研究资源：论文与软件
Autonomous GeoAI 相关代表性论文，以及 Python、R 和 GIS 集成工具。
来源与版本
本页整理自 AutoGeoAI4Sci/awesome-autonomous-geoai，上游 commit
2437033（2026-07-26）。 GIStudioNote 只调整文档结构与导航；资源描述和外部链接来自上游，时效性与准确性请以原项目及链接目标为准。
Selected Papers
A curated list of representative papers related to computer vision, GeoAI, and remote sensing, with an emphasis on CVPR and top-tier computer vision workshops.
-
EarthVision @ CVPR — Large Scale Computer Vision for Remote Sensing Imagery
A long-running CVPR workshop focusing on computer vision and machine learning methods for Earth observation and remote sensing.
📄 Proceedings: https://openaccess.thecvf.com/CVPR2025_workshops/EarthVision -
MORSE @ CVPR — Foundation and Large Vision Models in Remote Sensing
Workshop dedicated to foundation models, large vision models, and multimodal learning for remote sensing and geospatial analysis.
📄 Proceedings: https://openaccess.thecvf.com/CVPR2025_workshops/MORSE -
CV4EO @ WACV — Computer Vision for Earth Observation Applications
Focuses on practical and methodological advances in applying computer vision to Earth observation data.
🔗 Workshop hub: https://wacv.thecvf.com/Conferences/2026/Workshops
📌 This list will be continuously updated with papers from CVPR, ICCV, ECCV, ICLR, AAAI, and KDD workshops and tracks related to GeoAI, remote sensing, and AI4Science.
Software
Focus on accessible introductions to computational tools for GeoAI and AI4Science,
with an emphasis on open-source, reproducible, and research-oriented software ecosystems.
Python
-
Awesome Python — A curated list of Python frameworks, libraries, and software
https://github.com/vinta/awesome-python -
PyTorch — Deep learning framework widely used in GeoAI and AI4Science
https://pytorch.org/ -
Hugging Face Transformers — Foundation models and multimodal learning toolkit
https://huggingface.co/docs/transformers -
gfm-leaderboard — Open-source benchmark and leaderboard infrastructure for comparing geospatial foundation models, with emphasis on evaluation transparency, reproducibility, and cross-model assessment in Earth observation workflows.
https://github.com/taylor-geospatial/gfm-leaderboard -
GeoPandas — Geospatial data structures and spatial operations in Python
https://geopandas.org/ -
Rasterio — Raster data access and processing
https://rasterio.readthedocs.io/ -
xarray — Labeled multi-dimensional arrays for climate and Earth system data
https://docs.xarray.dev/ -
scikit-learn — Classical machine learning and data mining
https://scikit-learn.org/
R
-
Awesome R — Curated resources for R programming and data science
https://github.com/qinwf/awesome-R -
sf — Simple features for vector spatial data
https://r-spatial.github.io/sf/ -
terra — Modern raster data processing in R
https://rspatial.org/terra/ -
tidyverse — Data science tools for structured data analysis
https://www.tidyverse.org/
GIS-Integrated Tools
-
ArcGIS Pro / ArcGIS API for Python — GIS-native spatial analysis and GeoAI workflows
https://developers.arcgis.com/python/ -
QGIS — Open-source desktop GIS with extensive plugin ecosystem
https://www.qgis.org/ -
Google Earth Engine — Cloud-based geospatial analysis platform for Earth observation
https://earthengine.google.com/ -
Google Flood Hub: Public Google Research platform for AI-driven flood forecasting, river-gauge monitoring, flood alerts, and local inundation probability/history layers.
It combines hydrology and inundation models to support flood preparedness up to 7 days in advance. Intended users include disaster agencies, NGOs, researchers, and people in flood-prone areas. Forecasts are informational and should be checked against official local sources.
https://sites.research.google/floods/l/0/0/3 -
GRASS GIS — Advanced geospatial analysis and modeling system
https://grass.osgeo.org/ -
PostGIS — Spatial database extension for PostgreSQL
https://postgis.net/ -
CesiumJS — Open-source JavaScript library for 3D globes and maps; renders high-precision geospatial data and 3D tiles in the browser for visualization and digital-twin applications
https://cesium.com/platform/cesiumjs/