Paddle2ONNX icon indicating copy to clipboard operation
Paddle2ONNX copied to clipboard

Published 20 hours ago • verified publisher icon PaddlePaddle

paddledetection 训练的faster rcnn模型转换为onnx

Open gxqnba opened this issue 2 years ago • 11 comments

请问paddledetection 训练的faster rcnn模型能用paddle2onnx转换为onnx模型吗

gxqnba avatar May 15 '22 08:05 gxqnba

可以

jiangjiajun avatar May 16 '22 02:05 jiangjiajun

可以

谢谢！

gxqnba avatar May 16 '22 02:05 gxqnba

可以

可是我在转换faster rcnn r50 fpn训练的模型 遇到了问题， paddlepaddle 版本 2.2.2 paddledetection 2.4

输入命令： paddle2onnx --model_dir inference_model/faster_rcnn_r50_fpn_1x_coco
--model_filename model.pdmodel
--params_filename model.pdiparams
--save_file model.onnx --enable_onnx_checker True
--enable_dev_version True

报错如下： /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/onnx/mapping.py:27: DeprecationWarning: np.object is a deprecated alias for the builtin object. To silence this warning, use object by itself. Doing this will not modify any behavior and is safe. Deprecated in NumPy 1.20; for more details and guidance: https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations int(TensorProto.STRING): np.dtype(np.object) [Paddle2ONNX] Start to parse PaddlePaddle model(model file: inference_model/faster_rcnn_r50_fpn_1x_coco/model.pdmodel, parameters file: inference_model/faster_rcnn_r50_fpn_1x_coco/model.pdiparams [Paddle2ONNX] Model loaded, start to converting... [Paddle2ONNX] LodTensorArray is not supported. [Paddle2ONNX] LodTensorArray is not supported. [Paddle2ONNX] LodTensorArray is not supported. [Paddle2ONNX] Only support number of inputs equals to number of outputs for operator 'while'. [Paddle2ONNX] Oops, there are some operators not supported yet, including conditional_block,distribute_fpn_proposals,generate_proposals_v2,lod_array_length,select_input,tensor_array_to_tensor,while,write_to_array, [ERROR] Due to the unsupported operators, the conversion is aborted.

C++ Traceback (most recent call last):

No stack trace in paddle, may be caused by external reasons.

Error Message Summary:

FatalError: Process abort signal is detected by the operating system. [TimeInfo: *** Aborted at 1652668366 (unix time) try "date -d @1652668366" if you are using GNU date ***] [SignalInfo: *** SIGABRT (@0x3e8000006a1) received by PID 1697 (TID 0x7fa00f210700) from PID 1697 ***]

已放弃 (核心已转储)

gxqnba avatar May 16 '22 02:05 gxqnba

  1. 安装github上paddle2onnx的代码
  2. 转换时将--enable_dev_version设为False

PaddleDetection在export_model的时候，需在配置文件中添加export_onnx: True

jiangjiajun avatar May 16 '22 02:05 jiangjiajun

  1. 安装github上paddle2onnx的代码
  2. 转换时将--enable_dev_version设为False

PaddleDetection在export_model的时候，需在配置文件中添加export_onnx: True 我按照您的指导，进行了如下操作 1、安装github上paddle2onnx的代码 git clone https://github.com/PaddlePaddle/Paddle2ONNX.git cd Paddle2ONNX python setup.py install 最终安装的paddle2onnx版本为0.9.3

  1. 转换时将--enable_dev_version设为False paddle2onnx --model_dir ./
    --model_filename model.pdmodel
    --params_filename model.pdiparams
    --save_file model.onnx --enable_onnx_checker True
    --enable_dev_version False --opset_version 16 3、PaddleDetection在export_model的时候，需在配置文件中添加export_onnx: True image

