OCR-Yolov5-SwinIR-SVTR

OCR(Korean)

Run demo

  • Requirement
!python3 -m pip install paddlepaddle-gpu
!pip install "paddleocr>=2.0.1"
!pip install --upgrade PyYAML --ignore-installed
!pip install -r requirements.txt
  • Run demo.py
!python3 demo.py --image_dir='/content/test_img/' --rec_algorithm='SVTR' --rec_image_shape='3,64,256' --rec_char_dict_path='./korean_dict.txt'
  • Result : The result will be saved in './results'

Reference