I used the Intel FPGA server with Arria 10 to run the smart classroom demo, and it went like this:
InferenceEngine:
API version ............ 1.2
Build .................. 13911
[ INFO ] Parsing input parameters
Reading video '/root/classroom.mp4'
Loading plugin HETERO:FPGA
API version ............ 1.2
Build .................. heteroPlugin
Description ....... heteroPlugin
Loading plugin HETERO
API version ............ 1.2
Build .................. heteroPlugin
Description ....... heteroPlugin
[ ERROR ] Error reading network: cannot parse future versions: 5
I ran the sample with this script:
./smart_classroom_sample \
-d_act HETERO \
-d_fd HETERO \
-d_reid HETERO \
-d_lm HETERO \
-m_act ~/model/person-detection-action-recognition-0005/person-detection-action-recognition-0005.xml \
-m_fd ~/model/face-reidentification-retail-0095/face-reidentification-retail-0095.xml \
-m_reid ~/model/face-reidentification-retail-0095/face-reidentification-retail-0095.xml \
-m_lm ~/model/landmarks-regression-retail-0009/landmarks-regression-retail-0009.xml \
-fg ~/faces_gallery.json \
-i ~/classroom.mp4
Please tell me how to run the sample correctly, thanks!
Link Copied
Dear Hiiragi Iori,
This error:
[ ERROR ] Error reading network: cannot parse future versions: 5
Means that you are mixing and matching IR (Model Optimizer generated IR) versions and Inference Engine versions. They should be the same. Where are you getting your models from ? Are you getting them from here ?
To avoid this error, please use IR (whether generated by MO or Intel models IR) and Inference Engine from the same release package.
Hope it helps,
Thanks,
Shubha
Dear Shubha,
Thanks for your reply. It helps a lot. I'll try it again.
Thanks,
Hiiragi Iori
Dear Hiiragi Iori,
Sure ! Hope you get it working.
Thanks,
Shubha
For more complete information about compiler optimizations, see our Optimization Notice.