Finding Change in Satellite Archives from Text: How to Combine Before-and-After Images Efficiently
2026-07-30 • Computer Vision and Pattern Recognition
Computer Vision and Pattern RecognitionInformation Retrieval
AI summaryⓘ
The authors study how to quickly find pairs of satellite images showing changes, like new buildings, from large archives using natural language descriptions. They test eight different ways to combine 'before' and 'after' images, focusing on speed and accuracy. Their results show a two-step method that first narrows down candidates cheaply and then carefully checks them works well and saves a lot of time. They also find that some new methods don’t speed up as expected due to hardware limits, and compressing data can help but might lose important details. Their work helps improve searching changes in Earth observation images efficiently.
Earth observationbi-temporal satellite imageschange detectionfusion moduleCLIP modelattention mechanismstate-space modelslearned compressionMambaTemporal Bottleneck Fusion (TBF)
Authors
Simon Roy, Mark Bong, Giovanni Beltrame
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.