paddleocr

PaddleOCR : Awesome multilingual OCR toolkits based on PaddlePaddle

Input

(from https://github.com/PaddlePaddle/PaddleOCR/tree/dygraph/doc/imgs)

Output

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,

$ python paddleocr.py
(ex on CPU)  $ python paddleocr.py -e 0
(ex on BLAS) $ python paddleocr.py -e 1
(ex on GPU)  $ python paddleocr.py -e 2

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 paddleocr.py --input IMAGE_PATH --savepath SAVE_IMAGE_PATH�
$ python3 paddleocr.py -i IMAGE_PATH -s SAVE_IMAGE_PATH

By adding the --video option, you can input the video.

$ python3 paddleocr.py --video VIDEO_PATH --savepath SAVE_VIDEO_PATH
$ python3 paddleocr.py -v VIDEO_PATH -s SAVE_VIDEO_PATH
(ex) $ python3 paddleocr.py --video input.mp4 --savepath output.mp4

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

$ python3 paddleocr.py --lanuage LANGUAGE
$ python3 paddleocr.py -l LANGUAGE
(ex) $ python3 paddleocr.py --lanuage japanese

The prepared language specifications are as follows.

• japanese, jpn, jp (default)
• english, eng, en
• chinese, chi, ch
• german, ger, ge
• french, fre, fr
• korean, kor, ko

By adding the --case option, you can choose model size by case.

$ python3 paddleocr.py --case CASE
$ python3 paddleocr.py -c CASE
(ex) $ python3 paddleocr.py --case server

The prepared language specifications are as follows.

• mobile (default)
• server (only Japanese and Chinese have server size models)

By adding the --det_limit_type and --det_limit_side_len option, You can limit the size of the input image for text detection. --det_limit_type is max or min and --det_limit_side_len is a positive integer, generally set to a multiple of 32, such as 960. The default setting of the parameters is --det_limit_type='max', --det_limit_side_len=1920. Indicates that the longest side of the network input image cannot exceed 1920, If this value is exceeded, the image will be resized with the same width ratio to ensure that the longest side is det_limit_side_len. Set as --det_limit_type='min', --det_limit_side_len=960 it means that the shortest side of the image is limited to 960.

$ python3 paddleocr.py -lt DET_LIMIT_TYPE -ll DET_LIMIT_SIDE_LEN
$ python3 paddleocr.py  --det_limit_type DET_LIMIT_TYPE --det_limit_side_len DET_LIMIT_SIDE_LEN
(ex) $ python3 paddleocr.py -lt max -ll 1920
(ex) $ python3 paddleocr.py  --det_limit_type min --det_limit_side_len 960

The prepared language specifications are as follows.

• mobile (default)
• server (only Japanese and Chinese have server size models)

About font file

Requires Japanese font. The default specification for each OS is as follows.

    if sys.platform == "win32":
        # Windows
        dc['vis_font_path'] = 'C:/windows/Fonts/meiryo.ttc'
    elif sys.platform == "darwin":
        # Mac OS
        dc['vis_font_path'] = '/System/Library/Fonts/ヒラギノ丸ゴ ProN W4.ttc'
    else:
        # Linux
        dc['vis_font_path'] = '/usr/share/fonts/opentype/ipaexfont-gothic/ipaexg.ttf'

If the Japanese font of the above path does not exist, please prepare it as described or change the specification in the program. In case of linux, you can get the target font by executing the following command.

apt install fonts-ipaexfont

Reference

PaddleOCR : Awesome multilingual OCR toolkits based on PaddlePaddle

PaddleOCR2Pytorch : PaddleOCR推理的pytorch实现和模型转换

Research of PaddleOCR

Framework

Pytorch

Model Format

ONNX opset = 10

Netron

ch_ppocr_server_v2.0_det_train.onnx.prototxt ch_ppocr_mobile_v2.0_cls_train.onnx.prototxt japan_mobile_v2.0_rec_infer.onnx.prototxt