HIMEC：方向变化表示与固定接口解码用于遥感图像变化描述
原标题：HIMEC: Directional Change Representation and Fixed-Interface Decoding for Remote Sensing Image Change Captioning
AI 摘要
HIMEC 框架结合方向变化表示（DCR）与固定接口解码，用于遥感图像变化描述（RSICC）。DCR 将符号差异分离为出现、消失和共享上下文流，并通过学习查询编码器生成视觉条件变化查询令牌。在 LEVIR-CC 和 SECOND-CC 基准上，HIMEC 相比直接融合特征记忆显著提升 CIDEr 分数，并诊断了级联模型中的训练-推理不一致问题。源代码将在发表后公开。
正文节选
HIMEC: Directional Change Representation and Fixed-Interface Decoding for Remote Sensing Image Change Captioning Abstract Remote sensing image change captioning (RSICC) converts bitemporal imagery into a sentence describing semantic changes. Most RSICC methods condition caption decoders directly on fused visual features, leaving intermediate change structure and decoder-interface consistency less studied. We present HIMEC, a framework that combines Directional Change Representation (DCR) with fi