Tac-DINO: Learning Vision-Tactile Features with Patch Alignment
Not on the map. Why?
Abstract (arXiv)
Touch is the primary medium through which humans interact with the environment. Currently, tactile learning mainly focuses on image-level pretraining or alignment. However, tactile signals correspond to local object contact, while research into scale alignment and holographic matching remains limited and proper datasets and benchmarks also lack. To bridge this gap, we first construct a data collection system to acquire a large-scale tactile dataset, with over 20 K tactile contacts from 505 real-world objects. Building on this dataset, we design a Vis-Tac Holographic Matching Benchmark to evaluate vision-tactile local-to-global alignment ability. Then we propose Vision-Tactile Patch Alignment (VTPA) methods for vision-tactile representation learning. Experiments demonstrate that these exceed the performance of methods without alignment and align with whole-object images.
Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 55 of its 68 references that are in the map. First posted on arXiv 10 Jun 2026.
Builds on
55 of 68 references.
References · 55
- 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
- 2014Microsoft COCO: Common Objects in Contextcited by 17,608
- 2023DINOv2: Learning Robust Visual Features without Supervisioncited by 7,974
- 2021Masked Autoencoders Are Scalable Vision Learnerscited by 7,403
- 2022Generative Adversarial Networkscited by 5,627
- 2022Objaverse: A Universe of Annotated 3D Objectscited by 1,540
- 2025VGGT: Visual Geometry Grounded Transformercited by 1,539
- 2022DINO: DETR with Improved DeNoising Anchor Boxes for End-to-End Object Detectioncited by 1,319
- 2023Objaverse-XL: A Universe of 10M+ 3D Objectscited by 774
- 2024Structured 3D Latents for Scalable and Versatile 3D Generationcited by 770
- 2024Revisiting Feature Prediction for Learning Visual Representations from Videocited by 531
- 2020DIGIT: A Novel Design for a Low-Cost Compact High-Resolution Tactile Sensor with Application to In-Hand Manipulationcited by 167
38 more
- 2025Reactive Diffusion Policy: Slow-Fast Visual-Tactile Policy Learning for Contact-Rich Manipulationcited by 108
- 2024Binding Touch to Everything: Learning Unified Multimodal Tactile Representationscited by 95
- 2022Touch and Go: Learning from Human-Collected Vision and Touchcited by 87
- 2022ObjectFolder 2.0: A Multisensory Object Dataset for Sim2Real Transfercited by 75
- 20243D-ViTac: Learning Fine-Grained Manipulation with Visuo-Tactile Sensingcited by 72
- 2021ObjectFolder: A Dataset of Objects with Implicit Visual, Auditory, and Tactile Representationscited by 68
- 2017The Feeling of Success: Does Touch Sensing Help Predict Grasp Outcomes?cited by 67
- 2023Dexterity from Touch: Self-Supervised Pre-Training of Tactile Representations with Robotic Playcited by 62
- 2023Robot Synesthesia: In-Hand Manipulation with Visuotactile Sensingcited by 56
- 2025AnyTouch: Learning Unified Static-Dynamic Representation across Multiple Visuo-tactile Sensorscited by 48
- 2021GelSlim3.0: High-Resolution Measurement of Shape, Force and Slip in a Compact Tactile-Sensing Fingercited by 41
- 2023The ObjectFolder Benchmark: Multisensory Learning with Neural and Real Objectscited by 41
- 2025A Survey of Sim-to-Real Methods in RL: Progress, Prospects and Challenges with Foundation Modelscited by 38
- 2025ViTacFormer: Learning Cross-Modal Representation for Visuo-Tactile Dexterous Manipulationcited by 38
- 2025Vision in Action: Learning Active Perception from Human Demonstrationscited by 37
- 2017See, Hear, and Read: Deep Aligned Representationscited by 32
- 2025Touch in the Wild: Learning Fine-Grained Manipulation with a Portable Visuo-Tactile Grippercited by 29
- 2026Visuo-Tactile World Modelscited by 29
- 2017Improved GelSight Tactile Sensor for Measuring Geometry and Slipcited by 28
- 2024Touch100k: A Large-Scale Touch-Language-Vision Dataset for Touch-Centric Multimodal Representationcited by 27
- 2024Tactile-Augmented Radiance Fieldscited by 26
- 2024AnySkin: Plug-and-play Skin Sensing for Robotic Touchcited by 23
- 2023From Isolated Islands to Pangea: Unifying Semantic Space for Human Action Understandingcited by 19
- 2025exUMI: Extensible Robot Teaching System with Action-aware Task-agnostic Tactile Representationcited by 14
- 2024Learning Precise, Contact-Rich Manipulation through Uncalibrated Tactile Skinscited by 12
- 2024TacEx: GelSight Tactile Simulation in Isaac Sim -- Combining Soft-Body and Visuotactile Simulatorscited by 12
- 2025Adaptive Visuo-Tactile Fusion with Predictive Force Attention for Dexterous Manipulationcited by 12
- 2026AnyTouch 2: General Optical Tactile Representation Learning For Dynamic Tactile Perceptioncited by 12
- 2022Beyond Object Recognition: A New Benchmark towards Object Concept Learningcited by 11
- 2023PACE: A Large-Scale Dataset with Pose Annotations in Cluttered Environmentscited by 11
- 2025Tactile-Conditioned Diffusion Policy for Force-Aware Robotic Manipulationcited by 8
- 2026TacUMI: A Multi-Modal Universal Manipulation Interface for Contact-Rich Taskscited by 8
- 2024Generalize by Touching: Tactile Ensemble Skill Transfer for Robotic Furniture Assemblycited by 7
- 2024EyeSight Hand: Design of a Fully-Actuated Dexterous Robot Hand with Integrated Vision-Based Tactile Sensors and Compliant Actuationcited by 7
- 2023TacIPC: Intersection- and Inversion-free FEM-based Elastomer Simulation For Optical Tactile Sensorscited by 6
- 2024Tactile DreamFusion: Exploiting Tactile Sensing for 3D Generationcited by 6
- 2018Deformation Capture via Soft and Stretchable Sensor Arrayscited by 5
- 2024Contrastive Touch-to-Touch Pretrainingcited by 5
50 most cited of 55 All 55 (JSON)
Built on by
Citing papers per year: 2026: 1.
1 paper.