HySens: Sensor-Agnostic Foundation Models for Hyperspectral DataPaper
• Spectral indices outperform AlphaEarth foundation embeddings for aboveground biomass estimation in a regenerating tropical Andean forestPaper
• AlphaEarth Foundations: An embedding field model for accurate and efficient global mapping from sparse label dataPaper
• SpectralEarth: Training Hyperspectral Foundation Models at ScalePaper
• Prithvi-EO-2.0: A Versatile Multi-Temporal Foundation Model for Earth Observation ApplicationsPapercode
• Prithvi WxC: Foundation Model for Weather and ClimatePaper
• LandSegmenter: Towards a flexible foundation model for Land Use and Land Cover mappingPaper
• SatCLIP: Global, General-Purpose Location Embeddings with Satellite ImageryPaper
• EarthPT: a time series foundation model for Earth ObservationPaper
• Earth AI: Unlocking Geospatial Insights with Foundation Models and Cross-Modal ReasoningPaper
• TerraMind: Large-Scale Generative Multimodality for Earth ObservationPaper
• Leveraging Image Generators to Address Training Data Scarcity: The Gen4Regen Dataset for Forest Regeneration MappingPaperalternative version
• AnySat: One Earth Observation Model for Many Resolutions, Scales, and ModalitiesPaper
• RingMo-Agent: A Unified Remote Sensing Foundation Model for Multi-Platform and Multi-Modal ReasoningPaper
• Clay EO Non-Profit Embeddings: LGND Clay v1.5 Sentinel-2Model
• CROMA: Remote Sensing Representations with Contrastive Radar-Optical Masked AutoencodersPaper
• RemoteCLIP: A Vision Language Foundation Model for Remote SensingPaper
• OmniSat: Self-Supervised Modality Fusion for Earth ObservationPaper
• Panopticon: Advancing Any-Sensor Foundation Models for Earth ObservationPaper
• Efficient Self-Supervised Learning for Earth Observation via Dynamic Dataset CurationPaper

Earth Reasoning Models

• Earth Science Foundation Models: From Perception to Reasoning and DiscoveryPaper
• General Geospatial Inference with a Population Dynamics Foundation ModelPaper
• GeoSearch: Augmenting Worldwide Geolocalization with Web-Scale Reverse Image Search and Image MatchingPaper

World Models

• V-JEPA 2: Self-Supervised Video Models Enable Understanding, Prediction and PlanningPaper
• Causal-JEPA: Learning World Models through Object-Level Latent MaskingPaper
• Learning and Leveraging World Models in Visual Representation LearningPaper
• NVIDIA OmniDreams: Real-Time Generative World Model for Closed-Loop Autonomous Vehicle SimulationPaper
• You Don't Need Strong Assumptions: Visual Representation Learning via Temporal DifferencesPaper
• Skyfall-GS: Synthesizing Immersive 3D Urban Scenes from Satellite ImageryPaper
• Introduction to Latent Variable Energy-Based Models: A Path Towards Autonomous Machine IntelligencePaper

Earth-Observation; Weather / Climate Foundation Models

• GraphCast: Learning skillful medium-range global weather forecastingPaper
• Pangu-Weather: A 3D High-Resolution Model for Fast and Accurate Global Weather ForecastPaper
• GenCast: Diffusion-based ensemble forecasting for medium-range weatherPaper
• ClimaX: A foundation model for weather and climatePaper
• AtmoRep: A stochastic model of atmosphere dynamics using large scale representation learningPaper
• Aardvark weather: end-to-end data-driven weather forecastingPaper
• Neural General Circulation Models for Weather and ClimatePaper
• A Foundation Model for the Earth SystemPaper

Urban Intelligence

• Towards Urban General Intelligence: A Review and Outlook of Urban Foundation ModelsPaper
• Transferable human mobility network reconstruction with neuroGravityPaperNature
• Reconstructing urban mobility from the built environmentNature

Representation Learning

• Better Together: Evaluating the Complementarity of Earth Embedding ModelsPaper
• Spatial Representation Learning Beyond Pixels: Unifying Raster Data and Vector Semantics for Human-Centric Geospatial Foundation ModelsPaper
• MMEarth: Exploring Multi-Modal Pretext Tasks For Geospatial Representation LearningPaper
• SeCo: Exploring Sequence Supervision for Unsupervised Representation LearningPaper
• Scale-MAE: A Scale-Aware Masked Autoencoder for Multiscale Geospatial Representation LearningPaper
• Time2Vec: Learning a Vector Representation of TimePaper

Geographic Knowledge Representation & Spatial Reasoning / Urban Foundation Models

