PP-OCRV2

Key Features:

• Enhanced Mobile Optimization: Smaller model size with 20% faster inference than V1.

• Multi-Language Expansion: Supports 6+ languages (English, Chinese, Spanish, French, German, Japanese).

• Improved Detection (DB-Net): Better handles skewed and curved text.

• Balanced Speed/Accuracy: 15% higher accuracy than V1 with similar latency.

• Open-Source with Commercial Options: Free for research, paid for enterprise scaling.

Model Deployment Status:

General Availability Yes (Open-source)

Supported Data Types for Input Image (PNG, JPG, PDF)

Supported Data Types for Output Text

Supported # Tokens for Input Single image (max 2048x2048 px)

Supported # Tokens for Output 2k (denser text support)

Knowledge Cutoff June 2022

Tool Use None

Best For:

• Mobile OCR apps (Android/iOS)

• Multi-language document processing

• Availability:

  • GitHub (PaddleOCR)

  • AWS SageMaker