Forums - deeplab-mobilenetv2 convert error with snpe-1.17.0

2 posts / 0 new
or Register
Last post
deeplab-mobilenetv2 convert error with snpe-1.17.0
f10.liu
Join Date: 22 Mar 18
Posts: 15
Posted: Wed, 2018-07-18 19:17
Top
snpe version:snpe-1.17.0 model:deeplabv3-mobilenetv2 from https://github.com/tensorflow/models/blob/master/research/deeplab/g3doc/...command:snpe-tensorflow-to-dlc --graph mobilenetv2/mobilenet_v2_1.0_224_frozen.pb --input_dim input 1,224,224,3 --dlc mobilenetv2.dlc --out MobilenetV2/Predictions/Softmax After run above command I got the error log as below.

 

 

~/snpe-1.17.0/lib/python/converters/tensorflow/layers/convolution.py:319: RuntimeWarning: error_code=803; error_message=Layer parameters combination is invalid in GPU Layer MobilenetV2/expanded_conv_7/depthwise/depthwise: input depth must be divisible by 4 * number of groups; error_component=GPU Runtime; line_no=78; thread_id=140101827204928

I wonder will you support random input depth? If you will, when? Thank you very much!
  • Up0
  • Down0
Re: deeplab-mobilenetv2 convert error with snpe-1.17.0
gesqdn-forum
Join Date: 4 Nov 18
Posts: 184
Posted: Thu, 2019-05-09 23:57
Top

Hi FangLiu,

The error is displayed is for just inform that the particular layers will not support with the GPU runtime. So after conversion, You will get the converted DLC file. For Running the application properly please set CPU fallback option which will helps you to run layers on CPU which are not supported by GPU or DSP.
Hope This will resolve your problem.

  • 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.