XFeat Revisited: Reproducibility and Evaluation of a Lightweight Image Matcher
2026-08-10 • Computer Vision and Pattern Recognition
Computer Vision and Pattern RecognitionMachine Learning
AI summaryⓘ
The authors studied XFeat, a tool that finds matching points between images quickly on simple hardware. They redid the original design and tested both their version and the original to see how well it worked, finding similar or better performance on common image matching tasks. They examined parts of the design that were unclear in the original paper and found some benefits to a key feature, but less than initially claimed. They also tested the tool on different types of images and found it works well sometimes but struggles when the images are very different from what it was trained on. This suggests XFeat is good for typical uses but sensitive to changes in image types.
local feature extractorimage matchingreproducibility studyarchitectural ablationsMegaDepth-1500ScanNet-1500homography estimationvisual localizationzero-shot matchingcross-modal matching
Authors
Lazar Đoković, Aimee Lin
Abstract
We present a reproducibility study of XFeat, a lightweight local feature extractor and matcher designed to identify corresponding points across images efficiently on resource-constrained hardware. We re-implement the architecture based on the paper and supplementary material, re-evaluate the authors' released checkpoint alongside our re-implementation, and conduct additional architectural ablations to examine design choices that were not fully justified in the original work. This distinction between re-evaluation and reproduction is important, as the paper, supplement, and public code differ in several implementation details, including the backbone layout, fusion block, and training losses. Empirically, our reproduced models closely match and, in some cases, outperform the re-evaluated original checkpoint on MegaDepth-1500 and ScanNet-1500, supporting the main claim that XFeat provides a strong accuracy-efficiency trade-off for standard image-matching benchmarks. Our ablations provide a more nuanced view of two architectural arguments from the original paper. In particular, the parallel keypoint branch is important for semi-dense matching, but its benefit is less pronounced than originally claimed, while the evidence for the specific placement of the single skip-connection remains inconclusive. Finally, we reproduce the original downstream evaluations and find close agreement for homography estimation, while Aachen visual localization remains below the reported results, even for the released checkpoint, suggesting sensitivity to underspecified evaluation details. We then extend the analysis to zero-shot out-of-distribution and cross-modal matching across retinal, thermal-visible, and multimodal remote-sensing imagery, where XFeat remains effective in some settings but degrades sharply under severe modality shifts.