添加whl包的构建方式 #594
添加whl包的构建方式 #594
Conversation
mergepaddleocr
merge paddleocr
|
|
| package_dir={'paddleocr': ''}, | ||
| include_package_data=True, | ||
| entry_points={"console_scripts": ["paddleocr= paddleocr.paddleocr:main"]}, | ||
| version='0.0.3', |
There was a problem hiding this comment.
上面pip install写的是0.0.1，这块是0.0.3，这个是typo吗还是？
There was a problem hiding this comment.
这个我改一下，pypi上是已经提交到0.0.3了，本地也需要改为0.0.3
setup.py
Outdated
| 'Programming Language :: Python', 'Programming Language :: Python :: 2', | ||
| 'Programming Language :: Python :: 2.5', | ||
| 'Programming Language :: Python :: 2.6', | ||
| 'Programming Language :: Python :: 2.7', |
There was a problem hiding this comment.
这块确定都是支持的吗？py2的中文是否会有问题呢？要不测试一下？
|
嗯嗯去掉吧，repo也是只支持py3
在2020年08月23日 13:54，zhoujun 写道：
@WenmuZhou commented on this pull request.
In setup.py:
+ description='Awesome OCR toolkits based on PaddlePaddle （8.6M ultra-lightweight pre-trained model, support training and deployment among server, mobile, embeded and IoT devices',
+ long_description=readme(),
+ long_description_content_type='text/markdown',
+ author='Baidu PaddlePaddle',
+ url='https://github.com/PaddlePaddle/PaddleOCR',
+ download_url='https://github.com/PaddlePaddle/PaddleOCR.git',
+ keywords=[
+ 'ocr textdetection textrecognition paddleocr crnn east star-net rosetta ocrlite db chineseocr chinesetextdetection chinesetextrecognition'
+ ],
+ classifiers=[
+ 'Intended Audience :: Developers', 'Operating System :: OS Independent',
+ 'Natural Language :: Chinese (Simplified)',
+ 'Programming Language :: Python', 'Programming Language :: Python :: 2',
+ 'Programming Language :: Python :: 2.5',
+ 'Programming Language :: Python :: 2.6',
+ 'Programming Language :: Python :: 2.7',
要不去掉py2?
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub, or unsubscribe.
|
|
已完成修改 |
| long_description_content_type='text/markdown', | ||
| url='https://github.com/PaddlePaddle/PaddleOCR', | ||
| download_url='https://github.com/PaddlePaddle/PaddleOCR.git', | ||
| keywords=[ |
There was a problem hiding this comment.
应该需要的，毕竟whl包也属于repo的一部分，多加点相关的关键词被搜索到的几率大，可以导流
There was a problem hiding this comment.
OK，之前在repo增加的太多，在github上搜索ocr排名太靠后
paddleocr.py
Outdated
| } | ||
|
|
||
| SUPPORT_DET_MODEL = ['DB'] | ||
| SUPPORT_REC_MODEL = ['Rosetta', 'CRNN', 'STARNet', 'RARE'] |
There was a problem hiding this comment.
SUPPORT_REC_MODEL只支持CRNN，其他先不支持
paddleocr.py
Outdated
|
|
||
| class PaddleOCR(predict_system.TextSystem): | ||
| def __init__(self, | ||
| det_model_name='ch_det_mv3_db', |
There was a problem hiding this comment.
是否改成模型路径最好呢？这样通过路径配置，可以同时支持超轻量模型和通用模型
|
|
||
| pip安装 | ||
| ```bash | ||
| pip install paddleocr |
There was a problem hiding this comment.
如果PaddleOCR有更新，whl包是自动更新，还是需要手动更新呢？
There was a problem hiding this comment.
如果PaddleOCR有更新，whl包是自动更新，还是需要手动更新呢？
pypi上的需要我们自己构建了传上去，都是手动更新
There was a problem hiding this comment.
这块和下面的本地构建并安装的区别？
本地构建，用户可以自己打包自己的后处理进来。pip install paddleocr安装的是我们提前构建好传到pypi上的
doc/doc_ch/whl.md
Outdated
| * 检测+识别全流程 | ||
| ```python | ||
| from paddleocr import PaddleOCR, draw_ocr | ||
| ocr = PaddleOCR(model_storage_directory='./model') # need to run only once to load model into memory |
There was a problem hiding this comment.
最好把检测模型和识别模型的路径分开，这样便于后续配置
There was a problem hiding this comment.
也就是改成PaddleOCR(det_model_dir,rec_model_dir)这种形式吗
| ...... | ||
| ``` | ||
|
|
||
| * 单独执行检测 |
* add yolov4
whl包主要用于支持paddleocr的python预测,特点如下