invoice_ocr_mcp/configs/models.yaml

企业发票OCR识别MCP服务器，基于RapidOCR和ModelScope生态，支持多种发票类型的结构化识别

# ModelScope 模型配置文件

models:
  # 文本检测模型
  text_detection:
    name: "damo/cv_resnet18_ocr-detection-line-level_damo"
    version: "v1.0.2"
    cache_dir: "./cache/models/text_detection"
    task_type: "ocr-detection"
    description: "基于ResNet18的文本行检测模型"
    
  # 文本识别模型  
  text_recognition:
    name: "damo/cv_convnextTiny_ocr-recognition-general_damo"
    version: "v1.0.1"
    cache_dir: "./cache/models/text_recognition"
    task_type: "ocr-recognition"
    description: "基于ConvNext的通用文本识别模型"
    
  # 发票分类模型
  invoice_classification:
    name: "damo/cv_resnest50_ocr-invoice-classification"
    version: "v1.0.0"
    cache_dir: "./cache/models/classification"
    task_type: "image-classification"
    description: "基于ResNeSt50的发票类型分类模型"
    
  # 关键信息抽取模型
  info_extraction:
    name: "damo/nlp_structbert_document-classification_chinese-base"
    version: "v1.0.0"
    cache_dir: "./cache/models/info_extraction"
    task_type: "text-classification"
    description: "基于StructBERT的文档信息抽取模型"

# 发票类型映射
invoice_types:
  "01": "增值税专用发票"
  "03": "机动车增值税专用发票"
  "04": "增值税普通发票"
  "10": "增值税电子普通发票"
  "11": "增值税普通发票（卷式）"
  "14": "增值税普通发票（通行费）"
  "15": "二手车发票"
  "20": "增值税电子专用发票"
  "99": "数电发票（增值税专用发票）"
  "09": "数电发票（普通发票）"
  "61": "数电发票（航空运输电子客票行程单）"
  "83": "数电发票（铁路电子客票）"
  "100": "区块链发票（支持深圳、北京和云南地区）"

# 推理配置
inference:
  # GPU配置
  gpu:
    enabled: false
    device_id: 0
    memory_limit: 8192  # MB
    
  # 批处理配置
  batch:
    max_size: 32
    timeout: 30
    
  # 缓存配置
  cache:
    enabled: true
    max_entries: 1000
    ttl: 3600  # 秒
    
# 性能优化
optimization:
  # 模型量化
  quantization:
    enabled: false
    precision: "int8"
    
  # 模型并行
  parallel:
    enabled: true
    num_workers: 4
    
  # 内存优化
  memory:
    enable_offload: false
    max_memory_usage: 0.8  # 80% of available memory