when I convert SSD-mobilenet model(pre-trained model) to DLC
then run in Soc603 platform yocto Linux,but the detection result is very bad .
The same SSD-mobilenet model run in PC is ok.
Do anybody have similar experiance ?
when I convert SSD-mobilenet model(pre-trained model) to DLC
then run in Soc603 platform yocto Linux,but the detection result is very bad .
The same SSD-mobilenet model run in PC is ok.
Do anybody have similar experiance ?
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.
Hi,
Did you do Quantization to 8 bit on yocto linux? As knew, batch normailiztion and ReLU6 have quantization loss problem?
Also could you provide more infromation if any difference by your experiment between PC and yocto Linux, i.e., quantization?
Thanks.
Greetings,
yes I have the same bad results when run on SoC603.
Will very much appreciate any advice on how to improve mobilenetSSD.DLC execution accuracy/non-max suppression issue on SoC603.
Best,
Matt