Semi-supervised Deep Large-baseline Homography Estimation with Progressive Equivalence Constraint
This work addresses a specific challenge in computer vision for applications like image stitching and registration, but it is incremental as it builds on existing homography estimation methods.
The paper tackles the problem of erroneous homography estimation in large-baseline scenes by proposing a progressive estimation strategy and a semi-supervised loss, achieving state-of-the-art performance in large-baseline scenes while maintaining competitive results in small-baseline scenes.
Homography estimation is erroneous in the case of large-baseline due to the low image overlay and limited receptive field. To address it, we propose a progressive estimation strategy by converting large-baseline homography into multiple intermediate ones, cumulatively multiplying these intermediate items can reconstruct the initial homography. Meanwhile, a semi-supervised homography identity loss, which consists of two components: a supervised objective and an unsupervised objective, is introduced. The first supervised loss is acting to optimize intermediate homographies, while the second unsupervised one helps to estimate a large-baseline homography without photometric losses. To validate our method, we propose a large-scale dataset that covers regular and challenging scenes. Experiments show that our method achieves state-of-the-art performance in large-baseline scenes while keeping competitive performance in small-baseline scenes. Code and dataset are available at https://github.com/megvii-research/LBHomo.