Forums - Onnx 2 dlc ConvTranspose bugs

3 posts / 0 new
or Register
Last post
Onnx 2 dlc ConvTranspose bugs
yssongest
Join Date: 27 Mar 18
Posts: 4
Posted: Tue, 2018-05-08 20:56
Top

Hi, 

     There are so many bugs in snpe-onnx-to-dlc scripts, especially for ConvTranspose.

     1，‘’infer_output_shape‘’  should be ''infer_output_shapes'' in nn_translations.py, line 251.

    2,  Weight shape in deconvolution is CNHW,  so the bias length should be equal to weight.shape[1] rather than shape[0] . The mistake lies in nn_translations.py, line 213 : '' bias = numpy.zeros(weights.shape[0], dtype=numpy.float32)''

    3, Why not support output padding? How can I  get the right output_width and ouput_height? Through params directly?

   

  • Up0
  • Down0
Re: Onnx 2 dlc ConvTranspose bugs
yssongest
Join Date: 27 Mar 18
Posts: 4
Posted: Tue, 2018-05-08 21:07
Top

By the way, do you have a plan to support MXNet？

  • Up0
  • Down0
Re: Onnx 2 dlc ConvTranspose bugs
yssongest
Join Date: 27 Mar 18
Posts: 4
Posted: Tue, 2018-05-08 21:07
Top

By the way, do you have a plan to support MXNet？

  • Up0
  • Down0
or Register

Opinions expressed in the content posted here are the personal opinions of the original authors, and do not necessarily reflect those of Qualcomm Incorporated or its subsidiaries (“Qualcomm”). The content is provided for informational purposes only and is not meant to be an endorsement or representation by Qualcomm or any other party. This site may also provide links or references to non-Qualcomm sites and resources. Qualcomm makes no representations, warranties, or other commitments whatsoever about any non-Qualcomm sites or third-party resources that may be referenced, accessible from, or linked to this site.