zl程序教程

您现在的位置是:首页 >  其它

当前栏目

MMOCR使用指南

使用指南
2023-06-13 09:12:44 时间

MMOCR是通用视觉框架OpenMMLab的光学字符识别器。

安装配置环境

MMOCR github主页:GitHub - open-mmlab/mmocr: OpenMMLab Text Detection, Recognition and Understanding Toolbox

pip install mmcv-full -f https://download.openmmlab.com/mmcv/dist/cu113/torch1.11.0/index.html
pip install mmdet -i https://pypi.tuna.tsinghua.edu.cn/simple
pip install lmdb
pip install shapely
pip install rapidfuzz
pip install lanms
pip install pyclipper
pip install scikit-image
pip install imgaug

验证是否安装成功代码

import torch, torchvision
import mmcv
from mmcv.ops import get_compiling_cuda_version, get_compiler_version
import mmdet
import mmocr
from mmocr.utils.ocr import MMOCR

mmocr = MMOCR(det=None, recog='SAR', device='cpu')
print('mmocr载入成功')

文本检测与文本提取

import torch, torchvision
import mmcv
from mmcv.ops import get_compiling_cuda_version, get_compiler_version
import mmdet
import mmocr
from mmocr.utils.ocr import MMOCR

# mmocr = MMOCR(det=None, recog='SAR', device='cpu')
# print('mmocr载入成功')

if __name__ == '__main__':

    detector = MMOCR(det='TextSnake', recog='SAR', device='cuda')
    result = detector.readtext('demo/demo_densetext_det.jpg', output='output/demo_densetext_det.jpg')

中文检测与提取

在mmocr主目录下新建文件夹/data/chineseocr/labels

进入该文件夹执行

wget http://download.openmmlab.com/mmocr/textrecog/sar/dict_printed_chinese_english_digits.txt
wget http://download.openmmlab.com/mmocr/data/font.TTF

下载字体和字库

import torch, torchvision
import mmcv
from mmcv.ops import get_compiling_cuda_version, get_compiler_version
import mmdet
import mmocr
from mmocr.utils.ocr import MMOCR

# mmocr = MMOCR(det=None, recog='SAR', device='cpu')
# print('mmocr载入成功')

if __name__ == '__main__':

    detector = MMOCR(det='TextSnake', recog='SAR_CN', device='cuda')
    result = detector.readtext('demo/demo_densetext_det.jpg', output='output/demo_densetext_det.jpg')