notesum.ai

Published at November 28

GEOBench-VLM: Benchmarking Vision-Language Models for Geospatial Tasks

cs.CV

Released Date: November 28, 2024

Authors: Muhammad Sohail Danish¹, Muhammad Akhtar Munir¹, Syed Roshaan Ali Shah², Kartik Kuckreja¹, Fahad Shahbaz Khan³, Paolo Fraccaro⁴, Alexandre Lacoste⁵, Salman Khan⁶

Aff.: ¹Mohamed bin Zayed University of Artificial Intelligence; ²University College London; ³Mohamed bin Zayed University of Artificial Intelligence, Linkoping University, Sweden; ⁴IBM Research Europe, UK; ⁵ServiceNow Research; ⁶Mohamed bin Zayed University of Artificial Intelligence, Australian National University

Arxiv: http://arxiv.org/pdf/2411.19325v1

Model	Crop	Disaster	Farm	Land	Damaged
	Type Cls	Type Cls	Pond CD	Use Cls	Bldg Cnt
LLaVA-OneV [23]	0.1273	0.4493	0.1579	0.5672	0.2139
Qwen2-VL [53]	0.1273	0.5903	0.0921	0.5869	0.2270
GPT-4o	0.1818	0.6344	0.1447	0.6230	0.2420