topic Re: Running tinyYolo on Intel NCS error in Intel® Distribution of OpenVINO™ Toolkit

Running tinyYolo on Intel NCS error

idata — Mon, 15 Jan 2018 17:17:06 GMT

Hi everyone !

I am an undergrad student work who recently started out with intel NCS. I am trying to run TinyYolo on pi but an error as follows shows. I am also enclosing the paths (if that helps in any way)

pi@raspberrypi:~/workspace/ncappzoo/caffe/TinyYolo $ python run.py

Running NCS Caffe TinyYolo example

Traceback (most recent call last):

File "run.py", line 302, in

sys.exit(main())

File "run.py", line 268, in main

with open(tiny_yolo_graph_file, mode='rb') as f:

IOError: [Errno 2] No such file or directory: './graph'

Any advice will be highly appreciated. Sorry if this is a very amateur problem.

Ali

Re: Running tinyYolo on Intel NCS error

idata — Mon, 15 Jan 2018 18:12:40 GMT

Have you compiled ncappzoo after cloning from git? The problem is that graph file cannot be found.

You can see an example here:

https://software.intel.com/en-us/articles/build-an-image-classifier-in-5-steps-on-the-intel-movidius-neural-compute-stick

Re: Running tinyYolo on Intel NCS error

idata — Mon, 15 Jan 2018 20:30:10 GMT

I did but this shows up:

**Traceback (most recent call last):

File "convert_mean.py", line 3, in

import caffe

ImportError: No module named 'caffe'

Makefile:5: recipe for target 'all' failed

make[2]: *** [all] Error 1

make[2]: Leaving directory '/home/pi/workspace/ncappzoo/data/age_gender'

Makefile:17: recipe for target 'prereqs' failed

make[1]: *** [prereqs] Error 2

make[1]: Leaving directory '/home/pi/workspace/ncappzoo/caffe/GenderNet'

Makefile:11: recipe for target 'gendernet' failed

make: *** [gendernet] Error 2**

Please bear in mind that I am running this on raspberry pi 3

Re: Running tinyYolo on Intel NCS error

idata — Mon, 15 Jan 2018 20:59:05 GMT

You are doing this on a raspberry? If you don't have the whole SDK toolit on the raspberry only the API you cannot compile and profile networks.

https://movidius.github.io/blog/ncs-apps-on-rpi/

First, you need a machine with Ubuntu 16 where you install the SDK toolkit, you can find the details in the ncsdk github readme.

After you have compiled all the examples on this Ubuntu machine you can run the compiled graph files on the NCS even with a raspberry using the API.

If you don't have a linux machine with Ubuntu, you can compile everything on a virtual machine and after that you can copy the graph files to the raspberry.

If you have installed the whole toolkit on the raspberry then you have some problem with the path to the installed caffe package.