UniLoop: Unified Loop Closure Detection from Retrieval to Geometric Verification and Relative Pose Estimation
UniLoop: Unified Loop Closure Detection from Retrieval to Geometric Verification and Relative Pose Estimation
This research paper introduces UniLoop, a unified framework for loop closure detection that reuses hierarchical representations from a pretrained geometric foundation model across retrieval, verification, and pose estimation. By sharing features between a global image encoder, cross-view decoder, and pose-graph optimization, the method reduces redundant computation while improving robustness. Experiments on MSLS and Pitts30k benchmarks show UniLoop achieves state-of-the-art Recall@10 scores of 96.6 and 99.3 respectively, with verification branch performance reaching 95.63 AP and 96.29 ROC-AUC, outperforming existing modular baselines.
The paper introduces a unified framework that integrates retrieval, geometric verification, and relative pose estimation to improve loop-closure robustness and accuracy on standard benchmarks.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- paper: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
undant backbone computation relative to modular pipelines. UniLoop achieves the best Recall@10 on MSLS (96.6) and Pitts30k (99.3) among the compared methods, while its verification branch reaches 95.63 AP and 96.29 ROC-AUC, surpassing the existing geometric verification baselines.
Open source S4
Source：arXiv Robotics — research abstracts · arxiv.org