Upload README.md with huggingface_hub

README.md

@@ -1,77 +1,188 @@
 ---
 license: odbl
-Author: Dexrobot Team
+task_categories:
+  - robotics
+  - computer-vision
+  - video-classification
+tags:
+  - dexterous-manipulation
+  - hand-object-interaction
+  - motion-capture
+  - physics-simulation
+  - rgbd
+  - contact-forces
+size_categories:
+  - 10<n<1K
 ---
-# DexCanvas
-## - Human Dexterous Manipulation Dataset & Simulation Platform
 
-  🔬 Coming Soon - Expected Release: Mid-October 2025
+# DexCanvas: Dexterous Manipulation Dataset v0.1
 
-  We are excited to announce the upcoming release of MOCAP, a comprehensive dataset and simulation platform for human dexterous manipulation research.
+**⚠️ TEST RELEASE**: This is a preview version containing 1% of the full dataset. Contact force data is not included in v0.1.
 
-  📊 Dataset Overview
+DexCanvas is a large-scale hybrid dataset for robotic hand-object interaction research, combining real human demonstrations with physics-validated simulation data.
 
-  Our team has collected and processed approximately 10,000 recordings / 200 hours of human manipulation data, categorized by:
+## Dataset Statistics (v0.1 Test Release)
 
-  - 33 Common Grasping Gestures: Comprehensive coverage of human grasping patterns
-  - ~10 In-Hand Manipulation Actions: Essential dexterous manipulation skills
+- **Total Frames**: ~30 million multi-view RGB-D frames
+- **Total Duration**: ~70 hours of dexterous hand-object interactions
+- **Real Demonstrations**: ~0.7 hours of human mocap data (1/100 of collected data)
+- **Expansion Ratio**: 100× from real to simulated data
+- **Manipulation Types**: 21 types based on Cutkosky taxonomy
+- **Objects**: 30 objects (geometric primitives + YCB objects)
+- **Capture Rate**: 100 Hz optical motion capture
 
-  🛠️ Key Features
+## Manipulation Coverage
 
-  🔥 1M+ Physics-Accurate Records
+The dataset spans four primary grasp categories:
 
-  - 100x Amplified Dataset: Original 10K human recordings expanded to 1M+ samples through RL-based simulation
-  - Physics-Correct: Each simulated sample maintains physical accuracy and realistic dynamics
-  - Rich Contact Information: Complete contact points, forces, and interaction dynamics
+- **Power Grasps**: Full-hand wrapping grips
+- **Intermediate Grasps**: Mixed precision-power combinations
+- **Precision Grasps**: Fingertip-based manipulation
+- **In-Hand Manipulation**: Object reorientation and repositioning
 
-  Data Pipeline
+All 21 manipulation types follow the Cutkosky grasp taxonomy.
 
-  - ✅ Complete data cleaning and preprocessing pipeline
-  - ✅ Quality-assured human operation recordings
-  - ✅ Standardized annotation framework
+## Data Modalities
 
-  Real2Sim Achievement
+Each frame includes:
 
-  - 🎯 Precise replication of human hand operations in simulation environment
-  - 🤖 Reinforcement Learning-based real2sim conversion methodology
-  - 📈 100x data scaling with maintained physical fidelity
+- **RGB-D Data**: Multi-view color and depth images
+- **Hand Pose**: MANO hand parameters with high-precision tracking
+- **Object State**: 6-DoF pose and object wrenches
+- **Annotations**: Per-frame labels and metadata
 
-  🎯 Technical Significance
+**Note**: Contact force data is not included in v0.1. Contact forces will be available in future releases.
 
-  1. End-to-End Validation
+## Data Pipeline
 
-  Complete workflow from "how to collect" to "how to use", ensuring data collection and preprocessing accuracy.
+The dataset is generated through three stages:
 
-  2. Physics-Informed Insights
+1. **Real Capture**: Optical motion capture of human demonstrations at 30 Hz
+2. **Force Reconstruction**: RL-based physics simulation to infer contact forces
+3. **Physics Validation**: Verification of contact points, forces, and object dynamics
 
-  Obtain comprehensive contact points, force information, and interaction dynamics through physics simulation - data impossible to observe directly in real-world scenarios.
+This hybrid approach provides contact information impossible to observe directly in real-world scenarios while maintaining physical accuracy.
 
