Using movidius caffe?

[idata]

After asking my previous question I decided to check that movidius version of caffe using simple samples from caffe site. It seems that movidius caffe don't use movidius - only cpu:

ohad@ohad:/opt/movidius/ssd-caffe$ ./build/install/bin/caffe time -model examples/mnist/lenet_train_test.prototxt -iterations 10

I0106 17:11:05.843257 31674 caffe.cpp:347] Use CPU.I0106 17:11:05.844693 31674 net.cpp:322] The NetState phase (0) differed from the phase (1) specified by a rule in layer mnistI0106 17:11:05.844707 31674 net.cpp:322] The NetState phase (0) differed from the phase (1) specified by a rule in layer accuracyI0106 17:11:05.844805 31674 net.cpp:58] Initializing net from parameters: …I0106 17:11:06.636744 31674 caffe.cpp:412] Average Forward pass: 30.2072 ms.I0106 17:11:06.636747 31674 caffe.cpp:414] Average Backward pass: 37.2378 ms.I0106 17:11:06.636770 31674 caffe.cpp:416] Average Forward-Backward: 67.4 ms.I0106 17:11:06.636773 31674 caffe.cpp:418] Total Time: 674 ms.I0106 17:11:06.636776 31674 caffe.cpp:419] *** Benchmark ends ***

disconnecting movidius - caffe benchmark gives the same result. Is caffe there to use with movidius?

[idata]

@ohadcn To use the Intel NCS, you have to use the C/C++ and Python API that comes with the SDK. If you just use Caffe, it will default to processing on the CPU or GPU. Using the API that comes with the SDK, you can offload the inference processing to the NCS. More information on that here: https://movidius.github.io/ncsdk/Caffe.html and https://movidius.github.io/ncsdk/configure_network.html.