A Neural Compute Stick example program that shows how to determine the frames/inferences per second that you are achieving with your current configuration for a specified network and image set.
This directory contains a python example program that will run inferences and report the FPS
This code example requires that the following components are available:
- Movidius Neural Compute Sticks
- Movidius Neural Compute SDK
To run the example code against multiple networks do the following :
- Open a terminal and change directory to the benchmarkncs base directory
- Type the following command in the terminal: make run The included makefile will build the network graph files if needed as well as run the python program one time for each network
To run the program for a single network do the following:
- Open a terminal and change directory to the benchmarkncs base directory
- type the following command ./benchmarkncs.py
for example: ./benchmarkncs.py ../../caffe/GoogLeNet ../../data/images 224 224