$\mathbb{X}$Resolution Correspondence Networks
This work addresses image matching for computer vision applications, offering an incremental improvement by optimizing resolution and efficiency.
The paper tackles the problem of establishing dense correspondences between images under challenging conditions by discovering that key computational modules in state-of-the-art networks can be removed with minimal accuracy loss, enabling faster training and larger batch sizes. This allows systematic investigation of resolution effects, leading to an optimal resolution that surpasses state-of-the-art methods on public benchmarks, particularly in lower error bands.
In this paper, we aim at establishing accurate dense correspondences between a pair of images with overlapping field of view under challenging illumination variation, viewpoint changes, and style differences. Through an extensive ablation study of the state-of-the-art correspondence networks, we surprisingly discovered that the widely adopted 4D correlation tensor and its related learning and processing modules could be de-parameterised and removed from training with merely a minor impact over the final matching accuracy. Disabling these computational expensive modules dramatically speeds up the training procedure and allows to use 4 times bigger batch size, which in turn compensates for the accuracy drop. Together with a multi-GPU inference stage, our method facilitates the systematic investigation of the relationship between matching accuracy and up-sampling resolution of the native testing images from 1280 to 4K. This leads to discovery of the existence of an optimal resolution $\mathbb{X}$ that produces accurate matching performance surpassing the state-of-the-art methods particularly over the lower error band on public benchmarks for the proposed network.