Forums - Convert （deeplabv3+）MobilenetV2 error

2 posts / 0 new
or Register
Last post
Convert （deeplabv3+）MobilenetV2 error
f10.liu
Join Date: 22 Mar 18
Posts: 14
Posted: Tue, 2018-03-27 02:54
Top

I downloaded the mobilenetv2 model from deeplab model zoo :

https://github.com/tensorflow/models/blob/master/research/deeplab/g3doc/...

And when I run the command

"snpe-tensorflow-to-dlc --graph mobilenetv2_deeplab/frozen_inference_graph.pb --input_dim sub_7 513,513,3 --dlc mobilenetv2_deeplab.dlc --out logits/semantic/BiasAdd"

 

I got the erre like below:

Quote:
layers/eltwise.py:164: RuntimeWarning: error_code=1004; error_message=Layer parameters combination is invalid. Layer MobilenetV2/expanded_conv_7/add: dimensions mismatch between MobilenetV2/expanded_conv_6/project/BatchNorm/FusedBatchNorm:0 {65, 65, 64} and MobilenetV2/expanded_conv_7/add {61, 61, 64}; error_component=Model Validation; line_no=117; thread_id=140619208206144

But I printed the shapes of  inputs of "MobilenetV2/expanded_conv_7/add", I got [1, 65, 65, 64] and [1, 65, 65, 64].

Someone can help me?

  • Up0
  • Down0
Re: Convert (deeplabv3+) MobilenetV2 error
mistry Moderator
Join Date: 18 Apr 18
Posts: 56
Posted: Fri, 2019-06-14 10:36
Top

Hi Fang,

The input dimensions provided for arguments are incorrect.
Please find the below command for right usage,

$ snpe-tensorflow-to-dlc --graph mobilenetv2_deeplab/frozen_inference_graph.pb --input_dim sub_7 1,513,513,3 --out_node ArgMax --dlc deeplabv3.dlc --allow_unconsumed_nodes

Hope this helps.

  • 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.