面向泛化的多模态单应性估计
Towards Generalized Multimodal Homography Estimation
用单张图像合成带真值偏移的未对齐图像对来替代真实配对数据，可作为任何需要几何配对监督的视觉任务（单应性、光流、匹配）的通用数据增广手段；同时其颜色与特征解耦、跨尺度融合的网络设计可直接迁移到跨模态配准。
原文依据 we propose a training data synthesis method that generates unaligned image pairs with ground-truth offsets from a single input image
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论有监督与无监督的单应性估计方法依赖针对特定模态定制的图像对才能达到较高精度。然而，当应用于未见过的模态时，其性能会显著下降。为解决这一问题，我们提出一种训练数据合成方法，可从单张输入图像生成带有真值偏移的未对齐图像对。我们的方法在渲染图像对时赋予其多样的纹理与颜色，同时保留其结构信息。这些合成数据使训练出的模型在各类域上具有更强的鲁棒性和更好的泛化能力。此外，我们设计了一个网络，以充分利用跨尺度信息并将颜色信息从特征表示中解耦，从而提升估计精度。大量实验表明，我们的训练数据合成方法提升了泛化性能，结果也验证了所提出网络的有效性。
英文原文摘要 974 字符
Supervised and unsupervised homography estimation methods depend on image pairs tailored to specific modalities to achieve high accuracy. However, their performance deteriorates substantially when applied to unseen modalities. To address this issue, we propose a training data synthesis method that generates unaligned image pairs with ground-truth offsets from a single input image. Our approach renders the image pairs with diverse textures and colors while preserving their structural information. These synthetic data empower the trained model to achieve greater robustness and improved generalization across various domains. Additionally, we design a network to fully leverage cross-scale information and decouple color information from feature representations, thus improving estimation accuracy. Extensive experiments show that our training data synthesis method improves generalization performance. The results also confirm the effectiveness of the proposed network.