arXiv reaDer
InterNet: Unsupervised Cross-modal Homography Estimation Based on Interleaved Modality Transfer and Self-supervised Homography Prediction
We propose a novel unsupervised cross-modal homography estimation framework, based on interleaved modality transfer and self-supervised homography prediction, named InterNet. InterNet integrates modality transfer and self-supervised homography estimation, introducing an innovative interleaved optimization framework to alternately promote both components. The modality transfer gradually narrows the modality gaps, facilitating the self-supervised homography estimation to fully leverage the synthetic intra-modal data. The self-supervised homography estimation progressively achieves reliable predictions, thereby providing robust cross-modal supervision for the modality transfer. To further boost the estimation accuracy, we also formulate a fine-grained homography feature loss to improve the connection between two components. Furthermore, we employ a simple yet effective distillation training technique to reduce model parameters and improve cross-domain generalization ability while maintaining comparable performance. Experiments reveal that InterNet achieves the state-of-the-art (SOTA) performance among unsupervised methods, and even outperforms many supervised methods such as MHN and LocalTrans.
updated: Fri Sep 27 2024 02:35:47 GMT+0000 (UTC)
published: Thu Sep 26 2024 16:04:31 GMT+0000 (UTC)
参考文献 (このサイトで利用可能なもの) / References (only if available on this site)
被参照文献 (このサイトで利用可能なものを新しい順に) / Citations (only if available on this site, in order of most recent)
Amazon.co.jpアソシエイト