lishenyu/Plant3D
收藏Hugging Face2026-04-01 更新2026-04-12 收录
下载链接：
https://hf-mirror.com/datasets/lishenyu/Plant3D
下载链接
链接失效反馈官方服务：
资源简介：
---
license: mit
---
# Plant3D: A Self-Constructed Plant 3D Reconstruction Dataset
The **Plant3D** dataset is a **self-constructed dataset** created during my experiments on **RGB-D based 3D reconstruction of plants**.
This dataset focuses **only on plants**, each with multiple reconstructed instances. If you want more data about Multi-Object, please check out: https://huggingface.co/datasets/yixuan-huang/Obj3D.
These datasets will be **continuously updated**.
---
## 📂 Dataset Structure
Each plant instance has the following structure:
```
plant_xxx/
│── color/ # RGB images
│── depth/ # Depth images
│── sparse/ # Sparse reconstruction results
│── dense/ # Dense reconstruction results
│── database.db # COLMAP database file
```
### 🔹 Special Cases
```
plant_xxx/
│── mask_color/ # Masked RGB images
│── mask_depth/ # Masked Depth images
│── raw_color/ # Original RGB images
│── raw_depth/ # Original Depth images
│── sparse/
│── dense/
│── database.db
````
- `mask_color/`, `mask_depth/` → Images after applying **green mask** and **fixed depth-range filtering**
- `raw_color/`, `raw_depth/` → Original images
---
## 🚀 Usage
The **Plant3D** dataset can be useful for:
- Benchmarking **plant-specific 3D reconstruction algorithms**
- Research on **RGB-D plant modeling**
- Validating **sparse/dense reconstruction pipelines**
---
## 🛠️ Tool Scripts
👉 For reconstruction scripts and utilities, please check: https://github.com/yixuanhuangm/multiview-recon
---
## 📷 Example Visualization
todo...
---
## 📖 Citation
If you use **Plant3D** in your research, please consider citing it as:
```bibtex
@misc{huang2025plant3d,
author = {Yixuan Huang},
title = {Plant3D: A Self-Constructed Plant 3D Reconstruction Dataset},
year = {2025},
howpublished = {HuggingFace Datasets},
url = {https://huggingface.co/datasets/yixuan-huang/Plant3D/}
}
````
---
## 📬 Contact
If you have any questions, please contact: `yixuanhm@gmail.com`
提供机构：
lishenyu