最终转换模型时，输出如下： /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/onnx/mapping.py:27: DeprecationWarning: np.object is a deprecated alias for the builtin object. To silence this warning, use object by itself. Doing this will not modify any behavior and is safe. Deprecated in NumPy 1.20; for more details and guidance: https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations int(TensorProto.STRING): np.dtype(np.object) 2022-05-16 11:44:42 [WARNING] [Paddle2ONNX][WARN] This package didn't compile with deploykit, refer https://github.com/PaddlePaddle/Paddle2ONNX/tree/deploykit/docs/zh/deploykit.md for more details. 2022-05-16 11:44:42 [WARNING] [Paddle2ONNX][WARN] This package didn't compile with onnxruntime backend, refer https://github.com/PaddlePaddle/Paddle2ONNX/tree/deploykit/docs/zh/deploykit.md for more details. 2022-05-16 11:44:42 [WARNING] [Paddle2ONNX][WARN] This package didn't compile with TensorRT backend, refer https://github.com/PaddlePaddle/Paddle2ONNX/tree/deploykit/docs/zh/deploykit.md for more details. /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/layers/math_op_patch.py:341: UserWarning: /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle2onnx-0.9.3-py3.7.egg/paddle2onnx/op_mapper/custom_paddle_op/generate_proposals.py:107 The behavior of expression A - B has been unified with elementwise_sub(X, Y, axis=-1) from Paddle 2.0. If your code works well in the older versions but crashes in this version, try to use elementwise_sub(X, Y, axis=0) instead of A - B. This transitional warning will be dropped in the future. op_type, op_type, EXPRESSION_MAP[method_name])) /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/layers/math_op_patch.py:341: UserWarning: /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle2onnx-0.9.3-py3.7.egg/paddle2onnx/op_mapper/custom_paddle_op/generate_proposals.py:108 The behavior of expression A - B has been unified with elementwise_sub(X, Y, axis=-1) from Paddle 2.0. If your code works well in the older versions but crashes in this version, try to use elementwise_sub(X, Y, axis=0) instead of A - B. This transitional warning will be dropped in the future. op_type, op_type, EXPRESSION_MAP[method_name])) /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/layers/math_op_patch.py:341: UserWarning: /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle2onnx-0.9.3-py3.7.egg/paddle2onnx/op_mapper/custom_paddle_op/generate_proposals.py:109 The behavior of expression A + B has been unified with elementwise_add(X, Y, axis=-1) from Paddle 2.0. If your code works well in the older versions but crashes in this version, try to use elementwise_add(X, Y, axis=0) instead of A + B. This transitional warning will be dropped in the future. op_type, op_type, EXPRESSION_MAP[method_name])) /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/layers/math_op_patch.py:341: UserWarning: /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle2onnx-0.9.3-py3.7.egg/paddle2onnx/op_mapper/custom_paddle_op/generate_proposals.py:110 The behavior of expression A + B has been unified with elementwise_add(X, Y, axis=-1) from Paddle 2.0. If your code works well in the older versions but crashes in this version, try to use elementwise_add(X, Y, axis=0) instead of A + B. This transitional warning will be dropped in the future. op_type, op_type, EXPRESSION_MAP[method_name])) /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/layers/math_op_patch.py:341: UserWarning: /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle2onnx-0.9.3-py3.7.egg/paddle2onnx/op_mapper/custom_paddle_op/generate_proposals.py:116 The behavior of expression A * B has been unified with elementwise_mul(X, Y, axis=-1) from Paddle 2.0. If your code works well in the older versions but crashes in this version, try to use elementwise_mul(X, Y, axis=0) instead of A * B. This transitional warning will be dropped in the future. op_type, op_type, EXPRESSION_MAP[method_name])) /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/layers/math_op_patch.py:341: UserWarning: /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle2onnx-0.9.3-py3.7.egg/paddle2onnx/op_mapper/custom_paddle_op/generate_proposals.py:116 The behavior of expression A + B has been unified with elementwise_add(X, Y, axis=-1) from Paddle 2.0. If your code works well in the older versions but crashes in this version, try to use elementwise_add(X, Y, axis=0) instead of A + B. This transitional warning will be dropped in the future. op_type, op_type, EXPRESSION_MAP[method_name])) /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/layers/math_op_patch.py:341: UserWarning: /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle2onnx-0.9.3-py3.7.egg/paddle2onnx/op_mapper/custom_paddle_op/generate_proposals.py:117 The behavior of expression A * B has been unified with elementwise_mul(X, Y, axis=-1) from Paddle 2.0. If your code works well in the older versions but crashes in this version, try to use elementwise_mul(X, Y, axis=0) instead of A * B. This transitional warning will be dropped in the future. op_type, op_type, EXPRESSION_MAP[method_name])) /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/layers/math_op_patch.py:341: UserWarning: /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle2onnx-0.9.3-py3.7.egg/paddle2onnx/op_mapper/custom_paddle_op/generate_proposals.py:117 The behavior of expression A + B has been unified with elementwise_add(X, Y, axis=-1) from Paddle 2.0. If your code works well in the older versions but crashes in this version, try to use elementwise_add(X, Y, axis=0) instead of A + B. This transitional warning will be dropped in the future. op_type, op_type, EXPRESSION_MAP[method_name])) /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/layers/math_op_patch.py:341: UserWarning: /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle2onnx-0.9.3-py3.7.egg/paddle2onnx/op_mapper/custom_paddle_op/generate_proposals.py:120 The behavior of expression A * B has been unified with elementwise_mul(X, Y, axis=-1) from Paddle 2.0. If your code works well in the older versions but crashes in this version, try to use elementwise_mul(X, Y, axis=0) instead of A * B. This transitional warning will be dropped in the future. op_type, op_type, EXPRESSION_MAP[method_name])) /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/layers/math_op_patch.py:341: UserWarning: /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle2onnx-0.9.3-py3.7.egg/paddle2onnx/op_mapper/custom_paddle_op/generate_proposals.py:123 The behavior of expression A * B has been unified with elementwise_mul(X, Y, axis=-1) from Paddle 2.0. If your code works well in the older versions but crashes in this version, try to use elementwise_mul(X, Y, axis=0) instead of A * B. This transitional warning will be dropped in the