Learning to Make Keypoints Sub-Pixel Accurate
This addresses the accuracy gap for computer vision applications like pose estimation, though it is incremental as it builds on existing detectors rather than introducing a new paradigm.
The paper tackles the problem of sub-pixel accuracy in 2D local feature detection, a key challenge in computer vision, by proposing a network that learns offset vectors to enhance any detector, resulting in consistent accuracy improvements across datasets with only about 7 ms added computational time.
This work addresses the challenge of sub-pixel accuracy in detecting 2D local features, a cornerstone problem in computer vision. Despite the advancements brought by neural network-based methods like SuperPoint and ALIKED, these modern approaches lag behind classical ones such as SIFT in keypoint localization accuracy due to their lack of sub-pixel precision. We propose a novel network that enhances any detector with sub-pixel precision by learning an offset vector for detected features, thereby eliminating the need for designing specialized sub-pixel accurate detectors. This optimization directly minimizes test-time evaluation metrics like relative pose error. Through extensive testing with both nearest neighbors matching and the recent LightGlue matcher across various real-world datasets, our method consistently outperforms existing methods in accuracy. Moreover, it adds only around 7 ms to the time of a particular detector. The code is available at https://github.com/KimSinjeong/keypt2subpx .