Earth Vision AI Alliance

📰 News

• [2025-11] GeoVLM-R1 is available on arXiv.
• [2025-05] GEOBench-VLM is accepted at ICCV 2025.
• [2025-05] ThinkGeo is now publicly available.
• [2025-06] EarthDial paper accepted at CVPR 2025.

👩‍🔬 People

Dr. Salman Khan

Prof. MBZUAI

Dr. Fahad Khan

Prof. MBZUAI

Dr. Shantanu Godbole

Team Lead, IBM

Dr. Paolo Fraccaro

Senior Software Developer, IBM

Dr. Mustansar Fiaz

Researcher, IBM

Dr. Hiyam Debary

Researcher, IBM

Dr. Zaigham Zaheer

Researcher, MBZUAI

Dr. Muhammad Akhter

Researcher, MBZUAI

💡 Projects

GeoVLMs

GeoVLMs are multimodal models that analyze geospatial imagery, enabling users to extract insights and reason about spatial information through natural language queries.

GEOBench-VLM

A comprehensive benchmarking framework for evaluating vision-language models on geospatial tasks, including temporal analysis, fine-grained object detection, damage assessment, and spatial reasoning capabilities.

EarthVision Agents

Agent frameworks for Earth AI that enable autonomous systems to perform complex Earth Observation tasks through structured reasoning, tool use, and multi-step problem solving in remote sensing applications.

📄 Publications

• [2025] Simumba, Naomi, et al. "Geo-bench-2: From performance to capability, rethinking evaluation in geospatial ai." arXiv preprint arXiv:2511.15658 (2025).
• [2025] Fiaz, Mustansar, et al. "Geovlm-r1: Reinforcement fine-tuning for improved remote sensing reasoning." arXiv preprint arXiv:2509.25026 (2025).
• [2025] Danish, Muhammad, et al. "Geobench-vlm: Benchmarking vision-language models for geospatial tasks." Proceedings of the IEEE/CVF ICCV. 2025.
• [2025] Shabbir, Akashah, et al. "ThinkGeo: Evaluating Tool-Augmented Agents for Remote Sensing Tasks." arXiv preprint arXiv:2505.23752 (2025).
• [2025] Soni, Sagar, et al. "Earthdial: Turning multi-sensory earth observations to interactive dialogues." Proceedings of the CVPR. 2025.

🙏 Collaboration

📬 Contact