Dataset Viewer
The dataset viewer is not available because its heuristics could not detect any supported data files. You can try uploading some data files, or configuring the data files location manually.
YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
螺钿垂直模型训练数据集
仓库地址:https://huggingface.co/datasets/KaKa427/luodian-training-data
📌 项目简介
现代螺钿产品设计专用多模态AI模型训练数据集。
项目定位:
- 现代化、可落地、经济导向的螺钿产品设计生成器
- 不是:复古大漆螺钿、复杂工艺展示
- 而是:木胎贝母螺钿 + 简约几何 + 2026时尚审美 + 可量产
产品类型: 手持镜、首饰盒、挂件、手机壳、笔筒等小件产品
⚖️ 许可证
CC BY-NC-SA 4.0 - Creative Commons Attribution-NonCommercial-ShareAlike 4.0
你可以:
- ✅ 用于学术研究、课程学习、毕业设计
- ✅ 下载使用、修改改进
- ✅ 论文发表(需署名)
你不可以:
- ❌ 商业使用(最重要限制)
- ❌ 出售数据或模型
- ❌ 用于商业产品或服务
署名要求:
使用数据集或模型时请引用:
数据来源:KaKa427/luodian-training-data
许可证:CC BY-NC-SA 4.0
项目负责:KaKa427
详细许可证:https://creativecommons.org/licenses/by-nc-sa/4.0/
📁 数据集结构
luodian-training-data/
├── README.md # 本文档
├── schema/
│ ├── dataset_schema.json # 数据格式定义(17字段)
│ ├── tags_mapping.json # 口语化→专业术语映射
│ └── CLAUDE.md # 训练约束文档
├── images/
│ ├── seed/ # 教师提供的种子示例图片
│ │ ├── seed_001_mirror.jpg
│ │ ├── seed_002_jewelry_box.jpg
│ │ └── seed_003_pendant.jpg
│ └── students/ # 学生标注的图片
│ ├── 2024001_zhangsan/
│ ├── 2024002_lisi/
│ └── ...
└── annotations/
├── seed/ # 种子图片的JSON标注
│ ├── seed_001.json
│ └── ...
└── students/ # 学生的JSON标注
├── 2024001_zhangsan/
└── ...
🎨 数据格式
图片要求
- 格式:JPG/PNG
- 分辨率:至少1024x1024
- 内容:清晰的螺钿产品照片
- 背景:纯色或简洁背景
- 光线:均匀,突出材质光泽
JSON标注格式(17字段)
参考:schema/dataset_schema.json
核心字段:
{
"image_path": "images/students/2024001/mirror_001.jpg",
"product_type": "handheld_mirror",
"base_material": "紫光檀 (Pterocarpus santalinus)",
"inlay_material": "贝母螺钿 + 银丝",
"design_style": "geometric_silhouette",
"visual_subject": "京剧脸谱(简化版)",
"complexity_level": "medium",
"production_feasibility": "high",
"aesthetic_era": "modern_2026",
"market_orientation": "欧美市场",
"technical_specs": {
"dimensions": "12x8x0.5",
"inlay_technique": "平嵌 + 线嵌组合",
"carving_depth": "浅雕(2mm)",
"production_time": "6-8小时",
"cost_level": "中等"
},
"visual_description_cn": "黑色紫光檀圆镜,嵌了彩色贝壳做的京剧脸谱,很简洁像剪影,配了银丝线条",
"visual_description_en": "Modern handheld mirror, polished Pterocarpus santalinus base, geometric Peking Opera mask motif in iridescent nacre inlay with silver wire accents, minimalist silhouette style",
"reference_tags": ["京剧", "脸谱", "几何简化", "贝母", "银丝"],
"timestamp": "2026-02-08T12:00:00"
}
🚀 使用方法
下载数据集
方法1:使用CLI
# 安装工具
pip install huggingface_hub
# 下载完整数据集
huggingface-cli download \
KaKa427/luodian-training-data \
--repo-type dataset \
--local-dir ./luodian_data
方法2:使用Python
from huggingface_hub import snapshot_download
snapshot_download(
repo_id="KaKa427/luodian-training-data",
repo_type="dataset",
local_dir="./luodian_data"
)
加载数据集
import json
from pathlib import Path
# 读取种子标注
seed_annotations = []
for json_file in Path("./luodian_data/annotations/seed").glob("*.json"):
with open(json_file, 'r', encoding='utf-8') as f:
seed_annotations.append(json.load(f))
print(f"种子数据集: {len(seed_annotations)} 张图片")
# 读取学生标注
student_annotations = []
for json_file in Path("./luodian_data/annotations/students").rglob("*.json"):
with open(json_file, 'r', encoding='utf-8') as f:
student_annotations.append(json.load(f))
print(f"学生扩充: {len(student_annotations)} 张图片")
📊 数据集统计
| 类别 | 数量 | 说明 |
|---|---|---|
| 种子图片 | 5-10张 | 教师提供的详细标注示例 |
| 学生标注 | 目标100-200张 | 学生扩充的训练数据 |
| 产品类型 | 5-8类 | 镜子、首饰盒、挂件等 |
| 标注字段 | 17个 | 完整的工艺和设计信息 |
🎓 贡献指南
参与学生
欢迎参与训练的学生贡献标注数据!
流程:
- 注册HuggingFace账号
- 阅读学生操作手册
- 下载种子数据参考
- 标注图片(使用interactive_labeling.py工具)
- 通过Pull Request提交
质量要求:
- 图片清晰度:≥1024x1024
- 标注准确率:≥90%
- 工艺可行性:100%(必须可工厂生产)
- 术语一致性:参考tags_mapping.json
贡献者名单
指导教师:KaKa427 参与学生:
- 待更新...
📖 相关资源
项目文档
- 学生操作手册:
学生操作手册-HuggingFace协作版.md - 训练约束:
schema/CLAUDE.md - 数据格式:
schema/dataset_schema.json
相关仓库
- 模型仓库:https://huggingface.co/KaKa427/luodian-vertical-model
- 基座模型:https://huggingface.co/llava-hf/llava-1.5-7b-hf
🔬 引用
如在学术研究中使用本数据集,请引用:
@misc{luodian2026,
title={现代螺钿产品设计垂直AI模型训练数据集},
author={KaKa427 and 贡献学生},
year={2026},
publisher={HuggingFace},
howpublished={\url{https://huggingface.co/datasets/KaKa427/luodian-training-data}},
note={License: CC BY-NC-SA 4.0}
}
📞 联系方式
- GitHub Issues:技术问题、Bug报告
- 项目讨论:[项目群/论坛]
- 商业合作:[联系邮箱](遵守CC BY-NC-SA 4.0许可证)
注意:本数据集仅供学术研究使用,禁止商业用途。
最后更新:2026-02-08 维护者:KaKa427
- Downloads last month
- 9