-  3. Massive Scale with Physical Accuracy
+## Installation
 
-  - 1M+ physically-accurate samples generated through RL-enhanced simulation
-  - Complete contact information for every manipulation sequence
-  - Verified physics consistency across all synthesized data
+```bash
+pip install datasets huggingface_hub
+```
 
-  🚀 What's Coming
+For image processing and visualization:
 
-  - 📦 Complete Dataset Release: 1M+ physics-accurate manipulation recordings
-  - 🔧 Preprocessing Tools: Ready-to-use data pipeline
-  - 🎮 Simulation Environment: Physics-based replication platform
-  - 📊 Contact Analysis Tools: Contact force and interaction analysis utilities
-  - 📚 Documentation & Tutorials: Comprehensive usage guides
+```bash
+pip install pillow numpy torch
+```
 
-  📅 Release Timeline
+Authenticate with HuggingFace (required for private datasets):
 
-  Expected Release Date: Mid-October 2025
+```bash
+huggingface-cli login
+```
 
-  Stay tuned for updates!
+Or set your token as an environment variable:
 
-  ---
-  Developed by DexRobot TeamLast Updated: September 11, 2025
+```bash
+export HF_TOKEN="your_token_here"
+```
 
-  📬 Contact
+## Quick Start
 
-  For early access requests or collaboration inquiries, please reach out to our team at https://www.dex-robot.com/en
+### Data Structure
 
-  ---
-  ⭐ Star this space to get notified when the dataset becomes available!
+```json
+{
+  "trajectory_meta_data": {
+    "generated_data": "int",
+    "data_fps": "int",
+    "mocap_raw_data_source": {
+      "operator": "str",
+      "object": "str",
+      "gesture": "str"
+    },
+    "total_frames": "int",
+    "mano_hand_shape": "(10,)"
+    //...
+  },
+  "sequence_info": {
+    "timestamp": "(T,)",
+    "hand_joint": {
+      "position": "(T, 3)",
+      "rotation": "(T, 3)",
+      "finger_pose": "(T, 48)"
+    },
+    "object_info": {
+      "pose": "(T, 6)"
+    },
+    "mano_model_output": {
+      "joints": "(T, 63)"
+    }
+  }
+}
+```
+### Visualization
+
+Visualize trajectories using the **mocap_loader**:
+
+```bash
+# Install dependencies
+pip install open3d trimesh scipy
+
+# Visualize trajectory
+python -m hand_trajectory_loader.examples.visualize_trajectory \
+    dataset.parquet 0 \
+    --mano-model assets/mano/models/MANO_RIGHT.pkl \
+    --object assets/objects/cube1.stl \
+    --show-joints
+```
+
+Controls: **SPACE** pause/resume, **M** toggle hand mesh, **O** toggle object, **Q** quit
+
+## Version Information
+
+**v0.1 (Test Release)** includes:
+- 1% of collected real human demonstration data
+- MANO hand parameters
+- Object pose data
+- Manipulation type annotations
+
+**Coming in future releases**:
+- Complete dataset (100× larger than v0.1)
+- Contact force data with physics validation
+- Additional objects and manipulation types
+- Extended annotations and metadata
+
+## Contact
+
+**Research Collaboration**
+Academic inquiries: lyw@dex-robot.com
+
+**Business Inquiries**
+Business collaboration: info@dex-robot.com
+
+**Website**
+https://www.dex-robot.com/en
+https://dexcanvas.github.io/
+
+## Citation
+
+```bibtex
+@article{dexcanvas2025,
+  title={DexCanvas: A Large-Scale Hybrid Dataset for Dexterous Manipulation},
+  author={DexRobot Team},
+  year={2025},
+  eprint={2510.15786},
+  archivePrefix={arXiv},
+  url={https://arxiv.org/abs/2510.15786}
+}
+```
+
+## License
+
+This dataset is released under the Open Database License (ODbL).
+
+---
+
+**Developed by DexRobot Team**
+Last Updated: January 2025