PaddleX/latest/pipeline_usage/tutorials/ocr_pipelines/PP-StructureV3 #3651
Replies: 20 comments 33 replies
-
您好,请问怎么在这个代码里面集成生成双层pdf呢 |
Beta Was this translation helpful? Give feedback.
-
缺少pipeline: PP-StructureV3 ,找了好多文档都没找到如何安装,请问在哪里安装呢? |
Beta Was this translation helpful? Give feedback.
-
这个文字方向检测模型效果很差,这个应该怎么解决呢 |
Beta Was this translation helpful? Give feedback.
-
PP-StructureV3产线运行报错,显示:Exception: The pipeline (PP-StructureV3) does not exist! |
Beta Was this translation helpful? Give feedback.
-
您好,请问最近是否有做相关的更新呢?记得半个月前跑过一次代码是没有问题的,这两天重新部署了一版好像一直不成功。以下是在飞桨的AIstudio里运行报的错。使用的是paddlepaddle3.0beta框架。模型下载都是正常的,在pipeline执行的时候就崩了。 Destroyed a op_result that is still in use by pd_op.while. The owner op type is: pd_op.full (at /paddle/paddle/pir/src/core/op_result_impl.cc:37)C++ Traceback (most recent call last):0 paddle_infer::Predictor::Predictor(paddle::AnalysisConfig const&) Error Message Summary:FatalError: Aborted (core dumped) |
Beta Was this translation helpful? Give feedback.
-
1.请问这个是可以商用的吗?我看文档有的地方说需要购买,我点击进入又到了模型产线部署的页面,里面又说的需要获取序列号,但是序列号又是完全免费的。所以对于是否能免费使用和商用很混乱。 |
Beta Was this translation helpful? Give feedback.
-
使用paddlex --serve发布成服务后,看接口文档里说对于超过10页的PDF文件,只有前10页的内容会被使用。如何才能对整篇pdf文档整体解析?还是说要自己先对pdf文件做切分再去解析? |
Beta Was this translation helpful? Give feedback.
-
最新PaddleX3.0.0rc版本,使用上面的示例代码报错:ValueError: (InvalidArgument) The size of inputs must be equal to 2. |
Beta Was this translation helpful? Give feedback.
-
高性能推理没有支持PP-StructureV3,可否添加 我试用image_classification是能跑通的,用PP-StructureV3显示找不到模型(而PP-StructureV3不添加use_hpip也能跑通)。 |
Beta Was this translation helpful? Give feedback.
-
图片可以正常运行,PDF报错。
|
Beta Was this translation helpful? Give feedback.
-
你的邮件我已收到,谢谢合作!
|
Beta Was this translation helpful? Give feedback.
-
您好,我的paddlepaddle=3.0.0, paddlex=3.0.0rc1, Traceback (most recent call last): |
Beta Was this translation helpful? Give feedback.
-
希望产线处理时,能在提取印章后,从图片中去除印章效果(比如按颜色去除)再执行其他文字识别。否则印章文字会和原本的文本产生识别冲突。 |
Beta Was this translation helpful? Give feedback.
-
你的邮件我已收到,谢谢合作!
|
Beta Was this translation helpful? Give feedback.
-
在cpu版本的docker里边运行,推理样例都跑不通,一直提示C++ Traceback (most recent call last):0 paddle::AnalysisPredictor::ZeroCopyRun(bool) Error Message Summary:FatalError: Segmentation fault (core dumped) 请问一下这个问题是内存不够导致的么?如果是的话产线能支持在cpu上最小的设备要求是什么?docker容器生成的时候需要增加什么设置么? |
Beta Was this translation helpful? Give feedback.
-
您好,请问版面区域检测模型可以支持单独部署吗?目前文档的说明是版面区域检测和OCR子产线是必选的。 |
Beta Was this translation helpful? Give feedback.
-
您好,我成功部署了PP-structureV3,关于版面检测模型PP-DocLayout_plus-L有两点疑问,请指教:
|
Beta Was this translation helpful? Give feedback.
-
您好,直接调用产线模型PPStructureV3( ) 最后识别结果包含后处理 Reading Order Recovering以及Layout Recovering,单独使用PP-DocLayout_plus-L模型并没有,请问阅读顺序有单独模型么 |
Beta Was this translation helpful? Give feedback.
-
大文件解析 visualize 设置为false ,解析返回的数据也造成缓冲区溢出 ,有什么参数可以控制只返回markdown字段吗 |
Beta Was this translation helpful? Give feedback.
-
@Bobholamovic 请教一下:我运行pp_structure_v3,显存占用通常在9-10G左右,可以长期稳定运行。 但我想把24G的显存用尽,同时运行两个pipeline,则运行一段时间后,就会报错退出,看上去是GPU的某种冲突,但指向性不明。 请问paddlex支持同一个GPU下运行多个pipeline,或有什么方式充分利用显存吗?多谢。 |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
PaddleX/latest/pipeline_usage/tutorials/ocr_pipelines/PP-StructureV3
https://paddlepaddle.github.io/PaddleX/latest/pipeline_usage/tutorials/ocr_pipelines/PP-StructureV3.html
Beta Was this translation helpful? Give feedback.
All reactions