图片RAG系统

图片数据集 (JSONL) 
    ↓
数据预处理
    ↓
向量化 (HuggingFace Embeddings)
    ↓
向量数据库 (ChromaDB)
    ↓
检索器 (Similarity Search)
    ↓
问答链 (LLM + RAG)
    ↓
回答 + 相关图片

# 自动安装脚本 (Linux/macOS)
./install.sh

# 或Windows
install.bat

# 或手动安装
conda env create -f environment.yml
conda activate image-rag-system

pip install -r requirements.txt

export OPENAI_API_KEY="your-openai-api-key"

export SILICONFLOW_API_KEY="your-siliconflow-api-key"

# 激活conda环境
conda activate image-rag-system

# 或使用激活脚本
./activate_env.sh  # Linux/macOS
activate_env.bat   # Windows

# 启动统一主程序
python main.py

# 或使用启动脚本
python start.py

from image_rag_system import ImageRAGSystem

# 创建RAG系统
rag_system = ImageRAGSystem(
    data_path="path/to/your/data.jsonl",
    vector_db_path="./vector_db"
)

# 初始化系统
rag_system.setup()

# 查询
result = rag_system.query("什么是中性粒细胞？")
print(result['answer'])
print(f"相关图片: {len(result['related_images'])} 张")

python simple_example.py

python image_rag_system.py

# 使用SiliconFlow API
python siliconflow_example.py

# 使用SiliconFlow (需要设置API密钥)
python run_siliconflow.py

{
    "image_url": "图片URL",
    "scene_description": "场景描述",
    "characters_in_frame": [
        {
            "character_name": "角色名称",
            "biological_entity": "生物学实体",
            "confidence": "置信度"
        }
    ],
    "biological_processes_or_concepts": [
        {
            "name": "生物学概念名称",
            "explanation": "详细解释",
            "scene_relevance": "场景相关性"
        }
    ],
    "keywords": ["关键词1", "关键词2"],
    "quality_assessment": {
        "technical_quality_score": 0.95
    },
    "content_assessment": {
        "suitability_score": 1.0
    }
}

import logging
logging.basicConfig(level=logging.DEBUG)

from image_rag_system import ImageRAGSystem

# 使用SiliconFlow API
rag_system = ImageRAGSystem(
    data_path="path/to/data.jsonl",
    use_siliconflow=True,
    siliconflow_api_key="your-api-key",
    siliconflow_model="BAAI/bge-large-zh-v1.5"
)

rag_system.setup()
result = rag_system.query("什么是中性粒细胞？")