# UnsuperPoint: End-to-end Unsupervised Interest Point Detector and Descriptor

arXiv: https://arxiv.org/abs/1907.04011 · HTML: https://mlsanity.com/paper/1907.04011/

## Key facts

- Cited by 22 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 9 Jul 2019.

## Built on by (most cited first, 22 of 22)

- [Pixel-Perfect Structure-from-Motion with Featuremetric Refinement](https://mlsanity.com/paper/2108.08291/) (2021)
- [SiLK -- Simple Learned Keypoints](https://mlsanity.com/paper/2304.06194/) (2023)
- [SOLD2: Self-supervised Occlusion-aware Line Description and Detection](https://mlsanity.com/paper/2104.03362/) (2021)
- [P2-Net: Joint Description and Detection of Local Features for Pixel and Point Matching](https://mlsanity.com/paper/2103.01055/) (2021)
- [Decoupling Makes Weakly Supervised Local Feature Better](https://mlsanity.com/paper/2201.02861/) (2022)
- [Matching 2D Images in 3D: Metric Relative Pose from Metric Correspondences](https://mlsanity.com/paper/2404.06337/) (2024)
- [EventPoint: Self-Supervised Interest Point Detection and Description for Event-based Camera](https://mlsanity.com/paper/2109.00210/) (2021)
- [Reverse Knowledge Distillation: Training a Large Model using a Small One for Retinal Image Matching on Limited Data](https://mlsanity.com/paper/2307.10698/) (2023)
- [ZippyPoint: Fast Interest Point Detection, Description, and Matching through Mixed Precision Discretization](https://mlsanity.com/paper/2203.03610/) (2022)
- [SCIPaD: Incorporating Spatial Clues into Unsupervised Pose-Depth Joint Learning](https://mlsanity.com/paper/2407.05283/) (2024)
- [Deep Learning Reforms Image Matching: A Survey and Outlook](https://mlsanity.com/paper/2506.04619/) (2025)
- [Monocular visual simultaneous localization and mapping: (r)evolution from geometry to deep learning-based pipelines](https://mlsanity.com/paper/2503.02955/) (2025)
- [3D3L: Deep Learned 3D Keypoint Detection and Description for LiDARs](https://mlsanity.com/paper/2103.13808/) (2021)
- [Digging Into Self-Supervised Learning of Feature Descriptors](https://mlsanity.com/paper/2110.04773/) (2021)
- [RaCo: Ranking and Covariance for Practical Learned Keypoints](https://mlsanity.com/paper/2602.15755/) (2026)
- [EdgePoint2: Compact Descriptors for Superior Efficiency and Accuracy](https://mlsanity.com/paper/2504.17280/) (2025)
- [Soft Expectation and Deep Maximization for Image Feature Detection](https://mlsanity.com/paper/2104.10291/) (2021)
- [Gleo-Det: Deep Convolution Feature-Guided Detector with Local Entropy Optimization for Salient Points](https://mlsanity.com/paper/2204.12884/) (2022)
- [DDM-NET: End-to-end learning of keypoint feature Detection, Description and Matching for 3D localization](https://mlsanity.com/paper/2212.04575/) (2022)
- [Utilizing Radiomic Feature Analysis For Automated MRI Keypoint Detection: Enhancing Graph Applications](https://mlsanity.com/paper/2311.18281/) (2023)
- [Learning Visual Information Utility with PIXER](https://mlsanity.com/paper/2409.13151/) (2024)
- [XRefine: Attention-Guided Keypoint Match Refinement](https://mlsanity.com/paper/2601.12530/) (2026)

## Abstract (arXiv)

It is hard to create consistent ground truth data for interest points in natural images, since interest points are hard to define clearly and consistently for a human annotator. This makes interest point detectors non-trivial to build. In this work, we introduce an unsupervised deep learning-based interest point detector and descriptor. Using a self-supervised approach, we utilize a siamese network and a novel loss function that enables interest point scores and positions to be learned automatically. The resulting interest point detector and descriptor is UnsuperPoint. We use regression of point positions to 1) make UnsuperPoint end-to-end trainable and 2) to incorporate non-maximum suppression in the model. Unlike most trainable detectors, it requires no generation of pseudo ground truth points, no structure-from-motion-generated representations and the model is learned from only one round of training. Furthermore, we introduce a novel loss function to regularize network predictions to be uniformly distributed. UnsuperPoint runs in real-time with 323 frames per second (fps) at a resolution of 224×320 and 90 fps at 480×640. It is comparable or better than state-of-the-art performance when measured for speed, repeatability, localization, matching score and homography estimation on the HPatch dataset.