Forums - TensorFlow ResNet50 failed to benchmark using SNPE 1.13.0

3 posts / 0 new
or Register
Last post
TensorFlow ResNet50 failed to benchmark using SNPE 1.13.0
qqcao.cs
Join Date: 15 Feb 17
Posts: 1
Posted: Tue, 2018-04-10 16:15
Top

ResNet 50 v1 from TensorFlow models page can be compiled to DLC file. But when running the benchmark, it gives the following error:

error_code=403; error_message=No output is defined. Network has empty outputs; error_component=Dl Network; line_no=317; thread_id=-384236380

 I also tried 

snpe-dlc-info -i  resnet_v1_50.frozen.dlc

and it outputs the following:

DLC info for data/snpe-1.13.0/benchmarks/models/resnet_v1_50.frozen.dlc

Model Version: N/A
------------------------------------------------------------------------------------------
| Id | Name    | Type | Inputs  | Outputs | Out Dims  | Parameters                       |
------------------------------------------------------------------------------------------
| 0  | input:0 | data | input:0 | input:0 | 224x224x3 | input_preprocessing: passthrough |
|    |         |      |         |         |           | input_type: default              |
Total parameters 0 (0 MB assuming single precision float)
Total MACs 0 (0%)

 

Converter command: snpe-tensorflow-to-dlc verbose=False out_node=['output'] allow_unconsumed_nodes=True model_version=None input_dim=[['input', '224,224,3']] in_type=None

This probably indicates the DLC conversion is not correct. Does anyone have any ideas? Thanks.

  • Up0
  • Down0
Re: TensorFlow ResNet50 failed to benchmark using SNPE 1.13.0
qttiot
Join Date: 4 Dec 18
Posts: 2
Posted: Tue, 2019-07-09 09:43
Top

Did you solved the problem? I'm having a similar issue.

 

Model Version: N/A
Model Copyright:N/A
--------------------------------------------------------------------------------------------------------------
| Id | Name           | Type | Inputs         | Outputs        | Out Dims | Parameters                       |
--------------------------------------------------------------------------------------------------------------
| 0  | lstm_1_input:0 | data | lstm_1_input:0 | lstm_1_input:0 | 1x1x1    | input_preprocessing: passthrough |
|    |                |      |                |                |          | input_type: default              |
Total parameters: 0 (0 MB assuming single precision float)
Total MACs per inference: 0 (0%)
Converter command: snpe-tensorflow-to-dlc verbose=True out_node=['lstm_1_input'] allow_unconsumed_nodes=True model_version=None input_dim=[['lstm_1_input', '1,1,1']] copyright_file=None in_type=None
DLC created with converter version: 1.27.1.382
--------------------------------------------------------------------------------------------------------------
 
  • Up0
  • Down0
Re: TensorFlow ResNet50 failed to benchmark using SNPE 1.13.0
alexdwu13
Join Date: 11 Oct 19
Posts: 2
Posted: Wed, 2019-11-06 16:23
Top

Anyone solve this issue? I'm running into the exact same 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.