awesome local feature in 2023s

特征可视化与对比

mage Matching WebUI：find matches between 2 images https://github.com/Vincentqyw/image-matching-webui

大模型

SEGMENT ANYTHING MODEL IS A GOOD TEACHER FOR LOCAL FEATURE LEARNING
https://github.com/vignywang/SAMFeat

特征点

ALIKE: Accurate and Lightweight Keypoint Detection and Descriptor Extraction
https://github.com/Shiaoming/ALIKE

r2d2
https://github.com/naver/r2d2

D2-Net: A Trainable CNN for Joint Detection and Description of Local Features
https://github.com/mihaidusmanu/d2-net

aslfeat
https://github.com/lzx551402/ASLFeat

disk
https://github.com/cvlab-epfl/disk

Deformation-Aware Local Features (DALF)
verlab.dcc.ufmg.br/descriptors/dalf_cvpr23
https://github.com/verlab/DALF_CVPR_2023

MTLDesc: Looking Wider to Describe Better
https://github.com/vignywang/MTLDesc

CVPR 2022 "Self-Supervised Equivariant Learning for Oriented Keypoint Detection
https://github.com/bluedream1121/REKD

Learning Rotation-Equivariant Features for Visual Correspondence (CVPR 2023)
https://github.com/bluedream1121/RELF

FisheyeSuperPoint: Keypoint Detection and Description Network for Fisheye Images

NinjaDesc: Content-Concealing Visual Descriptors via Adversarial Learning

superpoint (训练复现) https://github.com/eric-yyjau/pytorch-superpoint https://github.com/rpautrat/SuperPoint

superpoint+语义head
https://github.com/Gabriel-SGama/Semantic-SuperPoint

Shared Coupling-bridge for Weakly Supervised Local Feature Learning https://github.com/sunjiayuanro/SCFeat

SiLK - Simple Learned Keypoints https://github.com/facebookresearch/silk

ZippyPoint: Fast Interest Point Detection, Description, and Matching through Mixed Precision Discretization https://github.com/menelaoskanakis/ZippyPoint

Detect, Don't Describe --- Describe, Don't Detect for Local Feature Matching https://github.com/Parskatt/DeDoDe

特征点降维：

Learning-Based Dimensionality Reduction for Computing Compact and Effective Local Feature Descriptors (ICRA2023) https://github.com/PRBonn/descriptor-dr

特征点协方差

DAC: Detector-Agnostic Spatial Covariances for Deep Local Features https://github.com/javrtg/DAC

结合语义 SFD2: Semantic-guided Feature Detection and Description

RoRD: Rotation-Robust Descriptors and Orthographic Views for Local Feature Matching

UR2KiD: Unifying Retrieval, Keypoint Detection, and Keypoint Description without Local Correspondence Supervision

车道线描述子

Eigenlanes: Data-Driven Lane Descriptors for Structurally Diverse Lanes

Eigencontours: Novel Contour Descriptors Based on Low-Rank Approximation

线特征

室内线特征点benchmark

https://github.com/prime-slam/evolin

L2D2: Learnable Line Detector and Descriptor https://github.com/hichem-abdellali/L2D2

SOLD² - Self-supervised Occlusion-aware Line Description and Detection https://github.com/cvg/SOLD2

DeepLSD: Line Segment Detection and Refinement with Deep Image Gradients, accepted at CVPR 2023 https://github.com/cvg/DeepLSD

Line as a Visual Sentence: Context-Aware Line Descriptor for Visual Localization https://github.com/yosungho/LineTR

L2D2: Learnable Line Detector and Descriptor https://github.com/hichem-abdellali/L2D2

SuperLine3D: Self-supervised Line Segmentation and Description for LiDAR Point Cloud https://github.com/zxrzju/SuperLine3D (lidar)

wireframe

LETR: Line Segment Detection Using Transformers without Edges https://github.com/mlpc-ucsd/LETR

PPGNet: Learning Point-Pair Graph for Line Segment Detection https://github.com/svip-lab/PPGNet

消失点

https://github.com/danini/progressive-x
https://github.com/B1ueber2y/JLinkage

线特征地图

3D Line Mapping Revisited, CVPR 2023 https://github.com/cvg/limap

dense匹配

LIFE: Lighting Invariant Flow Estimation https://github.com/wx-b/LIFE

DKM: Dense Kernelized Feature Matching for Geometry Estimation https://github.com/Parskatt/dkm

PMatch: Paired Masked Image Modeling for Dense Geometric Matching https://github.com/ShngJZ/PMatch

RoMa:Revisiting Robust Losses for Dense Feature Matching https://github.com/Parskatt/RoMa

PDC-Net+: Enhanced Probabilistic Dense Correspondence Network https://github.com/PruneTruong/DenseMatching

DiffMatch: Diffusion Model for Dense Matching https://ku-cvlab.github.io/DiffMatch/ https://github.com/KU-CVLAB/DiffMatch

特征点与GCN

Learning Feature Matching via Matchable Keypoint-Assisted Graph Neural Network

AirCode: A Robust Object Encoding Method. IEEE Robotics and Automation Letters (2022) https://github.com/wang-chen/AirCode

Graph-Based Contrastive Learning for Description and Detection of Local Features https://github.com/RealZihaoWang/GCLFeat

