SMM-Dataset / README.md
yinbq's picture
Add README
aa864c4 verified
---
title: SMM Dataset
license: mit
tags:
- dataset
- computer-vision
- 3d
- blocks
- spatial-reasoning
- minecraft
size_categories:
- 10G<n<100G
---
# SMM Dataset
This dataset contains 3D block manipulation and spatial reasoning data for training AI models on spatial understanding and block construction tasks.
## πŸ“¦ Download
The dataset is provided as a compressed archive:
- **File**: `semantic_blocks_part2.tar.gz`
- **Size**: 0.81 GB (compressed)
- **Format**: GZ compressed tar archive
### Extract the dataset:
```bash
# For .tar.gz files
tar -xzf semantic_blocks_part2.tar.gz
# For .tar.xz files
tar -xJf semantic_blocks_part2.tar.gz
# For .tar.bz2 files
tar -xjf semantic_blocks_part2.tar.gz
```
## πŸ“Š Dataset Structure
After extraction, you will find:
```
SMM/
β”œβ”€β”€ SMM_data/ # Main dataset files
β”œβ”€β”€ random_pipeline/ # Random block generation pipeline
β”œβ”€β”€ semantic_blocks_part1/ # First part of semantic block data
β”œβ”€β”€ semantic_blocks_part2/ # Second part of semantic block data
└── sementic_blocks.tar.gz # Additional compressed semantic blocks
```
## 🎯 Usage
This dataset can be used for:
- Training models on spatial reasoning
- 3D object manipulation and understanding
- Block construction and planning tasks
- Multi-view 3D reconstruction
- Sequential decision making in 3D environments
## πŸ“ Data Format
The dataset contains:
- **Images**: RGB images of block structures from multiple viewpoints
- **Annotations**: Spatial relationships and block configurations
- **Metadata**: Task descriptions and ground truth labels
## πŸ”— Related Resources
- **Paper**: [Link to paper if available]
- **Code**: [Link to training/inference code]
- **Model**: [Link to trained models]
## πŸ“„ License
This dataset is released under the MIT License.
## πŸ™ Citation
If you use this dataset in your research, please cite:
```bibtex
@dataset{smm_dataset,
title={SMM Dataset: Spatial Reasoning with Minecraft Blocks},
author={Your Name},
year={2025},
publisher={Hugging Face},
url={https://huggingface.co/datasets/yinbq/SMM-Dataset}
}
```
## πŸ“§ Contact
For questions or issues, please open an issue on the dataset repository.
---
**Note**: This is a large dataset. Make sure you have sufficient disk space (~40GB) before extracting.