future. op_type, op_type, EXPRESSION_MAP[method_name])) /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/layers/math_op_patch.py:341: UserWarning: /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle2onnx-0.9.3-py3.7.egg/paddle2onnx/op_mapper/custom_paddle_op/generate_proposals.py:124 The behavior of expression A * B has been unified with elementwise_mul(X, Y, axis=-1) from Paddle 2.0. If your code works well in the older versions but crashes in this version, try to use elementwise_mul(X, Y, axis=0) instead of A * B. This transitional warning will be dropped in the future. op_type, op_type, EXPRESSION_MAP[method_name])) /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/layers/math_op_patch.py:341: UserWarning: /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle2onnx-0.9.3-py3.7.egg/paddle2onnx/op_mapper/custom_paddle_op/generate_proposals.py:126 The behavior of expression A - B has been unified with elementwise_sub(X, Y, axis=-1) from Paddle 2.0. If your code works well in the older versions but crashes in this version, try to use elementwise_sub(X, Y, axis=0) instead of A - B. This transitional warning will be dropped in the future. op_type, op_type, EXPRESSION_MAP[method_name])) /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/layers/math_op_patch.py:341: UserWarning: /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle2onnx-0.9.3-py3.7.egg/paddle2onnx/op_mapper/custom_paddle_op/generate_proposals.py:127 The behavior of expression A - B has been unified with elementwise_sub(X, Y, axis=-1) from Paddle 2.0. If your code works well in the older versions but crashes in this version, try to use elementwise_sub(X, Y, axis=0) instead of A - B. This transitional warning will be dropped in the future. op_type, op_type, EXPRESSION_MAP[method_name])) /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/layers/math_op_patch.py:341: UserWarning: /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle2onnx-0.9.3-py3.7.egg/paddle2onnx/op_mapper/custom_paddle_op/generate_proposals.py:128 The behavior of expression A + B has been unified with elementwise_add(X, Y, axis=-1) from Paddle 2.0. If your code works well in the older versions but crashes in this version, try to use elementwise_add(X, Y, axis=0) instead of A + B. This transitional warning will be dropped in the future. op_type, op_type, EXPRESSION_MAP[method_name])) /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/layers/math_op_patch.py:341: UserWarning: /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle2onnx-0.9.3-py3.7.egg/paddle2onnx/op_mapper/custom_paddle_op/generate_proposals.py:129 The behavior of expression A + B has been unified with elementwise_add(X, Y, axis=-1) from Paddle 2.0. If your code works well in the older versions but crashes in this version, try to use elementwise_add(X, Y, axis=0) instead of A + B. This transitional warning will be dropped in the future. op_type, op_type, EXPRESSION_MAP[method_name])) /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/layers/math_op_patch.py:341: UserWarning: /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle2onnx-0.9.3-py3.7.egg/paddle2onnx/op_mapper/custom_paddle_op/generate_proposals.py:72 The behavior of expression A - B has been unified with elementwise_sub(X, Y, axis=-1) from Paddle 2.0. If your code works well in the older versions but crashes in this version, try to use elementwise_sub(X, Y, axis=0) instead of A - B. This transitional warning will be dropped in the future. op_type, op_type, EXPRESSION_MAP[method_name])) /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/layers/math_op_patch.py:341: UserWarning: /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle2onnx-0.9.3-py3.7.egg/paddle2onnx/op_mapper/custom_paddle_op/generate_proposals.py:73 The behavior of expression A - B has been unified with elementwise_sub(X, Y, axis=-1) from Paddle 2.0. If your code works well in the older versions but crashes in this version, try to use elementwise_sub(X, Y, axis=0) instead of A - B. This transitional warning will be dropped in the future. op_type, op_type, EXPRESSION_MAP[method_name])) /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/framework.py:744: DeprecationWarning: np.bool is a deprecated alias for the builtin bool. To silence this warning, use bool by itself. Doing this will not modify any behavior and is safe. If you specifically wanted the numpy scalar type, use np.bool_ here. Deprecated in NumPy 1.20; for more details and guidance: https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations elif dtype == np.bool: /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/layers/math_op_patch.py:341: UserWarning: /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle2onnx-0.9.3-py3.7.egg/paddle2onnx/op_mapper/custom_paddle_op/distribute_fpn_proposals.py:38 The behavior of expression A - B has been unified with elementwise_sub(X, Y, axis=-1) from Paddle 2.0. If your code works well in the older versions but crashes in this version, try to use elementwise_sub(X, Y, axis=0) instead of A - B. This transitional warning will be dropped in the future. op_type, op_type, EXPRESSION_MAP[method_name])) /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/layers/math_op_patch.py:341: UserWarning: /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle2onnx-0.9.3-py3.7.egg/paddle2onnx/op_mapper/custom_paddle_op/distribute_fpn_proposals.py:39 The behavior of expression A - B has been unified with elementwise_sub(X, Y, axis=-1) from Paddle 2.0. If your code works well in the older versions but crashes in this version, try to use elementwise_sub(X, Y, axis=0) instead of A - B. This transitional warning will be dropped in the future. op_type, op_type, EXPRESSION_MAP[method_name])) /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle/fluid/layers/math_op_patch.py:341: UserWarning: /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/paddle2onnx-0.9.3-py3.7.egg/paddle2onnx/op_mapper/custom_paddle_op/distribute_fpn_proposals.py:40 The behavior of expression A * B has been unified with elementwise_mul(X, Y, axis=-1) from Paddle 2.0. If your code works well in the older versions but crashes in this version, try to use elementwise_mul(X, Y, axis=0) instead of A * B. This transitional warning will be dropped in the future. op_type, op_type, EXPRESSION_MAP[method_name])) /opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/onnx/helper.py:343: DeprecationWarning: Using or importing the ABCs from 'collections' instead of from 'collections.abc' is deprecated, and in 3.8 it will stop working is_iterable = isinstance(value, collections.Iterable) 2022-05-16 11:44:49 [WARNING] Due to the operator:multiclass_nms3, the converted ONNX model will only supports input[batch_size] == 1. 2022-05-16 11:44:52 [INFO] ONNX model generated is valid. 2022-05-16 11:44:52 [INFO] ONNX model saved in model.onnx

