Toybrick

标题: onnx转rknn在build rknn时一直报错 [打印本页]

作者: meter_meter 时间: 2020-11-12 11:44
标题: onnx转rknn在build rknn时一直报错

目前我们利用百度paddlex开源算法训练了模型，算法训练好后，我们利用paddle2onnx包进行了模型转换，转为onnx格式。目前在由onnx格式转为rknn格式的过程中一直报错，报错信息如下：

T Traceback (most recent call last):

T File "rknn/api/rknn_base.py", line 344, in rknn.api.rknn_base.RKNNBase.load_onnx

T File "rknn/base/RKNNlib/converter/convert_onnx.py", line 523, in rknn.base.RKNNlib.converter.convert_onnx.convert_onnx.__init__

T File "rknn/base/RKNNlib/converter/convert_onnx.py", line 529, in rknn.base.RKNNlib.converter.convert_onnx.convert_onnx.__init__

T File "rknn/base/RKNNlib/converter/onnx_util.py", line 158, in rknn.base.RKNNlib.converter.onnx_util.ONNXProto_Util.polishing_model

T File "/Users/song.li/anaconda3/envs/pythonLearn/lib/python3.6/site-packages/onnx/utils.py", line 18, in polish_model

T onnx.checker.check_model(model)

T File "/Users/song.li/anaconda3/envs/pythonLearn/lib/python3.6/site-packages/onnx/checker.py", line 86, in check_model

T C.check_model(model.SerializeToString())

T onnx.onnx_cpp2py_export.checker.ValidationError: Your model ir_version is higher than the checker's.

done

--> Building model

Traceback (most recent call last):

File "rknn_build.py", line 35, in <module>

rknn.build(do_quantization=False)

File "/Users/song.li/anaconda3/envs/pythonLearn/lib/python3.6/site-packages/rknn/api/rknn.py", line 163, in build

inputs = self.rknn_base.net.get_input_layers()

AttributeError: 'NoneType' object has no attribute 'get_input_layers'

我们目前使用的版本信息如下：

tensorflow 1.14.0,

rknn_toolkit-1.2.0 ( rknn_toolkit-1.2.0-cp36-cp36m-macosx_10_9_x86_64.whl )

paddle2onnx 0. 4

请教相应的解决方法。谢谢！

欢迎光临 Toybrick (https://t.rock-chips.com/)