Not getting FPS as per the website in Movidius VPU

[jakesully]

For MobileNet_v2_1.0 (from openvino zoo) It is expected to get 91 fps but I am able to get only 70

converted to IR by 

sudo python3 open_model_zoo/tools/downloader/converter.py --name MobileNet_v2_1.0   --mo /opt/intel/openvino_2021/deployment_tools/model_optimizer/mo.py 

this command will give both FP16 and FP32
____________________________

python3 benchmark_app.py -i ~/Downloads/image_coco/ -m /opt/intel/openvino_2021/deployment_tools/public/mobilenet-v2-1.0-224/FP16/mobilenet-v2-1.0-224.xml -d MYRIAD 

NB: folder image_coco contains a single image of a person

____________________________
output was

Count: 4224 iterations
Duration: 60091.21 ms
Latency: 56.75 ms
Throughput: 70.29 FPS

why there is a drop of almost 20 frames?

[Zulkifli_Intel]

Hello Adarsh Sudarsan,

Can you share the link to the website that you mentioned earlier in this thread? And which Movidius device you used to run the inference (eg: NCS1, NCS2, or Myriad X)?

Here is my benchmark_app test result using the mobilenet-v2-1.0-224 model on NCS 2:

Count:     6908 iterations

Duration:  60045.27 ms

Latency:   34.64 ms

Throughput: 115.05 FPS

Sincerely,

Zulkifli

[Zulkifli_Intel]

Hello Adarsh Sudarsan,

This thread will no longer be monitored since we have provided a solution. If you need any additional information from Intel, please submit a new question.

Sincerely,

Zulkifli