Research map / Vision-language and multimodal / Human and embodied modeling
Seeing Through Touch: Tactile-Driven Visual Localization of Material Regions
Seongyu Kim, Seungwoo Lee, Hyeonggon Ryu, Joon Son Chung, Arda Senocak
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 |
|---|---|---|---|
| use local dense alignment instead of CLS global alignment (ablation) | STT · Web-Material | mIoU: 34.74 -> 52.34 | Improved |
| use local dense alignment instead of CLS global alignment (ablation) | STT · OpenSurfaces | mIoU: 19.07 -> 29.47 | Improved |
| add in-domain material diversity pairing (ablation) | STT-Local · Web-Material | mIoU: 52.34 -> 55.73 | Improved |
| add in-domain material diversity pairing (ablation) | STT-Local · OpenSurfaces | mIoU: 29.47 -> 34.10 | Improved |
| add out-domain in-the-wild images to pairing (ablation) | STT-Indomain · Web-Material | mIoU: 55.73 -> 60.94 | Improved |
| add out-domain in-the-wild images to pairing (ablation) | STT-Indomain · OpenSurfaces | mIoU: 34.10 -> 36.73 | Improved |
| add out-domain in-the-wild images to pairing (ablation) | STT-Indomain · TG-Test | mIoU: 77.58 -> 76.82 | Hurt |
| add in-domain pairing, End (weak) tactile frames (ablation) | STT-Local · Web-Material | mIoU: 48.72 -> 53.24 | Improved |
| add out-domain pairing, End (weak) tactile frames (ablation) | STT-Local · Web-Material | mIoU: 48.72 -> 58.56 | Improved |
| add out-domain pairing, Start (weak) tactile frames (ablation) | STT-Local · OpenSurfaces | mIoU: 26.77 -> 35.00 | Improved |
| initialize tactile backbone with DINOv3 instead of T3 (ablation) | STT-Local · Web-Material | mIoU: 39.22 -> 52.34 | Improved |
| initialize tactile backbone with DINOv3 instead of T3 (ablation) | STT out-domain · Web-Material | mIoU: 54.03 -> 60.94 | Improved |
Closest papers by meaning
- PhysTacGen: Physics-Aware Visual-Tactile Sensor Image Generation (NeurIPS 2026 · similarity 0.88)
- AnyTouch: Learning Unified Static-Dynamic Representation across Multiple Visuo-tactile Sensors (ICLR 2025 · similarity 0.87)
- Tactile-Augmented Radiance Fields (CVPR 2024 · similarity 0.87)
- Touch in the Wild: Learning Fine-Grained Manipulation with a Portable Visuo-Tactile Gripper (NeurIPS 2025 · similarity 0.87)
- Tactile DreamFusion: Exploiting Tactile Sensing for 3D Generation (NeurIPS 2024 · similarity 0.86)
- Universal Visuo-Tactile Video Understanding for Embodied Interaction (NeurIPS 2025 · similarity 0.86)
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.
