Просмотр исходного кода

feat(config): 添加 YUSYS-OCR v4.0 支持,更新多个文档的 OCR 工具配置

zhch158_admin 1 неделя назад
Родитель
Сommit
85626ae88e

+ 7 - 0
ocr_validator/config/A用户_单元格扫描流水.yaml

@@ -4,6 +4,13 @@ document:
 
   # 🎯 关键改进:定义该文档使用的 OCR 工具及其结果目录
   ocr_results:
+    # bank_statement_yusys_v4
+    - tool: "mineru"
+      result_dir: "bank_statement_yusys_v4"
+      image_dir: "bank_statement_yusys_v4/{{name}}"
+      description: "YUSYS-OCR框架 v4.0 GLM-OCR"
+      enabled: true
+
     # bank_statement_yusys_v3
     - tool: "mineru"
       result_dir: "bank_statement_yusys_v3"

+ 7 - 0
ocr_validator/config/B用户_扫描流水.yaml

@@ -4,6 +4,13 @@ document:
   
   # 🎯 关键改进:定义该文档使用的 OCR 工具及其结果目录
   ocr_results:
+    # bank_statement_yusys_v4
+    - tool: "mineru"
+      result_dir: "bank_statement_yusys_v4"
+      image_dir: "bank_statement_yusys_v4/{{name}}"
+      description: "YUSYS-OCR框架 v4.0 GLM-OCR"
+      enabled: true
+
     # bank_statement_yusys_v3
     - tool: "mineru"
       result_dir: "bank_statement_yusys_v3"

+ 0 - 2
ocr_validator/config/global.yaml

@@ -160,5 +160,3 @@ data_sources:
   - 张_微信图.yaml
   - 付_工商银行943825图.yaml
   - 许_民生银行图.yaml
-
-

+ 7 - 0
ocr_validator/config/乔_建设银行图.yaml

@@ -4,6 +4,13 @@ document:
   
   # 🎯 关键改进:定义该文档使用的 OCR 工具及其结果目录
   ocr_results:
+    # bank_statement_yusys_v4
+    - tool: "mineru"
+      result_dir: "bank_statement_yusys_v4"
+      image_dir: "bank_statement_yusys_v4/{{name}}"
+      description: "YUSYS-OCR框架 v4.0 GLM-OCR"
+      enabled: true
+
     # bank_statement_yusys_v3
     - tool: "mineru"
       result_dir: "bank_statement_yusys_v3"

+ 7 - 0
ocr_validator/config/付_工商银行943825图.yaml

@@ -4,6 +4,13 @@ document:
   
   # 🎯 关键改进:定义该文档使用的 OCR 工具及其结果目录
   ocr_results:
+    # bank_statement_yusys_v4
+    - tool: "mineru"
+      result_dir: "bank_statement_yusys_v4"
+      image_dir: "bank_statement_yusys_v4/{{name}}"
+      description: "YUSYS-OCR框架 v4.0 GLM-OCR"
+      enabled: true
+
     # bank_statement_yusys_v3
     - tool: "mineru"
       result_dir: "bank_statement_yusys_v3"

+ 7 - 0
ocr_validator/config/对公_招商银行图.yaml

@@ -4,6 +4,13 @@ document:
   
   # 🎯 关键改进:定义该文档使用的 OCR 工具及其结果目录
   ocr_results:
+    # bank_statement_yusys_v4
+    - tool: "mineru"
+      result_dir: "bank_statement_yusys_v4"
+      image_dir: "bank_statement_yusys_v4/{{name}}"
+      description: "YUSYS-OCR框架 v4.0 GLM-OCR"
+      enabled: true
+
     # bank_statement_yusys_v3
     - tool: "mineru"
       result_dir: "bank_statement_yusys_v3"

+ 7 - 0
ocr_validator/config/山西云集科技有限公司.yaml

@@ -4,6 +4,13 @@ document:
   
   # 🎯 关键改进:定义该文档使用的 OCR 工具及其结果目录
   ocr_results:
+    # bank_statement_yusys_v4
+    - tool: "mineru"
+      result_dir: "bank_statement_yusys_v4"
+      image_dir: "bank_statement_yusys_v4/{{name}}"
+      description: "YUSYS-OCR框架 v4.0 GLM-OCR"
+      enabled: true
+
     # bank_statement_yusys_v3
     - tool: "mineru"
       result_dir: "bank_statement_yusys_v3"

+ 7 - 0
ocr_validator/config/康强_北京农村商业银行.yaml

@@ -5,6 +5,13 @@ document:
   
   # 🎯 关键改进:定义该文档使用的 OCR 工具及其结果目录
   ocr_results:
+    # bank_statement_yusys_v4
+    - tool: "mineru"
+      result_dir: "bank_statement_yusys_v4"
+      image_dir: "bank_statement_yusys_v4/{{name}}"
+      description: "YUSYS-OCR框架 v4.0 GLM-OCR"
+      enabled: true
+
     # bank_statement_yusys_v3
     - tool: "mineru"
       result_dir: "bank_statement_yusys_v3"

+ 7 - 0
ocr_validator/config/张_微信图.yaml

@@ -4,6 +4,13 @@ document:
   
