Forums - MobilenetSSD on DSP

2 posts / 0 new
or Register
Last post
MobilenetSSD on DSP
Pragnesh
Join Date: 21 Jul 12
Posts: 5
Posted: Fri, 2018-08-10 15:45
Top
SNPE version : 1.17.0
Model : Tensorflow MobilenetSSD
Hardware: Snapdragon 820 based board with DSP
 
My intention is to run MobilenetSSD object detection model on DSP. I have converted coco model into a DLC file. 
 
My code is based on "examples/NativeCpp/SampleCode/" application and I am using ITENSOR method to load input images. I have CPU fallback option enabled as mentioned in the documentation. My code is able to do classification using a DLC file converted from Inception_v3 model on both GPU and DSP runtimes. I am also able to do object detection using a DLC file converted from MobilenetSSD model on GPU, but I am facing issue while running it on DSP. I am able to run it on DSP without any errors but I am getting weird detection results, I see ~100 objects getting detected with their probability set to 100%.
 
Not sure if MobilenetSSD is supported on DSP. I do see less network execution time compared to GPU as expected.
 
Any help or pointers will be helpful.
 
Thanks,
Pragnesh
 
  • Up0
  • Down0
Re: MobilenetSSD on DSP
nickzhuy
Join Date: 6 Mar 17
Posts: 1
Posted: Fri, 2018-10-19 00:46
Top

Hi Pragnesh,

Do you fix the issue?

I have the same issue.

  • 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.