DAC: Detector-Agnostic Spatial Covariances for Deep Local Features
Not on the map. Why?
Abstract (arXiv)
Current deep visual local feature detectors do not model the spatial uncertainty of detected features, producing suboptimal results in downstream applications. In this work, we propose two post-hoc covariance estimates that can be plugged into any pretrained deep feature detector: a simple, isotropic covariance estimate that uses the predicted score at a given pixel location, and a full covariance estimate via the local structure tensor of the learned score maps. Both methods are easy to implement and can be applied to any deep feature detector. We show that these covariances are directly related to errors in feature matching, leading to improvements in downstream tasks, including solving the perspective-n-point problem and motion-only bundle adjustment. Code is available at https://github.com/javrtg/DAC
Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Builds on 28 of its 67 references that are in the map. First posted on arXiv 20 May 2023.
Builds on
28 of 67 references.
References · 28
- 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,364
- 2017What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision?cited by 2,352
- 2017SuperPoint: Self-Supervised Interest Point Detection and Descriptioncited by 1,320
- 2021LoFTR: Detector-Free Local Feature Matching with Transformerscited by 984
- 2020ORB-SLAM3: An Accurate Open-Source Library for Visual, Visual-Inertial and Multi-Map SLAMcited by 878
- 2016ORB-SLAM2: an Open-Source SLAM System for Monocular, Stereo and RGB-D Camerascited by 806
- 2018From Coarse to Fine: Robust Hierarchical Localization at Large Scalecited by 559
- 2019D2-Net: A Trainable CNN for Joint Detection and Description of Local Featurescited by 326
- 2020DISK: Learning local features with policy gradientcited by 312
- 2016LIFT: Learned Invariant Feature Transformcited by 306
- 2017HPatches: A benchmark and evaluation of handcrafted and learned local descriptorscited by 264
- 2021COTR: Correspondence Transformer for Matching Across Imagescited by 183
16 more
- 2019Evaluating Scalable Bayesian Deep Learning Methods for Robust Computer Visioncited by 182
- 2022OnePose: One-Shot Object Pose Estimation without CAD Modelscited by 157
- 2020ASLFeat: Learning Local Features of Accurate Shape and Localizationcited by 130
- 2022QuadTree Attention for Vision Transformerscited by 129
- 2014TILDE: A Temporally Invariant Learned DEtectorcited by 55
- 2022Is Geometry Enough for Matching in Visual Localization?cited by 48
- 2020Multi-View Optimization of Local Feature Geometrycited by 24
- 2021Uncertainty-Aware Camera Pose Estimation from Points and Linescited by 17
- 2018Calibration Wizard: A Guidance System for Camera Calibration Based on Modelling Geometric and Corner Uncertaintycited by 15
- 2023Learning Correspondence Uncertainty via Differentiable Nonlinear Least Squarescited by 14
- 2016MLPnP - A Real-Time Maximum Likelihood Solution to the Perspective-n-Point Problemcited by 13
- 2022AirVO: An Illumination-Robust Point-Line Visual Odometrycited by 9
- 2022The Probabilistic Normal Epipolar Constraint for Frame-To-Frame Rotation Optimization under Uncertain Feature Positionscited by 8
- 2020Geometric Interpretations of the Normalized Epipolar Errorcited by 6
- 2022SuperPoint features in endoscopycited by 6
- 2022A Model for Multi-View Residual Covariances based on Perspective Deformationcited by 3
Built on by
Citing papers per year: 2024: 1; 2025: 0; 2026: 2.
3 papers.