Research map / Label-efficient and robust learning / Test-time adaptation
UL-VIO: Ultra-lightweight Visual-Inertial Odometry with Noise Robust Test-time Adaptation
Jinho Park, Se Young Chun, Mingoo Seok
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| compress NASVIO: pool after last conv in visual encoder, shrink inertial channels, replace LSTM with FC | NASVIO · KITTI Seq 05/07/10 | model size vs relative translation error: 36.45x smaller, t_rel +1.11%, r_rel +1.05 deg | Improved |
| replace LSTM with fully connected layers in fused decoder (ablation) | UL-VIO · KITTI | pose error: 161x smaller D_fused, 1% error increase | No clear effect |
| add inertial-pseudo-label BN-only test-time adaptation | UL-VIO · KITTI blur noise | pose-wise t_rmse [m]: 0.261 -> 0.230 | Improved |
| add TTA under multiplicative noise | UL-VIO · KITTI multiplicative noise | pose-wise t_rmse [m]: 0.154 -> 0.156 | No clear effect |
| use TTA instead of fine-tuning on the matching noise | UL-VIO · KITTI rain noise | pose-wise t_rmse [m]: 0.095 (fine-tuned) vs 0.143 (TTA) | Hurt |
| add online TTA with gating under single blur shift | UL-VIO · KITTI Seq 07 blur | pose-wise t_rmse [m]: 0.133 -> 0.093 | Improved |
| add continual TTA with BN dictionary and domain matching | UL-VIO · KITTI continual shifts | pose-wise t_rmse [m]: 0.137 -> 0.113 | Improved |
| add continual TTA on EuRoC | UL-VIO · EuRoC MH_4_difficult | pose-wise t_rmse [m]: 0.0262 -> 0.0254 | Improved |
| add continual TTA on real-world Marulan shifts | UL-VIO · Marulan | pose-wise t_rmse [m]: 0.237 -> 0.232 | Improved |
Closest papers by meaning
- Adaptive VIO: Deep Visual-Inertial Odometry with Online Continual Learning (CVPR 2024 · similarity 0.91)
- Efficient Deep Visual and Inertial Odometry with Adaptive Visual Modality Selection (ECCV 2022 · similarity 0.89)
- Tartan IMU: A Light Foundation Model for Inertial Positioning in Robotics (CVPR 2025 · similarity 0.86)
- Leveraging Consistent Spatio-Temporal Correspondence for Robust Visual Odometry (AAAI 2025 · similarity 0.85)
- EqNIO: Subequivariant Neural Inertial Odometry (ICLR 2025 · similarity 0.85)
- LEAP-VO: Long-term Effective Any Point Tracking for Visual Odometry (CVPR 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
