easyocr

EasyOCR

Ready-to-use OCR with 80+ supported languages and all popular writing scripts including Latin, Chinese, Arabic, Devanagari, Cyrillic and etc.

Input

(from https://github.com/JaidedAI/EasyOCR/tree/master/examples)

Output

recognized text and visualized image

  word=西 bbox=[[86, 80], [134, 80], [134, 128], [86, 128]] confidence=0.5982276948352592
  word=愚园路 bbox=[[187, 75], [469, 75], [469, 165], [187, 165]] confidence=0.7516731254579294
  word=东 bbox=[[517, 81], [565, 81], [565, 123], [517, 123]] confidence=0.9942460813641709
  word=315 bbox=[[78, 126], [136, 126], [136, 156], [78, 156]] confidence=0.9999935304043551
  word=309 bbox=[[514, 124], [574, 124], [574, 156], [514, 156]] confidence=0.9999759111084772
  word=I bbox=[[81, 175], [125, 175], [125, 211], [81, 211]] confidence=0.9417670402058889
  word=Yyan Rd bbox=[[226, 171], [414, 171], [414, 220], [226, 220]] confidence=0.6218364490642484
  word=E bbox=[[529, 173], [569, 173], [569, 213], [529, 213]] confidence=0.038355074509201836

Usage

Automatically downloads the onnx and prototxt files on the first run. It is necessary to be connected to the Internet while downloading.

For the sample image,

$ python3 easyocr.py

If you want to specify the input image, put the image path after the --input option.
You can use --savepath option to change the name of the output file to save.

$ python3 easyocr.py --input IMAGE_PATH
$ python3 easyocr.py -i IMAGE_PATH -s SAVE_IMAGE_PATH

By adding the --video option, you can input the video.
If you pass 0 as an argument to VIDEO_PATH, you can use the webcam input instead of the video file.

$ python3 easyocr.py --video VIDEO_PATH

By adding the --language option, you can choose the language.

$ python3 easyocr.py --language LANGUAGE
$ python3 easyocr.py -l LANGUAGE
(ex) $ python3 easyocr.py --language chinese
(ex) $ python3 easyocr.py -l chinese

The prepared language are as follows.

• chinese (default)
• japanese
• english
• french
• korean
• thai

(ex) $ python3 easyocr.py --language chinese --input example/chinese.jpg
(ex) $ python3 easyocr.py --language japanese --input example/japanese.jpg
(ex) $ python3 easyocr.py --language english --input example/english.png
(ex) $ python3 easyocr.py --language french --input example/french.jpg
(ex) $ python3 easyocr.py --language korean --input example/korean.png
(ex) $ python3 easyocr.py --language thai --input example/thai.jpg

Reference

Jaided AI: EasyOCR demo
Ready-to-use OCR with 80+ supported languages and all popular writing scripts including Latin, Chinese, Arabic, Devanagari, Cyrillic and etc.

Framework

Pytorch

Model Format

ONNX opset = 11

Netron

detector_craft.onnx.prototxt
recognizer_zh_sim_g2.onnx.prototxt
recognizer_japanese_g2.onnx.prototxt
recognizer_english_g2.onnx.prototxt
recognizer_latin_g2.onnx.prototxt
recognizer_korean_g2.onnx.prototxt
recognizer_thai.onnx.prototxt