invoice_ocr_mcp
Version:
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
69 lines (61 loc) • 1.89 kB
Markdown
# 期望结果
这个目录存放对应测试图片的期望识别结果，用于验证OCR识别的准确性。
## 文件结构
每个测试图片都对应一个JSON文件，包含期望的识别结果：
- `vat_special_invoice.json` - 增值税专用发票期望结果
- `vat_common_invoice.json` - 增值税普通发票期望结果
- `electronic_invoice.json` - 电子发票期望结果
- `digital_invoice.json` - 数电发票期望结果
## JSON格式说明
```json
{
"invoice_type": {
"code": "01",
"name": "增值税专用发票",
"confidence": 0.95
},
"basic_info": {
"invoice_number": "12345678",
"invoice_date": "2024-01-15",
"total_amount": "1000.00",
"tax_amount": "130.00",
"amount_without_tax": "870.00"
},
"seller_info": {
"name": "示例科技有限公司",
"tax_id": "91110000123456789X",
"address": "北京市朝阳区示例街道123号",
"phone": "010-12345678",
"bank_account": "1234567890123456789"
},
"buyer_info": {
"name": "客户公司",
"tax_id": "91110000987654321Y",
"address": "北京市海淀区客户路456号",
"phone": "010-87654321",
"bank_account": "9876543210987654321"
},
"items": [
{
"name": "测试商品",
"specification": "标准版",
"unit": "台",
"quantity": "1",
"unit_price": "870.00",
"amount": "870.00",
"tax_rate": "13%",
"tax_amount": "113.10"
}
],
"verification": {
"check_code": "12345678",
"machine_number": "499098765432",
"is_valid": true
}
}
```
## 使用说明
1. 期望结果用于自动化测试的结果验证
2. 可以设置容差范围，允许识别结果在合理范围内波动
3. 关键字段（如发票号码、金额）应该完全匹配
4. 置信度字段可以有适当的误差范围