PaddleOCR中文表格模型转onnx缺失算子支持

[746891300-org]

PaddleOCR新出的中文表格识别模型转onnx报错： NotImplementedError: There's 5 ops are not supported yet =========== tensor_array_to_tensor =========== =========== lod_array_length =========== =========== while =========== =========== write_to_array =========== =========== one_hot_v2 ===========

模型地址： https://paddleocr.bj.bcebos.com/ppstructure/models/slanet/ch_ppstructure_mobile_v2.0_SLANet_infer.tar

[yeliang2258]

你好，这个模型要转换成功，需要OCR模型去除控制流，建议去PaddleOCR repo下提issue，去除该模型的控制流之后再进行转换

[746891300-org]

好的，谢谢！

[mingo-doer]

@yeliang2258 你好 这个模型也是一样的问题吗？https://paddleocr.bj.bcebos.com/dygraph_v2.0/table/en_ppocr_mobile_v2.0_table_structure_infer.tar [Paddle2ONNX] Start to parse PaddlePaddle model... [Paddle2ONNX] Model file path: /root/ocr/inference/en_ppocr_mobile_v2.0_table_structure_infer/inference.pdmodel [Paddle2ONNX] Paramters file path: /root/ocr/inference/en_ppocr_mobile_v2.0_table_structure_infer/inference.pdiparams [Paddle2ONNX] Start to parsing Paddle model... [Paddle2ONNX] LodTensorArray is not supported. [Paddle2ONNX] Oops, there are some operators not supported yet, including conditional_block,lod_array_length,select_input,tensor_array_to_tensor,while,write_to_array, [ERROR] Due to the unsupported operators, the conversion is aborted. Aborted (core dumped)

[WenmuZhou]

fixed in https://github.com/PaddlePaddle/PaddleOCR/pull/7485, the inference model has also been updated, you can re-download it for paddle2onnx

[Zheng-Bicheng]

• https://github.com/PaddlePaddle/Paddle2ONNX/pull/1324

[Zheng-Bicheng]

已经更新支持 while 算子，等待 CI 结束后 安装 Paddle2ONNX v1.2.10 使用:

• https://github.com/PaddlePaddle/Paddle2ONNX/actions/runs/11362622694