Finding Change in Satellite Archives from Text: How to Combine Before-and-After Images Efficiently

2026-07-30

AI Summary

This study addresses the task of retrieving satellite image "changes" described by natural language queries. Under a frozen CLIP vision encoder and a unified training framework, it systematically evaluates the trade-offs between efficiency and accuracy across eight temporal fusion architectures. The authors propose a two-stage, training-free retrieval pipeline that substantially reduces query cost. They further introduce a Temporal Bottleneck Fusion (TBF) module, which cuts model parameters by 2.3× and latency by 1.6× at the expense of only a 0.007 drop in BLEU-1 score. Experiments demonstrate that Mamba offers no speed advantage for typical remote sensing sequence lengths. On the LEVIR-CC dataset, the method achieves comparable or superior recall with 10–15× lower query cost, and results are statistically significant across ten random seeds.

Abstract

Operational Earth observation increasingly calls for answering queries such as ``find the image pairs where a new building appeared.'' This means searching an archive of before-and-after (bi-temporal) satellite image pairs and ranking each pair by how well it matches a natural-language description of the change. The component that performs this match, the fusion module that combines the ``before'' and ``after'' views, must be run at query time across many candidate pairs, so its speed largely sets the cost of every search. We present a controlled comparison of how to build that module. Using one fixed image encoder (a frozen CLIP model) and one training recipe for all variants, we evaluate eight designs drawn from three families: attention, state-space models (Mamba), and learned compression (our Temporal Bottleneck Fusion, TBF). Each design is tested on two benchmarks (LEVIR-CC and Dubai-CC) with ten random seeds, so the reported differences are statistically grounded. We outline three findings: first, a training-free two-stage search (a cheap difference model that shortlists candidates, followed by attention fusion that re-ranks them) matches or exceeds full-fusion recall on LEVIR-CC while cutting query cost $10$-$15\times$, with comparable R@1/R@5 on Dubai-CC; second, the linear-time scan of Mamba, attractive on paper, gives no speed benefit at the patch counts typical of vision transformers ($L{=}196$): the scan is limited by memory bandwidth, whereas attention maps cleanly onto parallel hardware; and third, compressing the fused representation (TBF) reduces parameters by $2.3\times$ and latency by $1.6\times$ for a change-only BLEU-1 cost of $0.007$, although more aggressive compression quietly discards change-relevant detail that aggregate metrics fail to reveal.

Problem

• satellite image retrieval
• bi-temporal change detection
• natural language query
• image fusion
• efficient search

Innovation

• bi-temporal fusion
• Temporal Bottleneck Fusion
• efficient satellite change detection
• vision-language retrieval
• Mamba vs attention