PP-OCRv6宣布了重大模型扩展和多语言OCR功能
PaddlePaddle宣布推出PP-OCRv6，这是一个具有三个层级的新OCR模型，采用PPLCNetV4主干网络，结合RepLKFPN用于文本检测，EncoderWithLightSVTR用于识别，并支持统一的多语言。
01
证据
- HHugging Face Blog公司披露2026-06-22 21:18
The medium and small tiers support 50 languages in one model family, covering Simplified Chinese, Traditional Chinese, English, Japanese, and 46 Latin-script languages.
- HHugging Face Blog公司披露2026-06-22 21:18
What’s new in PP-OCRv6 Three model tiers PPLCNetV4 backbone RepLKFPN for text detection EncoderWithLightSVTR for recognition Unified multilingual OCR Quick start with PaddleOCR Available inference backends Conclusion Evaluate PP-OCRv6 online, then integrate lightweight, producti…
- HHugging Face Blog公司披露2026-06-22 21:18
What’s new in PP-OCRv6 Three model tiers PPLCNetV4 backbone RepLKFPN for text detection EncoderWithLightSVTR for recognition Unified multilingual OCR Quick start with PaddleOCR Available inference backends Conclusion Evaluate PP-OCRv6 online, then integrate lightweight, producti…