donghyun
Add OCR code, modules, and weights
8672bad
# -*- coding: utf-8 -*-
"""
================================================================================
Epitext AI Unified Preprocessing Module
================================================================================
톡합 이미지 μ „μ²˜λ¦¬ νŒ¨ν‚€μ§€ (Swin Gray + OCR λ™μ‹œ 생성)
ν•œ 번의 ν•¨μˆ˜ 호좜둜 두 κ°€μ§€ μ „μ²˜λ¦¬ μ™„λ£Œ:
1️⃣ Swin Gray: 그레이 비이진화 (정보 손싀 μ΅œμ†Œ) β†’ JPG 3채널
2️⃣ OCR: 이진화 (λͺ…ν™•ν•œ 흑백) β†’ PNG 1채널
버전: 1.0.0
μƒνƒœ: βœ… Production Ready
μ£Όμš” νŠΉμ§•:
βœ… νš¨μœ¨μ„±: μ˜μ—­ κ²€μΆœ 1회 (두 κ°€μ§€ λͺ¨λ‘ μ‚¬μš©)
βœ… λ°°κ²½ 보μž₯: Swin (밝음) + OCR (ν•˜μ–€μƒ‰)
βœ… 탁본 지원: μžλ™ κ²€μΆœ μ˜΅μ…˜
βœ… μ„€μ • κ°€λŠ₯: JSON 기반 μ»€μŠ€ν„°λ§ˆμ΄μ§•
"""
from .preprocessor_unified import (
UnifiedImagePreprocessor,
get_preprocessor,
preprocess_image_unified
)
from .ocr_engine import (
get_ocr_engine,
OCREngine,
ocr_and_detect
)
from .nlp_engine import (
get_nlp_engine,
NLPEngine,
process_text_with_nlp
)
__version__ = "1.0.0"
__author__ = "Epitext Team"
__all__ = [
"UnifiedImagePreprocessor",
"get_preprocessor",
"preprocess_image_unified",
"get_ocr_engine",
"OCREngine",
"ocr_and_detect",
"get_nlp_engine",
"NLPEngine",
"process_text_with_nlp"
]