OCR MCP Service

"""多引擎对比验证脚本 - 对比不同OCR引擎的性能 注意：这是验证脚本，不是pytest单元测试。 pytest单元测试位于 tests/ 目录下。 """ import sys import glob from pathlib import Path # Add project root to path before importing scripts.common project_root = Path(__file__).parent.parent sys.path.insert(0, str(project_root)) # Setup script environment from scripts.common import setup_script setup_script() # 查找测试图片 test_images = glob.glob('*/IMG_20251124_220855.jpg') if not test_images: test_images = glob.glob('**/IMG_20251124_220855.jpg') if not test_images: print("未找到测试图片") sys.exit(1) image_path = Path(test_images[0]).resolve() print("=" * 80) print("多引擎 OCR 对比测试") print("=" * 80) print(f"测试图片: {image_path}") print() # 要测试的引擎列表 engines_to_test = ["paddleocr", "paddleocr_mcp", "deepseek"] # 尝试添加 EasyOCR 和 Tesseract（如果可用） try: from ocr_mcp_service.ocr_engine import OCREngineFactory test_engine = OCREngineFactory.get_engine("easyocr") engines_to_test.append("easyocr") except: pass try: test_engine = OCREngineFactory.get_engine("tesseract") engines_to_test.append("tesseract") except: pass print(f"将测试的引擎: {', '.join(engines_to_test)}") print() from ocr_mcp_service.ocr_engine import OCREngineFactory from ocr_mcp_service.utils import validate_image import time # 验证图片 validate_image(str(image_path)) results = {} for engine_name in engines_to_test: print("-" * 80) print(f"测试引擎: {engine_name}") print("-" * 80) try: start_time = time.time() # 加载引擎 print(f" 加载 {engine_name} 引擎...") engine = OCREngineFactory.get_engine(engine_name) # 识别 print(f" 开始识别...") result = engine.recognize_image(str(image_path)) elapsed_time = time.time() - start_time results[engine_name] = { "success": True, "text": result.text, "confidence": result.confidence, "processing_time": result.processing_time, "total_time": elapsed_time, "boxes_count": len(result.boxes), "text_length": len(result.text), } print(f" ✓ 成功") print(f" 处理时间: {result.processing_time:.2f}s") print(f" 总时间: {elapsed_time:.2f}s") print(f" 置信度: {result.confidence:.2f}") print(f" 文本框数量: {len(result.boxes)}") print(f" 文本长度: {len(result.text)} 字符") print(f" 文本预览 (前200字符):") print(f" {result.text[:200]}...") except Exception as e: results[engine_name] = { "success": False, "error": str(e) } print(f" ✗ 失败: {e}") print() # 对比总结 print("=" * 80) print("对比总结") print("=" * 80) successful = {k: v for k, v in results.items() if v.get("success")} if len(successful) > 0: print("\n性能对比:") print(f"{'引擎':<20} {'处理时间(s)':<15} {'总时间(s)':<15} {'置信度':<12} {'文本长度':<12} {'文本框数':<12}") print("-" * 80) for name, result in successful.items(): print(f"{name:<20} {result['processing_time']:<15.2f} {result['total_time']:<15.2f} {result['confidence']:<12.2f} {result['text_length']:<12} {result['boxes_count']:<12}") if len(successful) > 1: # 找出最快的 fastest = min(successful.items(), key=lambda x: x[1]['processing_time']) print(f"\n最快: {fastest[0]} ({fastest[1]['processing_time']:.2f}s)") # 找出置信度最高的 best_conf = max(successful.items(), key=lambda x: x[1]['confidence']) print(f"最高置信度: {best_conf[0]} ({best_conf[1]['confidence']:.2f})") # 找出文本最长的（可能表示识别更完整） longest = max(successful.items(), key=lambda x: x[1]['text_length']) print(f"最长文本: {longest[0]} ({longest[1]['text_length']} 字符)") print("\n" + "=" * 80) print("详细结果") print("=" * 80) for engine_name, result in results.items(): print(f"\n【{engine_name}】") if result.get("success"): print(f"文本内容:\n{result['text']}") else: print(f"错误: {result.get('error', 'Unknown error')}") def main(): """Main entry point for script execution.""" # Script logic is already executed at module level pass if __name__ == "__main__": main()