Towards Generalized Multimodal Homography Estimation
This work addresses the problem of homography estimation for computer vision applications, particularly for scenarios where the model needs to generalize across different modalities.
The authors tackled the problem of homography estimation across different modalities, achieving improved generalization and robustness through a novel training data synthesis method and network design. The approach enables the model to perform well across various domains.
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.