# Knowledge Distillation for Feature Extraction in Underwater VSLAM

arXiv: https://arxiv.org/abs/2303.17981 · HTML: https://mlsanity.com/paper/2303.17981/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- Placed in Detection and dense prediction distillation, one of 12 directions in Knowledge distillation.
- Builds on 14 of its 32 references that are in the map.
- First posted on arXiv 31 Mar 2023.

## Placed in

- Knowledge distillation / [Detection and dense prediction distillation](https://mlsanity.com/distill/detection-and-dense-prediction-distillation/)

## Builds on (most cited first, 14 of 14)

- [SuperPoint: Self-Supervised Interest Point Detection and Description](https://mlsanity.com/paper/1712.07629/) (2017)
- [ORB-SLAM3: An Accurate Open-Source Library for Visual, Visual-Inertial and Multi-Map SLAM](https://mlsanity.com/paper/2007.11898/) (2020)
- [ORB-SLAM2: an Open-Source SLAM System for Monocular, Stereo and RGB-D Cameras](https://mlsanity.com/paper/1610.06475/) (2016)
- [Direct Sparse Odometry](https://mlsanity.com/paper/1607.02565/) (2016)
- [HPatches: A benchmark and evaluation of handcrafted and learned local descriptors](https://mlsanity.com/paper/1704.05939/) (2017)
- [Learning Deep Representations with Probabilistic Knowledge Transfer](https://mlsanity.com/paper/1803.10837/) (2018)
- [R2D2: Repeatable and Reliable Detector and Descriptor](https://mlsanity.com/paper/1906.06195/) (2019)
- [Cross Modal Distillation for Supervision Transfer](https://mlsanity.com/paper/1507.00448/) (2015)
- [ASLFeat: Learning Local Features of Accurate Shape and Localization](https://mlsanity.com/paper/2003.10071/) (2020)
- [WaterGAN: Unsupervised Generative Network to Enable Real-time Color Correction of Monocular Underwater Images](https://mlsanity.com/paper/1702.07392/) (2017)
- [DIML/CVL RGB-D Dataset: 2M RGB-D Images of Natural Indoor and Outdoor Scenes](https://mlsanity.com/paper/2110.11590/) (2021)
- [Online Invariance Selection for Local Feature Descriptors](https://mlsanity.com/paper/2007.08988/) (2020)
- [GCNv2: Efficient Correspondence Prediction for Real-Time SLAM](https://mlsanity.com/paper/1902.11046/) (2019)
- [Experimental Comparison of Open Source Visual-Inertial-Based State Estimation Algorithms in the Underwater Domain](https://mlsanity.com/paper/1904.02215/) (2019)

## Built on by (most cited first, 5 of 5)

- [Deep Learning Reforms Image Matching: A Survey and Outlook](https://mlsanity.com/paper/2506.04619/) (2025)
- [Knowledge Distillation for Underwater Feature Extraction and Matching via GAN-synthesized Images](https://mlsanity.com/paper/2504.08253/) (2025)
- [Optical Ocean Recipes: Creating Realistic Datasets to Facilitate Underwater Vision Research](https://mlsanity.com/paper/2509.20171/) (2025)
- [DUViN: Diffusion-Based Underwater Visual Navigation via Knowledge-Transferred Depth Features](https://mlsanity.com/paper/2509.02983/) (2025)
- [AquaFlow: A Monocular Gaussian Splatting SLAM for Underwater Streaming Reconstruction](https://mlsanity.com/paper/2608.22906/) (2026)

## Abstract (arXiv)

In recent years, learning-based feature detection and matching have outperformed manually-designed methods in in-air cases. However, it is challenging to learn the features in the underwater scenario due to the absence of annotated underwater datasets. This paper proposes a cross-modal knowledge distillation framework for training an underwater feature detection and matching network (UFEN). In particular, we use in-air RGBD data to generate synthetic underwater images based on a physical underwater imaging formation model and employ these as the medium to distil knowledge from a teacher model SuperPoint pretrained on in-air images. We embed UFEN into the ORB-SLAM3 framework to replace the ORB feature by introducing an additional binarization layer. To test the effectiveness of our method, we built a new underwater dataset with groundtruth measurements named EASI (https://github.com/Jinghe-mel/UFEN-SLAM), recorded in an indoor water tank for different turbidity levels. The experimental results on the existing dataset and our new dataset demonstrate the effectiveness of our method.