   # 🎯 关键改进:定义该文档使用的 OCR 工具及其结果目录
   ocr_results:
+    # bank_statement_yusys_v4
+    - tool: "mineru"
+      result_dir: "bank_statement_yusys_v4"
+      image_dir: "bank_statement_yusys_v4/{{name}}"
+      description: "YUSYS-OCR框架 v4.0 GLM-OCR"
+      enabled: true
+
     # bank_statement_yusys_v3
     - tool: "mineru"
       result_dir: "bank_statement_yusys_v3"

+ 7 - 0
ocr_validator/config/德_内蒙古银行照.yaml

@@ -5,6 +5,13 @@ document:
   
   # 🎯 关键改进:定义该文档使用的 OCR 工具及其结果目录
   ocr_results:
+    # bank_statement_yusys_v4
+    - tool: "mineru"
+      result_dir: "bank_statement_yusys_v4"
+      image_dir: "bank_statement_yusys_v4/{{name}}"
+      description: "YUSYS-OCR框架 v4.0 GLM-OCR"
+      enabled: true
+
     # bank_statement_yusys_v3
     - tool: "mineru"
       result_dir: "bank_statement_yusys_v3"

+ 7 - 0
ocr_validator/config/提取自赤峰黄金2023年报.yaml

@@ -4,6 +4,13 @@ document:
   
   # 🎯 关键改进:定义该文档使用的 OCR 工具及其结果目录
   ocr_results:
+    # bank_statement_yusys_v4
+    - tool: "mineru"
+      result_dir: "bank_statement_yusys_v4"
+      image_dir: "bank_statement_yusys_v4/{{name}}"
+      description: "YUSYS-OCR框架 v4.0 GLM-OCR"
+      enabled: true
+
     # bank_statement_yusys_v3
     - tool: "mineru"
       result_dir: "bank_statement_yusys_v3"

+ 7 - 0
ocr_validator/config/施博深.yaml

@@ -4,6 +4,13 @@ document:
   
   # 🎯 关键改进:定义该文档使用的 OCR 工具及其结果目录
   ocr_results:
+    # bank_statement_yusys_v4
+    - tool: "mineru"
+      result_dir: "bank_statement_yusys_v4"
+      image_dir: "bank_statement_yusys_v4/{{name}}"
+      description: "YUSYS-OCR框架 v4.0 GLM-OCR"
+      enabled: true
+
     # bank_statement_yusys_v3
     - tool: "mineru"
       result_dir: "bank_statement_yusys_v3"

+ 7 - 0
ocr_validator/config/朱_中信银行图.yaml

@@ -4,6 +4,13 @@ document:
   
   # 🎯 关键改进:定义该文档使用的 OCR 工具及其结果目录
   ocr_results:
+    # bank_statement_yusys_v4
+    - tool: "mineru"
+      result_dir: "bank_statement_yusys_v4"
+      image_dir: "bank_statement_yusys_v4/{{name}}"
+      description: "YUSYS-OCR框架 v4.0 GLM-OCR"
+      enabled: true
+
     # bank_statement_yusys_v3
     - tool: "mineru"
       result_dir: "bank_statement_yusys_v3"

+ 7 - 0
ocr_validator/config/湛_平安银行图.yaml

@@ -4,6 +4,13 @@ document:
   
   # 🎯 关键改进:定义该文档使用的 OCR 工具及其结果目录
   ocr_results:
+    # bank_statement_yusys_v4
+    - tool: "mineru"
+      result_dir: "bank_statement_yusys_v4"
+      image_dir: "bank_statement_yusys_v4/{{name}}"
+      description: "YUSYS-OCR框架 v4.0 GLM-OCR"
+      enabled: true
+
     # bank_statement_yusys_v3
     - tool: "mineru"
       result_dir: "bank_statement_yusys_v3"

+ 9 - 2
ocr_validator/config/至远彩色_2023年报.yaml

@@ -4,6 +4,13 @@ document:
   
   # 🎯 关键改进:定义该文档使用的 OCR 工具及其结果目录
   ocr_results:
+    # bank_statement_yusys_v4
+    - tool: "mineru"
+      result_dir: "bank_statement_yusys_v4"
+      image_dir: "bank_statement_yusys_v4/{{name}}"
+      description: "YUSYS-OCR框架 v4.0 GLM-OCR"
+      enabled: true
+
     # bank_statement_yusys_v3
     - tool: "mineru"
       result_dir: "bank_statement_yusys_v3"
@@ -13,8 +20,8 @@ document:
 
     # bank_statement_yusys_v2
     - tool: "mineru"
-      result_dir: "bank_statement_yusys_v2"
-      image_dir: "mineru_vllm_results/{{name}}"
+      result_dir: "bank_statement_glm_vl"
+      image_dir: "bank_statement_glm_vl/{{name}}"
       description: "YUSYS统一OCR框架"
       enabled: true
 

+ 7 - 0
ocr_validator/config/许_民生银行图.yaml

@@ -4,6 +4,13 @@ document:
   
   # 🎯 关键改进:定义该文档使用的 OCR 工具及其结果目录
   ocr_results:
+    # bank_statement_yusys_v4
+    - tool: "mineru"
+      result_dir: "bank_statement_yusys_v4"
+      image_dir: "bank_statement_yusys_v4/{{name}}"
+      description: "YUSYS-OCR框架 v4.0 GLM-OCR"
+      enabled: true
+
     # bank_statement_yusys_v3
     - tool: "mineru"
       result_dir: "bank_statement_yusys_v3"