Research map / Vision-language and multimodal / Robot action and navigation
EgoLoc: A Generalizable Solution for Temporal Interaction Localization in Egocentric Videos
Junyi Ma, Erhang Zhang, Yin-Dong Zheng, Yuchen Xie, Yixuan Zhou, Hesheng Wang
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 self-adaptive 3D hand-dynamics sampling instead of random sampling (ablation) | EgoLoc · EgoPAT3D-DT | SR (gamma=5): 0.412 -> 0.819 | Improved |
| use 3D instead of 2D hand dynamics for sampling (ablation) | EgoLoc SASS · EgoPAT3D-DT | SR (gamma=5): 0.726 -> 0.819 | Improved |
| use boundary hand-region frames instead of the single anchor frame for attribute identification (ablation) | EgoLoc VLM discriminator · EgoPAT3D-DT | SR (gamma=5): 0.667 -> 0.819 | Improved |
| use boundary frames instead of the entire grid image for attribute identification (ablation) | EgoLoc VLM discriminator · EgoPAT3D-DT | SR (gamma=5): 0.750 -> 0.819 | Improved |
| add visual cue assessment feedback (ablation) | EgoLoc closed loop · DeskTIL | SR (gamma=5): 0.921 -> 0.976 | Improved |
| add in-context negative examples to the second round (ablation) | EgoLoc closed loop · DeskTIL | SR (gamma=5): 0.976 -> 1.000 | Improved |
| add closed-loop feedback, measured as MAE standard deviation across trials (ablation) | EgoLoc · EgoPAT3D-DT | MAE standard deviation: 0.327 -> 0.138 | Improved |
| increase N_adj from 2 to 4 (ablation) | EgoLoc · EgoPAT3D-DT | SR (gamma=5): 0.819 -> 0.705 | Hurt |
| increase N_adj from 2 to 4 (ablation) | EgoLoc · DeskTIL | SR (gamma=5): 1.000 -> 0.846 | Hurt |
| increase N_adj from 2 to 4 (ablation) | EgoLoc · ManiTIL | SR (gamma=5): 0.761 -> 0.651 | Hurt |
| use GPT-4o instead of GPT-4o mini as the VLM (ablation) | EgoLoc · EgoPAT3D-DT | SR (gamma=5): 0.678 -> 0.819 | Improved |
| use GPT-4o instead of open-source Janus-Pro-7B (ablation) | EgoLoc · EgoPAT3D-DT | SR (gamma=5): 0.469 -> 0.819 | Improved |
Closest papers by meaning
- MEgoHand: Multimodal Egocentric Hand-Object Interaction Motion Generation (NeurIPS 2025 · similarity 0.89)
- Do Egocentric Video-Language Models Truly Understand Hand-Object Interactions? (ICLR 2025 · similarity 0.88)
- EgoM2P: Egocentric Multimodal Multitask Pretraining (ICCV 2025 · similarity 0.88)
- EgoChoir: Capturing 3D Human-Object Interaction Regions from Egocentric Views (NeurIPS 2024 · similarity 0.88)
- HOT3D: Hand and Object Tracking in 3D from Egocentric Multi-View Videos (CVPR 2025 · similarity 0.87)
- Modeling Fine-Grained Hand-Object Dynamics for Egocentric Video Representation Learning (ICLR 2025 · similarity 0.87)
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.
