面向变化描述的干扰免疫表示学习与跨模态对比正则化
Distractors-Immune Representation Learning with Cross-modal Contrastive Regularization for Change Captioning
AI总结：
本文提出干扰免疫表示学习网络，通过自监督通道关联与去相关获得稳定图像表示，并设计跨模态对比正则化提升变化描述生成，在四个公共数据集上超越现有方法。
摘要：
变化描述旨在简洁地描述一对相似图像之间的语义变化，同时免受干扰（光照和视角变化）的影响。在这些干扰下，未变化的物体往往在位置和尺度上出现伪变化，且某些物体可能相互重叠，导致两幅图像之间的特征产生扰动和判别性退化。然而，大多数现有方法直接捕获它们之间的差异，这存在获得易出错差异特征的风险。本文提出了一种干扰免疫表示学习网络，该网络以自监督方式关联两个图像表示的对应通道并去相关不同通道，从而在干扰下获得一对稳定的图像表示。然后，模型可以更好地交互它们以捕获可靠的差异特征用于描述生成。为了基于最相关的差异特征生成单词，我们进一步设计了一种跨模态对比正则化，通过最大化注意力差异特征与生成单词之间的对比对齐来规范跨模态对齐。大量实验表明，我们的方法在四个公共数据集上优于最先进的方法。代码可在 https://github.com/tuyunbin/DIRL 获取。
英文摘要：
Change captioning aims to succinctly describe the semantic change between a pair of similar images, while being immune to distractors (illumination and viewpoint changes). Under these distractors, unchanged objects often appear pseudo changes about location and scale, and certain objects might overlap others, resulting in perturbational and discrimination-degraded features between two images. However, most existing methods directly capture the difference between them, which risk obtaining error-prone difference features. In this paper, we propose a distractors-immune representation learning network that correlates the corresponding channels of two image representations and decorrelates different ones in a self-supervised manner, thus attaining a pair of stable image representations under distractors. Then, the model can better interact them to capture the reliable difference features for caption generation. To yield words based on the most related difference features, we further design a cross-modal contrastive regularization, which regularizes the cross-modal alignment by maximizing the contrastive alignment between the attended difference features and generated words. Extensive experiments show that our method outperforms the state-of-the-art methods on four public datasets. The code is available at https://github.com/tuyunbin/DIRL.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
