Forums - Tensorflow object detection api model failed to convert to dlc

2 posts / 0 new
or Register
Last post
Tensorflow object detection api model failed to convert to dlc
a.annaiyan
Join Date: 2 Jul 21
Posts: 3
Posted: Fri, 2022-02-04 00:36
Top

Hi everyone,

I have trained an object detection model (CenterNet with MobileNet) using the Tensorflow object detection api. While converting to dlc format using the savedmodel format I get the warning message as "not able to consume layers" and at the end the conversion fails with message "Model is empty".

snpe-tensorflow-to-dlc --input_network /path/to/saved_model/ --input_dim "input_tensor" "1,224,224,3" --out_node detection_boxes --out_node detection_scores --out_node detection_classes --output_path centerNet.dlc --show_unconsumed_nodes --debug

I have seen similar issues posted on this group and have tried multiple solution:

  1.  checked the name of the input node : In my case it is "input tensor"
  2. Some suggested to check tensorflow dependencies, revet to a previous commit of graph generation function in object detection api. I have tried this approach as well, but still stuck. 

My configuration:

  1. Tensorflow: 2.4
  2. Python: 3.8
  3. SNPE: 1.58.0_3160

Any help with this issue will be higly appreciated. 

kind regards

 

 

 

  • Up0
  • Down0
Re: Tensorflow object detection api model failed to convert...
weihuan
Join Date: 12 Apr 20
Posts: 270
Posted: Thu, 2022-02-10 20:02
Top

Dear customer,

The version of TF we're recommended is TF1.15. Please try to remove the opiton of --show_unconsumed_nodes and convert it again.

BR.

Wei

  • 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.