Ready-to-use OCR for 80+ languages and major writing scripts
- Stars29.6k
- Forks3.6k
- Open Issues528
Apache-2.0
- Python
- C++
- Cuda

About EasyOCR
EasyOCR is an OCR library for reading text from images, with ready-to-use recognition for more than 80 languages and all popular writing scripts, including Latin, Chinese, Arabic, Devanagari, and Cyrillic. One tool handles many scripts without building a custom pipeline for each.
It runs a detection and recognition pipeline in Python with PyTorch. The text reader accepts image files, OpenCV image objects, image bytes, and raw image URLs. You can read multiple languages at once, disable the GPU, reduce output detail, and swap in different detection or recognition models.
EasyOCR is developed by Jaided AI, which also offers enterprise support for custom OCR systems. It installs with pip for local use or deployment in your own environment, and a browser demo is hosted on the project website.
Key features
- 80+ supported languages and multiple writing scripts
- Reads image files, OpenCV arrays, bytes, and raw image URLs
- Python API with multi-language reading
- GPU can be disabled with gpu=False
- Supports alternate detection and recognition model choices
Details
- First released
- 2020
- Platforms
- Web · CLI
- Language
- Python
- Framework
- PyTorch
- Languages
- 80+
- Self-hosting
- Local use with pip
