请问一下，支持PP-Structure中的关键信息提取吗？ #96
Replies: 2 comments 6 replies
-
|
不支持哈，这个仓库只是做表格提取的 |
Beta Was this translation helpful? Give feedback.
-
|
建议可以考虑轻量级llm模型实现任意情况的信息提取 |
Beta Was this translation helpful? Give feedback.
-
|
rapidTable是来源百度pp-structure系列的模型和算法，这边最开始是阿里读光的模型，后来又加了一些东西，两个仓库都是表格识别，可以试试哪个更适合你的场景 |
Beta Was this translation helpful? Give feedback.
-
|
WiredTableRecognition() 这里面默认用的是PP-OCRv4 ？ |
Beta Was this translation helpful? Give feedback.
-
|
嗯，用的rapidOCR转的onnx，但是输入参数和paddleOCR不一样，可以看RapidOCR的文档了解 |
Beta Was this translation helpful? Give feedback.
-
|
再问一下，怎么用GPU加速？ |
Beta Was this translation helpful? Give feedback.
-
|
在faq里面写了哈，主要还是OCR耗时明显，可以使用rapidOCR-paddle来gpu加速，表格模型的话因为不是静态输入,onnx加速效果不好，但是cpu推理有线100ms,无线500ms左右 |
Beta Was this translation helpful? Give feedback.
-
如题
Beta Was this translation helpful? Give feedback.
All reactions