AdaDINO: Pair-Aware In-Backbone Adaptation of Frozen DINO for Efficient Remote Sensing Change Detection

2026-08-08

AI Summary

This work addresses the limitation of existing vision foundation model–based remote sensing change detection methods, which process bi-temporal images independently and thus fail to effectively model cross-temporal relationships. To overcome this, the authors propose AdaDINO, a framework that introduces a bi-temporal interaction mechanism into a frozen DINO backbone by coupling dual streams at selected layers and injecting shared temporal residuals with opposite signs to explicitly capture temporal dependencies. Key innovations include Change-aware Gated Local Adaptation (CGLA), Batch-shared Block Selection (BSCS), and a CGLA Prior-guided Refinement (CPGR) decoder, enabling efficient computation while keeping the backbone frozen. Experiments demonstrate state-of-the-art performance across four benchmarks, achieving an F1 score of 85.29% on SYSU-CD, a 62.5% reduction in FFN width, and a 1.41× throughput improvement.

Abstract

Vision foundation models (VFMs) such as DINO are pretrained for single-image representation, whereas remote sensing change detection requires reasoning over a bi-temporal pair. Existing VFM-based methods usually encode the two images independently and compare them only afterward, leaving the VFM backbone unaware of cross-temporal relations. To bridge this mismatch, we present AdaDINO, a pair-aware in-backbone adaptation framework that equips a frozen DINO encoder with bi-temporal interaction for efficient change detection. Its core component, Change-aware Gated Local Adaptation (CGLA), couples the two streams after selected frozen blocks and injects a shared temporal residual into them with opposite signs, enhancing genuine change responses while preserving the pair midpoint. Batch-Shared Chunk Selection (BSCS) further reduces feed-forward network (FFN) computation by retaining a batch-shared subset of channel chunks that can be executed as a compact dense FFN. A CGLA-Prior-Guided Refinement (CPGR) decoder reuses encoder-side change responses for coarse-to-fine prediction. Experiments on four remote sensing change detection benchmarks show that AdaDINO achieves competitive or superior performance against VFM-based baselines, with the largest gain on the category-agnostic SYSU-CD dataset. With 62.5% of the FFN hidden width removed, AdaDINO still achieves an F1 score of 85.29% on SYSU-CD while delivering a 1.41$\times$ throughput speedup. The code will be released.

Problem

• remote sensing change detection
• vision foundation models
• bi-temporal image pairs
• cross-temporal relations
• frozen backbone adaptation

Innovation

• pair-aware adaptation
• frozen vision foundation model
• bi-temporal interaction
• efficient change detection
• channel chunk selection