Saved in:
Bibliographic Details
Main Authors: Pourhadi, Abtin, Swoboda, Paul
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2503.17715
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913089720418304
author Pourhadi, Abtin
Swoboda, Paul
author_facet Pourhadi, Abtin
Swoboda, Paul
contents We introduce the Normalized Matching Transformer (NMT), a deep learning approach for efficient and accurate sparse semantic keypoint matching between image pairs. NMT consists of a strong visual backbone, geometric feature refinement via SplineCNN, followed by a normalized Transformer for computing matching features. Central to NMT is our hyperspherical normalization strategy: we enforce unit-norm embeddings at every Transformer layer and train with a combined contrastive InfoNCE and hyperspherical uniformity loss to yield more discriminative keypoint representations. This novel architecture/loss combination encourages close alignment of matching image features and large distances between non-matching ones not only at the output level, but for each layer. Despite its architectural simplicity, NMT sets a new state-of-the-art performance on PascalVOC and SPair-71k, outperforming BBGM, ASAR, COMMON and GMTR by 5.1% and 2.2%, respectively, while converging in at least 1.7x fewer epochs compared to other state-of-the-art baselines. These results underscore the power of combining pervasive normalization with hyperspherical learning for matching tasks.
format Preprint
id arxiv_https___arxiv_org_abs_2503_17715
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Normalized Matching Transformer
Pourhadi, Abtin
Swoboda, Paul
Computer Vision and Pattern Recognition
Machine Learning
We introduce the Normalized Matching Transformer (NMT), a deep learning approach for efficient and accurate sparse semantic keypoint matching between image pairs. NMT consists of a strong visual backbone, geometric feature refinement via SplineCNN, followed by a normalized Transformer for computing matching features. Central to NMT is our hyperspherical normalization strategy: we enforce unit-norm embeddings at every Transformer layer and train with a combined contrastive InfoNCE and hyperspherical uniformity loss to yield more discriminative keypoint representations. This novel architecture/loss combination encourages close alignment of matching image features and large distances between non-matching ones not only at the output level, but for each layer. Despite its architectural simplicity, NMT sets a new state-of-the-art performance on PascalVOC and SPair-71k, outperforming BBGM, ASAR, COMMON and GMTR by 5.1% and 2.2%, respectively, while converging in at least 1.7x fewer epochs compared to other state-of-the-art baselines. These results underscore the power of combining pervasive normalization with hyperspherical learning for matching tasks.
title Normalized Matching Transformer
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2503.17715