I followed the installation procedure: https://developer.movidius.com/start
I started with fresh raspbian stretch and followed the steps. I was able to successfully install the NC SDK. However, it failed to test the installation.
Make installation is successful. But, the make examples is failing.
Error importing caffe
make: Leaving directory '/home/mc/workspace/ncsdk/examples/caffe/GoogLeNet'
cp: cannot stat '../../caffe/GoogLeNet/graph': No such file or directory
Makefile:9: recipe for target 'googlenet' failed
make: *** [googlenet] Error 1
make: Leaving directory '/home/mc/workspace/ncsdk/examples/apps/multistick_cpp'
Makefile:12: recipe for target 'multistick_cpp/.' failed
make: *** [multistick_cpp/.] Error 2
make: Leaving directory '/home/mc/workspace/ncsdk/examples/apps'
Makefile:12: recipe for target 'apps/.' failed
make: *** [apps/.] Error 2
make: Leaving directory '/home/mc/workspace/ncsdk/examples'
Makefile:57: recipe for target 'examples' failed
make: *** [examples] Error 2
it is super frustrating! NCSDK version 1 works on virtualbox and raspberry pi. But version 2 has so many problems. Virtual box is not able to detect the usb while installing the second (latest) version. And the raspberry pi is also throwing some errors while implementing the yolo model. Very immature device!
@jhurley @lucas.baiao Please go to one of the samples in the Caffe folder like AlexNet and do a
make run and post the output here. Make sure to open a new terminal window after installation to set the PYTHONPATH environment variable. Instead of using
make examples, I recommend
install-opencv.shscript and then building each example as you need using
make compileinside of each of the sample network folders.
The NCSDK 2 Tiny Yolo V2 project wont build in Raspian Stretch on the RPi because the makefile clones the Darkflow repo and downloads the cfg and bin file and then converts it to a Tensorflow PB file using Darkflow. This conversion isn't possible on the Pi at the moment because of the Darkflow incompatibility, but if you build the Tiny Yolo v2 project on a laptop running Ubuntu and then move the graph file over, the sample code can run on the RPi using the graph file.