用门控注意力提升遥感变化检测精度，更省参数。
GRAD-Former: Gated Robust Attention-based Differential Transformer for Change Detection
- 引入门控机制与差分注意力，动态筛选关键特征。
- 在三个数据集上超越现有模型，参数量更少。
- 适合高分辨率遥感图像变化检测任务。
遥感变化检测旨在识别不同时相卫星图像间的语义差异。尽管深度学习显著推进了该领域，基于卷积神经网络（CNN）、Transformer和选择性状态空间模型（SSM）的方法仍难以精准划分变化区域。传统Transformer方法在处理超高清（VHR）遥感图像时面临二次计算复杂度问题，且在训练数据有限时表现不佳，导致未能充分利用VHR图像中的丰富空间信息。本文提出GRAD-Former，一种新型框架，在保持高效的同时增强上下文理解能力，并减少模型规模。该框架包含创新编码器、融合与解码模块。编码器中的自适应特征相关性与重构（AFRAR）模块通过选择性嵌入增强（SEA）和全局-局部特征重构（GLFR）组件实现全局-局部上下文感知。SEA利用门控机制，GLFR采用差分注意力，生成多个softmax堆栈以捕获重要特征并抑制无关特征。在三个挑战性数据集（LEVIR-CD、CDD、DSIFN-CD）上的多项实验表明，GRAD-Former在所有指标和数据集上均优于现有方法，且参数更少。本框架为遥感变化检测树立了新基准。代码将发布于：https://github.com/Ujjwal238/GRAD-Former
原文摘要 · Abstract (English)
Change detection (CD) in remote sensing aims to identify semantic differences between satellite images captured at different times. While deep learning has significantly advanced this field, existing approaches based on convolutional neural networks (CNNs), transformers and Selective State Space Models (SSMs) still struggle to precisely delineate change regions. In particular, traditional transformer-based methods suffer from quadratic computational complexity when applied to very high-resolution (VHR) satellite images and often perform poorly with limited training data, leading to under-utilization of the rich spatial information available in VHR imagery. We present GRAD-Former, a novel framework that enhances contextual understanding while maintaining efficiency through reduced model size. The proposed framework consists of a novel encoder with Adaptive Feature Relevance and Refinement (AFRAR) module, fusion and decoder blocks. AFRAR integrates global-local contextual awareness through two proposed components: the Selective Embedding Amplification (SEA) module and the Global-Local Feature Refinement (GLFR) module. SEA and GLFR leverage gating mechanisms and differential attention, respectively, which generates multiple softmax heaps to capture important features while minimizing the captured irreverent features. Multiple experiments across three challenging CD datasets (LEVIR-CD, CDD, DSIFN-CD) demonstrate GRAD-Former's superior performance compared to existing approaches. Notably, GRAD-Former outperforms the current state-of-the-art models across all the metrics and all the datasets while using fewer parameters. Our framework establishes a new benchmark for remote sensing change detection performance. Our code will be released at: https://github.com/Ujjwal238/GRAD-Former
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。