Key.Net: Keypoint Detection by Handcrafted and Learned CNN Filters

Axel Barroso-Laguna, Edgar Riba, Daniel Ponsa, Krystian Mikolajczyk
4/1/2019
cs.CV

Official code: axelBarroso/Key.Net (213 stars, last commit 2021-12-08).

Abstract

We introduce a novel approach for keypoint detection task that combines handcrafted and learned CNN filters within a shallow multi-scale architecture. Handcrafted filters provide anchor structures for learned filters, which localize, score and rank repeatable features. Scale-space representation is used within the network to extract keypoints at different levels. We design a loss function to detect robust features that exist across a range of scales and to maximize the repeatability score. Our Key.Net model is trained on data synthetically created from ImageNet and evaluated on HPatches benchmark. Results show that our approach outperforms state-of-the-art detectors in terms of repeatability, matching performance and complexity.

View on arXivView PDF

Code Implementations(1)

axelBarroso/Key.NetOfficial100%

[ICCV 2019] Key.Net: Keypoint Detection by Handcrafted and Learned CNN Filters

21337PythonAug 6, 20194 years agoNOASSERTION

Similar Papers

GMM-IKRS: Gaussian Mixture Models for Interpretable Keypoint Refinement and Scoring

83% match2024

DaD: Distilled Reinforcement Learning for Diverse Keypoint Detection

83% match1 repo1052025

SiLK: Simple Learned Keypoints

82% match1 repo6832023

Adversarial Focal Loss: Asking Your Discriminator for Hard Examples

82% match1 repo52022

DeDoDe: Detect, Don't Describe -- Describe, Don't Detect for Local Feature Matching

82% match2 repos4912023

ELF: Embedded Localisation of Features in pre-trained CNN

82% match1 repo52019

Cite this paper

@article{barrosolaguna2019keynet,
  title  = {Key.Net: Keypoint Detection by Handcrafted and Learned CNN Filters},
  author = {Axel Barroso-Laguna and Edgar Riba and Daniel Ponsa and Krystian Mikolajczyk},
  year   = {2019},
  eprint = {1904.00889},
  archivePrefix = {arXiv},
  url    = {https://arxiv.org/abs/1904.00889}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/1904.00889.svg)](https://codeofpaper.com/paper/1904.00889)

Discussion