GoodPoint: unsupervised learning of keypoint detection and description

TL;DR

GoodPoint presents an unsupervised method for keypoint detection and description that converges quickly and performs well across diverse datasets, including natural and medical images, by leveraging image transformations during training.

Contribution

It introduces a novel unsupervised training approach for keypoint detection and description using homographic transformations, with a model architecture based on SuperPoint for easy comparison.

Findings

Achieves fast convergence in training.

Performs well on natural images from HPatches.

Outperforms on retina images with fewer features.

Abstract

This paper introduces a new algorithm for unsupervised learning of keypoint detectors and descriptors, which demonstrates fast convergence and good performance across different datasets. The training procedure uses homographic transformation of images. The proposed model learns to detect points and generate descriptors on pairs of transformed images, which are easy for it to distinguish and repeatedly detect. The trained model follows SuperPoint architecture for ease of comparison, and demonstrates similar performance on natural images from HPatches dataset, and better performance on retina images from Fundus Image Registration Dataset, which contain low number of corner-like features. For HPatches and other datasets, coverage was also computed to provide better estimation of model quality.