基于回归

patch2pix https://github.com/GrumpyZhou/patch2pix

caps：Learning Feature Descriptors using Camera Pose Supervision https://github.com/qianqianwang68/caps

posfeat：Decoupling Makes Weakly Supervised Local Feature Better https://github.com/The-Learning-And-Vision-Atelier-LAVA/PoSFeat

PATS: Patch Area Transportation with Subdivision for Local Feature Matching https://zju3dv.github.io/pats/

特征匹配与位姿估计

IMP: Iterative Matching and Pose Estimation with Adaptive Pooling https://github.com/feixue94/imp-release

特征匹配

image-matching-toolbox

https://github.com/GrumpyZhou/image-matching-toolbox

重复纹理的匹配

Doppelgangers: Learning to Disambiguate Images of Similar Structures
doppelgangers-3d.github.io

detection-based

ResMatch: Residual Attention Learning for Local Feature Matching
https://github.com/ACuOoOoO/ResMatch

Scene-Aware Feature Matching

FeatureBooster: Boosting Feature Descriptors with a Lightweight Neural Network (CVPR 2023) https://github.com/SJTU-ViSYS/FeatureBooster

superglue https://github.com/magicleap/SuperGluePretrainedNetwork

SGMNET ICCV'21 "Learning to Match Features with Seeded Graph Matching Network https://github.com/vdvchen/SGMNet

FGCNet: Fast Graph Convolution for Matching Features

openglue https://github.com/ucuapps/OpenGlue

点云superglue Keypoint Matching for Point Cloud Registration using Multiplex Dynamic Graph Attention Networks https://github.com/nubot-nudt/MDGAT-matcher

(点云superglue)Geometric Transformer for Fast and Robust Point Cloud Registration https://github.com/minhaolee/GeoTransformer

点线段匹配统一架构 gluestick https://github.com/cvg/GlueStick

AffineGlue: Joint Matching and Robust Estimation

2023 LightGlue: Local Feature Matching at Light Speed github.com/cvg/LightGlue

detection-free

loftr https://github.com/zju3dv/LoFTR

Local Feature Matching with Transformers for low-end devices. LoFTR method adaptation approach. https://github.com/Kolkir/Coarse_LoFTR_TRT

se-loftr：A case for using rotation invariant features in state of the art feature matchers https://github.com/georg-bn/se2-loftr

MatchFormer: Interleaving Attention in Transformers for Feature Matching https://github.com/jamycheung/MatchFormer

deepmatcher https://github.com/XT-1997/DeepMatcher

OETR: Guide Local Feature Matching by Overlap Estimation https://github.com/TencentYoutuResearch/ImageMatching-OETR

OAMatcher: An Overlapping Areas-based Network for Accurate Local Feature Matching https://github.com/DK-HU/OAMatcher

AdaMatcher: Adaptive Assignment for Geometry Aware Local Feature Matching https://github.com/TencentYoutuResearch/AdaMatcher

ASTR: Adaptive Spot-Guided Transformer for Consistent Local Feature Matching(CVPR 2023) https://astr2023.github.io/

ECCV'22 “ASpanFormer: Detector-Free Image Matching with Adaptive Span Transformer” https://github.com/apple/ml-aspanformer

ParaFormer: Parallel Attention Transformer for Efficient Feature Matching

AAAI23 TopicFM: Robust and Interpretable Topic-Assisted Feature Matching https://github.com/TruongKhang/TopicFM

AAAI23 "ConvMatch: Rethinking Network Design for Two-View Correspondence Learning" https://github.com/SuhZhang/ConvMatch

Improving Transformer-based Image Matching by Cascaded Capturing Spatially Informative Keypoints
https://github.com/ewrfcas/CasMTR

极线优化

E3CM: Epipolar-Constrained Cascade Correspondence Matching

2d-3dmatch:

https://github.com/GabbySuwichaya/2D-3D-Matching

Efficient & Effective Prioritized Matching for Large-Scale Image-Based Localization

Recalling Direct 2D-3D Matches for Large-Scale Visual Localization

Efficient Global 2D-3D Matching for Camera Localization in a Large-Scale 3D Map

Efficient 2D-3D Matching for Multi-Camera Visual Localization

2D3D-MatchNet: Learning to Match Keypoints Across 2D Image and 3D Point Cloud

BPNPnet：Solving the Blind Perspective-n-Point Problem End-To-End With Robust Differentiable Geometric Optimization

P2-Net: Joint Description and Detection of Local Features for Pixel and Point Matching https://github.com/BingCS/P2-Net

GoMatch -- "Is Geometry Enough for Matching in Visual Localization" [ECCV2022] https://github.com/dvl-tum/gomatch

(GAM,BMnet)Improving Feature-based Visual Localization by Geometry-Aided Matching https://github.com/openxrlab/xrlocalization

2D3D-MATR: 2D-3D Matching Transformer for Detection-free Registration between Images and Point Clouds https://github.com/minhaolee/2D3DMATR

外点过滤

BANSAC: A dynamic BAyesian Network for adaptive SAmple Consensus
https://github.com/merlresearch/bansac

其他

多视图几何稳定性研究
On the Instability of Relative Pose Estimation and RANSAC’s Role