EasyOCR

Deep learning OCR with high accuracy.

---

Overview

Tasks	OCR
Backends	PyTorch
Speed	1-2s/page
Quality	Very Good
Languages	80+

---

Why EasyOCR

• Higher accuracy than Tesseract on diverse text
• Deep learning - handles varied fonts and styles
• GPU acceleration - optional but faster
• Easy setup - pip install, no system dependencies

---

Basic Usage

from omnidocs.tasks.ocr_extraction import EasyOCR, EasyOCRConfig
from PIL import Image

image = Image.open("document.png")

ocr = EasyOCR(
    config=EasyOCRConfig(
        languages=["en"],
        gpu=True,
    )
)

result = ocr.extract(image)

for block in result.text_blocks:
    print(f"'{block.text}' @ {block.bbox}")

---

Configuration

config = EasyOCRConfig(
    languages=["en"],     # Language codes
    gpu=True,             # Use GPU if available
)

---

Multi-Language

# Multiple languages
config = EasyOCRConfig(
    languages=["en", "fr", "de"],
    gpu=True,
)

Common language codes: en, ch_sim, ch_tra, ja, ko, ar, hi, fr, de, es, pt, ru

---

EasyOCR vs Tesseract

	EasyOCR	Tesseract
Accuracy	Higher	Good
Speed	1-2s/page	0.5-1s/page
GPU	Optional	No
Setup	pip install	System package
Languages	80+	100+

---

When to Use

✅ Need higher accuracy than Tesseract ✅ Have GPU available ✅ Diverse fonts and styles

❌ CPU-only, need speed → Use Tesseract ❌ Asian languages priority → Use PaddleOCR