SuperPoint-SLAM3: Augmenting ORB-SLAM3 with Deep Features, Adaptive NMS, and Learning-Based Loop Closure

2025/06/16 by Shahram Najam Syed, Ishir Roongta, Syed, Shahram Najam +5 · 1 citation
Computer Science · #Advanced Image and Video Retrieval Techniques #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Handwritten Text Recognition Techniques #I.2.10 #I.2.9 #I.4.8 #Natural Language Processing Techniques #Robotics (cs.RO)

paper · pdf · doi:10.48550/arxiv.2506.13089

openalex publication_date 2025/06/16 · openalex created_date 2025/10/12 · openalex updated_date 2026/07/28

Abstract

Visual simultaneous localization and mapping (SLAM) must remain accurate under extreme viewpoint, scale and illumination variations. The widely adopted ORB-SLAM3 falters in these regimes because it relies on hand-crafted ORB keypoints. We introduce SuperPoint-SLAM3, a drop-in upgrade that (i) replaces ORB with the self-supervised SuperPoint detector--descriptor, (ii) enforces spatially uniform keypoints via adaptive non-maximal suppression (ANMS), and (iii) integrates a lightweight NetVLAD place-recognition head for learning-based loop closure. On the KITTI Odometry benchmark SuperPoint-SLAM3 reduces mean translational error from 4.15% to 0.34% and mean rotational error from 0.0027 deg/m to 0.0010 deg/m. On the EuRoC MAV dataset it roughly halves both errors across every sequence (e.g., V2_03: 1.58% -> 0.79%). These gains confirm that fusing modern deep features with a learned loop-closure module markedly improves ORB-SLAM3 accuracy while preserving its real-time operation. Implementation, pretrained weights and reproducibility scripts are available at https://github.com/shahram95/SuperPointSLAM3.

Citations

• Joint Learning of Depth, Pose, and Local Radiance Field for Large Scale Monocular 3D Reconstruction
• ExpReS-VLA: Specializing Vision-Language-Action Models Through Experience Replay and Retrieval
• Visuo-Acoustic Hand Pose and Contact Estimation
• TensoRF: Tensorial Radiance Fields
• Block-NeRF: Scalable Large Scene Neural View Synthesis
• Mega-NeRF: Scalable Construction of Large-Scale NeRFs for Virtual Fly-Throughs
• Plenoxels: Radiance Fields without Neural Networks
• Self-Calibrating Neural Radiance Fields
• Pixel-Perfect Structure-from-Motion with Featuremetric Refinement
• The Temporal Opportunist: Self-Supervised Multi-Frame Monocular Depth
• BARF: Bundle-Adjusting Neural Radiance Fields
• Learning by Aligning Videos in Time
• INeRF: Inverting Neural Radiance Fields for Pose Estimation
• NeRF++: Analyzing and Improving Neural Radiance Fields
• ORB-SLAM3: An Accurate Open-Source Library for Visual, Visual–Inertial, and Multimap SLAM
• SuperGlue: Learning Feature Matching with Graph Neural Networks
• Unsupervised Scale-consistent Depth and Ego-motion Learning from Monocular Video
• LF-Net: Learning Local Features from Images
• SuperPoint: Self-Supervised Interest Point Detection and Description
• Unsupervised Learning of Depth and Ego-Motion from Video
• NetVLAD: CNN architecture for weakly supervised place recognition

Cited by

• Joint Learning of Depth, Pose, and Local Radiance Field for Large Scale Monocular 3D Reconstruction

Related

• ORB-SLAM3: An Accurate Open-Source Library for Visual, Visual–Inertial, and Multimap SLAM
• DK-SLAM: Monocular Visual SLAM with Deep Keypoint Learning, Tracking and Loop-Closing
• MLP-SLAM: Multilayer Perceptron-Based Simultaneous Localization and Mapping
• IL-SLAM: Intelligent Line-assisted SLAM Based on Feature Awareness for Dynamic Environments
• SuperVINS: A Real-Time Visual-Inertial SLAM Framework for Challenging Imaging Conditions