PP-OCRv6 returnWordBox 框不精准

Search before asking

• I have searched the PaddleOCR Docs and found no similar bug report.
• I have searched the PaddleOCR Issues and found no similar bug report.
• I have searched the PaddleOCR Discussions and found no similar bug report.

Bug (问题描述)

pp-ocr medium v6 使用returnWordBox 对比 Tesseract；pp-ocr v6 medium 完全不精准， 同时也变换很多参数， 情况还是一样。

pp-ocr v6.0 page18_ppocrv6.pdf

Tesseract page18_tesseract.pdf

‍♂️ Environment (运行环境)

OS: Windows 11 + Docker Python: 3.12 PP-OCR v6 Medium RAM: 32GB CPU: 275HX GPU: 5080

Minimal Reproducible Example (最小可复现问题的Demo)

` def _ocr(file: str, file_type: str = "PDF", url: str = "http://localhost:8080/ocr"): #with open(image_path, "rb") as f: # img_b64 = base64.b64encode(f.read()).decode("utf-8")

payload = { "file": file,
            "fileType": 0 if file_type == "PDF" else 1,    # 0=PDF，1=图片
            "lang": "en",
            "useDocOrientationClassify": False,
            "useDocUnwarping": False,
            "useTextlineOrientation": False,
            "textDetLimitSideLen": 1280,
            "textDetLimitType": "max",
            "textDetThresh": 0.25,
            "textDetBoxThresh": 0.6,
            "textDetUnclipRatio": 2.2,
            "textRecScoreThresh": 0,
            "returnWordBox": True,
            "visualize": False,
            "logId": "string"

} response = requests.post(url, json=payload) assert response.status_code == 200, (response.status_code, response.text) #resp.raise_for_status() return response.json()['result']

`