Revisiting Change VQA in Remote Sensing with Structured and Native Multimodal Qwen Models

Yakoub Bazi, Mohamad M. Al Rahhal, Mansour Zuair, Faroun Mohamed · 2026

Read Revisiting Change VQA in Remote Sensing with Structured and Native Multimodal Qwen Models by Yakoub Bazi, Mohamad M. Al Rahhal, Mansour Zuair, Faroun Mohamed (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-18429.txt.

Description

Bazi, Al Rahhal, Zuair, and Mohamed revisit change question-answering for remote sensing with modern Qwen vision-language models. Change VQA asks natural-language questions about semantic shifts between two satellite images taken at different times. Under a unified low-rank adaptation setup, the structured Qwen3-VL pipeline faces the native multimodal Qwen3.5 model on the CDVQA benchmark. Recent models beat older specialized baselines, size alone does not predict success, and native multimodal designs prove more effective than structured pipelines.

Catalog record

CREATORYakoub Bazi, Mohamad M. Al Rahhal, Mansour Zuair, Faroun Mohamed
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE22236
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.18429

Browse the full library · About the Tower of Babel