请问这是转换成功了吗？我看输出有些红色的提示 image image

gxqnba avatar May 16 '22 03:05 gxqnba

前三条提示忽略。 第四条提示表示导出的模型在部署预测时，只能支持单张图片的输入进行预测

jiangjiajun avatar May 16 '22 04:05 jiangjiajun

前三条提示忽略。 第四条提示表示导出的模型在部署预测时，只能支持单张图片的输入进行预测

我将导出的model.onnx放到netron中查看，其输入有三个，其中scale_factor是啥意思啊？还有输出有两个，concat_6.tmp_0 是[N,6]表示N个输出结果，每个结果包含4个坐标和两个种类概率吗？（我任务前景就两类） 那么tmp_42是啥意思啊？ image

gxqnba avatar May 16 '22 07:05 gxqnba

模型的输入输出含义请咨询PaddleDetection @gxqnba

jiangjiajun avatar May 16 '22 08:05 jiangjiajun

模型的输入输出含义请咨询PaddleDetection @gxqnba

好的，非常感谢！

gxqnba avatar May 16 '22 08:05 gxqnba

  1. 安装github上paddle2onnx的代码
  2. 转换时将--enable_dev_version设为False

PaddleDetection在export_model的时候，需在配置文件中添加export_onnx: True

请问一下转换时--enable_dev_version设为False， 会出现错误， 提示ModuleNotFoundError: No module named 'onnx', 这是什么问题 @jiangjiajun

zhjian831129 avatar Jan 04 '24 02:01 zhjian831129