Research map / Point clouds / Point cloud classification
Egocentric Action-aware Inertial Localization in Point Clouds with Vision-Language Guidance
Mingfang Zhang, Ryo Yonetani, Yifei Huang, Liangyang Ouyang, Ruicong Liu, Yoichi Sato
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 |
|---|---|---|---|
| add vision-language guidance in Stage 1 alignment (ablation) | EAIL · EgoExo4D unseen rooms | localization success @0.2m: 20.41 -> 26.86 | Improved |
| add vision-language guidance in Stage 1 alignment (ablation) | EAIL · EgoExo4D unseen rooms | action top-1: 9.53 -> 15.03 | Improved |
| add vision-language guidance in Stage 1 alignment (ablation) | EAIL · EgoExo4D seen rooms | localization success @0.2m: 39.75 -> 43.86 | Improved |
| add action classification loss in Stage 2 (ablation) | EAIL · EgoExo4D seen rooms | localization success @0.2m: 41.92 -> 43.86 | Improved |
| add action classification loss in Stage 2 (ablation) | EAIL · EgoExo4D unseen rooms | localization success @0.2m: 25.37 -> 26.86 | Improved |
| add spatial reasoning module (ablation) | EAIL · EgoExo4D seen rooms | localization success @0.2m: 38.68 -> 43.86 | Improved |
| add temporal reasoning module (ablation) | EAIL · EgoExo4D seen rooms | localization success @0.2m: 41.44 -> 43.86 | Improved |
| add location attention over point cloud features for action recognition (ablation) | EAIL · EgoExo4D seen rooms | action top-1: 17.88 -> 21.48 | Improved |
| naively add global point cloud features without location attention (ablation) | EAIL · EgoExo4D seen rooms | action top-1: 18.81 -> 17.88 | Hurt |
| add Stage 2 spatiotemporal reasoning on top of Stage 1 retrieval (ablation) | EAIL · EgoExo4D seen rooms | localization success @0.2m: 18.29 -> 43.86 | Improved |
| add residual connections on IMU path (ablation) | EAIL · EgoExo4D seen rooms | localization success @0.2m: 36.29 -> 43.86 | Improved |
| add residual connections on point cloud path (ablation) | EAIL · EgoExo4D seen rooms | localization success @0.2m: 39.14 -> 43.86 | Improved |
Closest papers by meaning
- Spherical World-Locking for Audio-Visual Localization in Egocentric Videos (ECCV 2024 · similarity 0.86)
- iMoT: Inertial Motion Transformer for Inertial Navigation (AAAI 2025 · similarity 0.84)
- EMHI: A Multimodal Egocentric Human Motion Dataset with HMD and Body-Worn IMUs (AAAI 2025 · similarity 0.84)
- Ego4o: Egocentric Human Motion Capture and Understanding from Multi-Modal Input (CVPR 2025 · similarity 0.84)
- EgoThinker: Unveiling Egocentric Reasoning with Spatio-Temporal CoT (NeurIPS 2025 · similarity 0.83)
- EgoLM: Multi-Modal Language Model of Egocentric Motions (CVPR 2025 · similarity 0.83)
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.
