Research map / Video / Video segmentation and tracking
EgoTactile: Learning Grasp Pressure for Everyday Objects from Egocentric Video
Yuan Zeng, Yujia Shi, Tiao Tan, Xingting Li, Yaqi Qin, Zongqing Lu, Wenming Yang, Jing-Hao Xue, Qingmin Liao
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 bin classification instead of raw pressure regression (ablation) | EgoPressureFormer · Gloved Object-Held-Out | Contact IoU: -14.4% | Improved |
| add bin classification instead of raw pressure regression (ablation) | EgoPressureFormer · Gloved Object-Held-Out | MAE: 6.2 -> 12.8 | Improved |
| add frame-level contact head (ablation) | EgoPressureFormer · Gloved Object-Held-Out | Contact IoU: 29.5 -> 36.8 | Improved |
| add mask conditioning (ablation) | EgoPressureDiff · Gloved Object-Held-Out | Contact IoU: 51.5 -> 56.3 | Improved |
| add hand pressure prototype (ablation) | EgoPressureDiff · Gloved Object-Held-Out | CoP error: 5.5 -> 3.1 | Improved |
| add text prompts (ablation) | EgoPressureDiff · Gloved Object-Held-Out | Volumetric IoU: 30.5 -> 38.9 | Improved |
| add text prompts (ablation) | EgoPressureDiff · Gloved Object-Held-Out | MAE: 5.1 -> 3.4 | Improved |
| add PIFR layer instead of naive fusion (ablation) | EgoPressureDiff · Gloved Object-Held-Out | Volumetric IoU: 31.9 -> 38.9 | Improved |
| add spatial mask loss (ablation) | EgoPressureDiff · Gloved Object-Held-Out | MAE: 3.8 -> 3.4 | Improved |
| fine-tune on bare-hand data after gloved pre-training | EgoPressureDiff · Bare-hand | Contact IoU: 47.2 -> 57.8 | Improved |
| use diffusion model instead of EgoPressureFormer regressor | EgoPressureDiff · Gloved Object-Held-Out | Volumetric IoU: 26.5 -> 38.9 | Improved |
| distill EgoPressureDiff to 8 steps (LCD) | EgoPressureDiff · Gloved Object-Held-Out | Contact IoU: 56.3 -> 54.9 | Mixed |
Closest papers by meaning
- EgoTac: In-the-wild Tactile Prediction from Egocentric Vision (NeurIPS 2026 · similarity 0.92)
- EgoPressure: A Dataset for Hand Pressure and Pose Estimation in Egocentric Vision (CVPR 2025 · similarity 0.92)
- MEgoHand: Multimodal Egocentric Hand-Object Interaction Motion Generation (NeurIPS 2025 · similarity 0.87)
- Estimating Ego-Body Pose from Doubly Sparse Egocentric Video Data (NeurIPS 2024 · similarity 0.87)
- Touch in the Wild: Learning Fine-Grained Manipulation with a Portable Visuo-Tactile Gripper (NeurIPS 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.
