# GeoDesc: Learning Local Descriptors by Integrating Geometry Constraints

arXiv: https://arxiv.org/abs/1807.06294 · HTML: https://mlsanity.com/paper/1807.06294/

## Key facts

- Cited by 52 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 17 Jul 2018.

## Built on by (most cited first, 30 of 52)

- [DROID-SLAM: Deep Visual SLAM for Monocular, Stereo, and RGB-D Cameras](https://mlsanity.com/paper/2108.10869/) (2021)
- [ASpanFormer: Detector-Free Image Matching with Adaptive Span Transformer](https://mlsanity.com/paper/2208.14201/) (2022)
- [PointDSC: Robust Point Cloud Registration using Deep Spatial Consistency](https://mlsanity.com/paper/2103.05465/) (2021)
- [Learning to Match Features with Seeded Graph Matching Network](https://mlsanity.com/paper/2108.08771/) (2021)
- [FS6D: Few-Shot 6D Pose Estimation of Novel Objects](https://mlsanity.com/paper/2203.14628/) (2022)
- [FlowMap: High-Quality Camera Poses, Intrinsics, and Depth via Gradient Descent](https://mlsanity.com/paper/2404.15259/) (2024)
- [Iterative Deep Homography Estimation](https://mlsanity.com/paper/2203.15982/) (2022)
- [LocalTrans: A Multiscale Local Transformer Network for Cross-Resolution Homography Estimation](https://mlsanity.com/paper/2106.04067/) (2021)
- [Local Feature Matching Using Deep Learning: A Survey](https://mlsanity.com/paper/2401.17592/) (2024)
- [Learning to Estimate 6DoF Pose from Limited Data: A Few-Shot, Generalizable Approach using RGB Images](https://mlsanity.com/paper/2306.07598/) (2023)
- [Dense-SfM: Structure from Motion with Dense Consistent Matching](https://mlsanity.com/paper/2501.14277/) (2025)
- [SACReg: Scene-Agnostic Coordinate Regression for Visual Localization](https://mlsanity.com/paper/2307.11702/) (2023)
- [HumanGPS: Geodesic PreServing Feature for Dense Human Correspondences](https://mlsanity.com/paper/2103.15573/) (2021)
- [Enhancing Deformable Local Features by Jointly Learning to Detect and Describe Keypoints](https://mlsanity.com/paper/2304.00583/) (2023)
- [FeatureBooster: Boosting Feature Descriptors with a Lightweight Neural Network](https://mlsanity.com/paper/2211.15069/) (2022)
- [SIFT Matching by Context Exposed](https://mlsanity.com/paper/2106.09584/) (2021)
- [DenseGAP: Graph-Structured Dense Correspondence Learning with Anchor Points](https://mlsanity.com/paper/2112.06910/) (2021)
- [Improving Feature-based Visual Localization by Geometry-Aided Matching](https://mlsanity.com/paper/2211.08712/) (2022)
- [Extracting Deformation-Aware Local Features by Learning to Deform](https://mlsanity.com/paper/2111.10617/) (2021)
- [Learning-Based Dimensionality Reduction for Computing Compact and Effective Local Feature Descriptors](https://mlsanity.com/paper/2209.13586/) (2022)
- [3D reconstruction from spherical images: A review of techniques, applications, and prospects](https://mlsanity.com/paper/2302.04495/) (2023)
- [Deep Learning Reforms Image Matching: A Survey and Outlook](https://mlsanity.com/paper/2506.04619/) (2025)
- [SDGMNet: Statistic-based Dynamic Gradient Modulation for Local Descriptor Learning](https://mlsanity.com/paper/2106.04434/) (2021)
- [PanoVPR: Towards Unified Perspective-to-Equirectangular Visual Place Recognition via Sliding Windows across the Panoramic View](https://mlsanity.com/paper/2303.14095/) (2023)
- [Learning Affine Correspondences by Integrating Geometric Constraints](https://mlsanity.com/paper/2504.04834/) (2025)
- [Self-Supervised Endoscopic Image Key-Points Matching](https://mlsanity.com/paper/2208.11424/) (2022)
- [RenderNet: Visual Relocalization Using Virtual Viewpoints in Large-Scale Indoor Environments](https://mlsanity.com/paper/2207.12579/) (2022)
- [pySLAM: An Open-Source, Modular, and Extensible Framework for SLAM](https://mlsanity.com/paper/2502.11955/) (2025)
- [Aerial-Ground Image Feature Matching via 3D Gaussian Splatting-based Intermediate View Rendering](https://mlsanity.com/paper/2509.19898/) (2025)
- [Descriptor Distillation: a Teacher-Student-Regularized Framework for Learning Local Descriptors](https://mlsanity.com/paper/2209.11795/) (2022)

## Abstract (arXiv)

Learned local descriptors based on Convolutional Neural Networks (CNNs) have achieved significant improvements on patch-based benchmarks, whereas not having demonstrated strong generalization ability on recent benchmarks of image-based 3D reconstruction. In this paper, we mitigate this limitation by proposing a novel local descriptor learning approach that integrates geometry constraints from multi-view reconstructions, which benefits the learning process in terms of data generation, data sampling and loss computation. We refer to the proposed descriptor as GeoDesc, and demonstrate its superior performance on various large-scale benchmarks, and in particular show its great success on challenging reconstruction tasks. Moreover, we provide guidelines towards practical integration of learned descriptors in Structure-from-Motion (SfM) pipelines, showing the good trade-off that GeoDesc delivers to 3D reconstruction tasks between accuracy and efficiency.