• A Survey on Spatio-Temporal Knowledge Graph ModelsPaper
• Geometric Feature Enhanced Knowledge Graph Embedding and Spatial ReasoningPaper
• Spatio-Temporal Graph Neural Networks for Predictive Learning in Urban Computing: A SurveyPaper
• Learning urban region representations with POIs and hierarchical graph infomaxPaper
• GN-GCN: Grid neighborhood-based graph convolutional network for spatio-temporal knowledge graph reasoningPaper

Human Geographic Intelligence Models

• From Points to Places: Towards Human Mobility-Driven Spatiotemporal Foundation Models via Understanding PlacesPaper
• Learning Universal Human Mobility Patterns with a Foundation Model for Cross-domain Data FusionPaper
• MoveGPT: Scaling Mobility Foundation Models with Spatially-Aware Mixture of ExpertsPaper

Spatial Foundation Models

• PlaceRep: Geospatial Place Representation Learning from Large-Scale Point-of-Interest DataPaper
• POI2Vec: Geographical Latent Representation for Predicting Future VisitorsPaper
• Enriching Location Representation with Detailed Semantic InformationPaper
• Hex2vec -- Context-Aware Embedding H3 Hexagons with OpenStreetMap TagsPaper
• Activity-aware urban area embedding with contrastive learning for intelligent transportation systems applicationsPaper

Geospatial Retrieval Papers

• Spatial-RAG: Spatial Retrieval Augmented Generation for Real-World Geospatial Reasoning QuestionsPaper
• Retrieval-Augmented Search for Large-Scale Map Collections with ColPaliPaper

Geospatial AI Agents

• An Autonomous GIS Agent Framework for Geospatial Data RetrievalPaper
• Earth-Agent: Unlocking the Full Landscape of Earth Observation with AgentsPaper
• Intelligent Multimodal Retrieval and Reasoning for Geospatial Knowledge Discovery on the I-GUIDE PlatformPaper

Benchmarks

• EarthShift: a benchmark for measuring robustness to real-world distribution shifts in Earth observationPaper
• EarthVQA: Towards Queryable Earth via Relational Reasoning-Based Remote Sensing Visual Question AnsweringPaper
• GEO-Bench: Toward Foundation Models for Earth MonitoringPaper
• GEO-Bench-2: From Performance to Capability, Rethinking Evaluation in Geospatial AIPaper
• MMEarth-Bench: Global Model Adaptation via Multimodal Test-Time TrainingPaper
• Assessing the value of Geo-Foundational Models for Flood Inundation Mapping: Benchmarking models for Sentinel-1, Sentinel-2, and Planetscope for end-usersPaper
• GeoQA: A Geometric Question Answering Benchmark Towards Multimodal Numerical ReasoningPaper
• CityBench: Evaluating the Capabilities of Large Language Models for Urban TasksPaper

Spatial Intelligence & 3D World Reconstruction Papers by Nvidia

• ArtiFixer: Enhancing and Extending 3D Reconstruction with Auto-Regressive Diffusion ModelsPaper
• ArtisanGS: Interactive Tools for Gaussian Splat Selection with AI and Human in the LoopPaper
• Déjà View: Looping Transformers for Multi-View 3D ReconstructionPaper
• TRON: Tracing Rays to Orchestrate a Neural Renderer for 3D Gaussian ReconstructionsPaper
• MetaEarth3D: Unlocking World-scale 3D Generation with Spatially Scalable Generative ModelingPaper

Geotuned LLMs and VLMs

• GeoLLM: Extracting Geospatial Knowledge from Large Language ModelsPaper
• GeoLLaVA-8K: Scaling Remote-Sensing Multimodal Large Language Models to 8K ResolutionPaper
• GeoChat: Grounded Large Vision-Language Model for Remote SensingPaper
• GeoLM: Empowering Language Models for Geospatially Grounded Language UnderstandingPaper
• GeoX: Geometric Problem Solving Through Unified Formalized Vision-Language Pre-trainingPaper

Geospatial Surveys for models

• Foundation Models for Remote Sensing and Earth Observation: A SurveyPaper
• Towards Vision-Language Geo-Foundation Model: A SurveyPaper
• On the Foundations of Earth and Climate Foundation ModelsPaper
• Self-supervised Learning in Remote Sensing: A ReviewPaper
• SSL4EO-S12: A Large-Scale Multi-Modal, Multi-Temporal Dataset for Self-Supervised Learning in Earth ObservationPaper
• Survey of Multimodal Geospatial Foundation Models: Techniques, Applications, and ChallengesPaper

Public Labeled Geospatial Datasets

Foundation Model Pretraining

• SSL4EO-S12: A Large-Scale Multi-Modal, Multi-Temporal Dataset for Self-Supervised Learning in Earth Observationcode
• SSL4EO-L: Datasets and Foundation Models for Landsat ImageryPaper
• Clay Open DataDataset