Benchmarking Composed Image Retrieval for Applied Earth Observation

Hugging Face Daily Papers Papers

Summary

This paper presents a unified benchmark for composed image retrieval in Earth observation, evaluating vision-language backbones and introducing a change-centric dataset (xView2-CIR) for disaster monitoring, highlighting distinct challenges compared to attribute-based retrieval.

Remote sensing composed image retrieval (RSCIR) enables search in large satellite image archives using composed queries that combine a reference image with a textual modifier. Although RSCIR offers a flexible interface for expressing targeted retrieval intent, the transferability of modern composition methods to Earth observation (EO) imagery and their relevance to operational EO workflows remain underexplored. We address this gap through a unified benchmark and an application-oriented study. First, we systematically adapt and evaluate representative composed image retrieval methods with six vision-language backbones on PatternCom under a standardized protocol, analyzing their behavior across backbones, composition strategies, and query types. Second, we introduce xView2-CIR, a change-centric dataset for disaster and damage monitoring, where retrieval is conditioned on scene identity and a target post-event state. Our results show that training-free composition methods provide strong and scalable baselines for EO retrieval, while change-centric retrieval presents different challenges from attribute-based retrieval, particularly due to the need to preserve scene identity. Overall, this study establishes a practical benchmark for RSCIR and positions composed retrieval as a complementary tool for remote sensing image retrieval, archive exploration, and change analysis. The dataset and code are available at https://github.com/billpsomas/rscir.
Original Article
View Cached Full Text

Cached at: 06/01/26, 11:21 AM

Paper page - Benchmarking Composed Image Retrieval for Applied Earth Observation

Source: https://huggingface.co/papers/2605.24442

Abstract

Remote sensing composed image retrieval methods are evaluated across vision-language backbones and a new change-centric dataset, demonstrating their effectiveness for Earth observation applications while highlighting distinct challenges compared to traditional attribute-based retrieval.

Remote sensing composed image retrieval(RSCIR) enables search in large satellite image archives using composed queries that combine a reference image with a textual modifier. Although RSCIR offers a flexible interface for expressing targeted retrieval intent, the transferability of modern composition methods to Earth observation (EO) imagery and their relevance to operational EO workflows remain underexplored. We address this gap through a unified benchmark and an application-oriented study. First, we systematically adapt and evaluate representative composed image retrieval methods with sixvision-language backbonesonPatternComunder a standardized protocol, analyzing their behavior across backbones, composition strategies, and query types. Second, we introducexView2-CIR, a change-centric dataset for disaster and damage monitoring, where retrieval is conditioned onscene identityand a target post-event state. Our results show that training-free composition methods provide strong and scalable baselines for EO retrieval, whilechange-centric retrievalpresents different challenges fromattribute-based retrieval, particularly due to the need to preservescene identity. Overall, this study establishes a practical benchmark for RSCIR and positions composed retrieval as a complementary tool for remote sensing image retrieval, archive exploration, and change analysis. The dataset and code are available at https://github.com/billpsomas/rscir.

View arXiv pageView PDFAdd to collection

Get this paper in your agent:

hf papers read 2605\.24442

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2605.24442 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2605.24442 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2605.24442 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

DataComp-VLM: Improved Open Datasets for Vision-Language Models

Hugging Face Daily Papers

This paper introduces DataComp-VLM (DCVLM), a comprehensive benchmark for evaluating data curation strategies for vision-language models. The authors find that data mixing, rather than filtering, significantly improves performance, and their resulting DCVLM-Baseline dataset achieves state-of-the-art results on 33 downstream tasks.