tensorflow make issue

idata · ‎01-19-2018

pi@rpi3:~ $ cat /etc/issue

Raspbian GNU/Linux 9 \n \l

pi@rpi3:~ $ uname -a

Linux rpi3 4.9.59-v7+ #1047 SMP Sun Oct 29 12:19:23 GMT 2017 armv7l GNU/Linux

mkdir -p ~/workspace

cd ~/workspace

git clone https://github.com/movidius/ncsdk.git

cd ~/workspace/ncsdk

make install

pi@rpi3:~/workspace/ncsdk/examples/apps/hello_ncs_py $ python3 ./hello_ncs.py

Hello NCS! Device opened normally.

Goodbye NCS! Device closed normally.

NCS device working.

(All is good up to this point. continuing on…)

cd ~/workspace/ncsdk

make examples

-- Installing: /usr/local/bin/opencv_version

-- Set runtime path of "/usr/local/bin/opencv_version" to "/usr/local/lib"

make examples starting.

(cd examples; make)

make[1]: Entering directory '/home/pi/workspace/ncsdk/examples'

make -C apps/.

make[2]: Entering directory '/home/pi/workspace/ncsdk/examples/apps'

make -C hello_ncs_py/.

make[3]: Entering directory '/home/pi/workspace/ncsdk/examples/apps/hello_ncs_py'

nothing to make, use 'make run' to run.

make[3]: Leaving directory '/home/pi/workspace/ncsdk/examples/apps/hello_ncs_py'

make -C multistick_cpp/.

make[3]: Entering directory '/home/pi/workspace/ncsdk/examples/apps/multistick_cpp'

get_ilsvrc_aux.sh 100%

caffe_ilsvrc12.tar.gz 100%

ilsvrc_2012_mean.npy 100%

deploy.prototxt 100%

bvlc_googlenet.caffemodel 100%

making compile

mvNCCompile -w bvlc_googlenet.caffemodel -s 12 deploy.prototxt

mvNCCompile v02.00, Copyright @ Movidius Ltd 2016

Layer inception_5b/1x1 forced to im2col_v2, because its output is used in concat

Layer inception_5b/pool_proj forced to im2col_v2, because its output is used in concat

/usr/local/bin/ncsdk/Controllers/FileIO.py:52: UserWarning: You are using a large type. Consider reducing your data sizes for best performance

"Consider reducing your data sizes for best performance\033[0m")

make[4]: Leaving directory '/home/pi/workspace/ncsdk/examples/caffe/GoogLeNet'

making squeezenet

making prereqs

LICENSE 100%

making prototxt

deploy.prototxt 100%

making caffemodel

squeezenet_v1.0.caffemodel 100%

making compile

making multistick_cpp

making hello_ncs_cpp

making prereqs

making prototxt

deploy.prototxt 100%

making profile

mvNCProfile deploy.prototxt -s 12

mvNCProfile v02.00, Copyright @ Movidius Ltd 2016

__ WARNING: using empty weights __

/usr/local/bin/ncsdk/Controllers/FileIO.py:52: UserWarning: You are using a large type. Consider reducing your data sizes for best performance

"Consider reducing your data sizes for best performance\033[0m")

making caffemodel

Downloading caffemodel file

bvlc_alexnet.caffemodel 100%

making check

mvNCCheck -w bvlc_alexnet.caffemodel -i ../../data/images/cat.jpg -s 12 -id 281 deploy.prototxt -M 110 -S 255 -metric top1

mvNCCheck v02.00, Copyright @ Movidius Ltd 2016

[Error 28] Caffe Error: MemoryError. Potential Cause: Available RAM not sufficient for Network to be loaded into Caffe

making compile

mvNCCompile -w bvlc_alexnet.caffemodel -s 12 deploy.prototxt

mvNCCompile v02.00, Copyright @ Movidius Ltd 2016

[Error 28] Caffe Error: MemoryError. Potential Cause: Available RAM not sufficient for Network to be loaded into Caffe

making cpp

making prereqs

making prototxt

making profile

mvNCProfile deploy.prototxt -s 12

mvNCProfile v02.00, Copyright @ Movidius Ltd 2016

__ WARNING: using empty weights __

/usr/local/bin/ncsdk/Controllers/FileIO.py:52: UserWarning: You are using a large type. Consider reducing your data sizes for best performance

making caffemodel

caffemodel file already exists

making check

Result: Validation Pass

making compile

making cpp

making prereqs

making prototxt

making profile

making check

making compile

mvNCCompile -w bvlc_googlenet.caffemodel -s 12 deploy.prototxt

mvNCCompile v02.00, Copyright @ Movidius Ltd 2016

making cpp

g++ cpp/run.cpp cpp/fp16.c -o cpp/run_cpp -lmvnc

make[3]: Leaving directory '/home/pi/workspace/ncsdk/examples/caffe/GoogLeNet'

make[2]: Leaving directory '/home/pi/workspace/ncsdk/examples/caffe'

make -C tensorflow/.

make[2]: Entering directory '/home/pi/workspace/ncsdk/examples/tensorflow'

make -C inception_v3/.

make[3]: Entering directory '/home/pi/workspace/ncsdk/examples/tensorflow/inception_v3'

test -f output/inception-v3.meta || ((wget http://download.tensorflow.org/models/inception_v3_2016_08_28.tar.gz && tar zxf inception_v3_2016_08_28.tar.gz && rm inception_v3_2016_08_28.tar.gz) && ./inception-v3.py)

--2018-01-19 10:18:25-- http://download.tensorflow.org/models/inception_v3_2016_08_28.tar.gz

Resolving download.tensorflow.org (download.tensorflow.org)… 216.58.216.208, 2607:f8b0:4009:809::2010

Connecting to download.tensorflow.org (download.tensorflow.org)|216.58.216.208|:80… connected.

HTTP request sent, awaiting response… 200 OK

Length: 100885009 (96M) [application/x-tar]

Saving to: ‘inception_v3_2016_08_28.tar.gz’

inception_v3_2016_08_28.tar.gz 100%[================================================================================================================================>] 96.21M 2.79MB/s in 68s

2018-01-19 10:19:33 (1.41 MB/s) - ‘inception_v3_2016_08_28.tar.gz’ saved [100885009/100885009]

Traceback (most recent call last):

File "./inception-v3.py", line 4, in

import tensorflow as tf

ImportError: No module named 'tensorflow'

Makefile:41: recipe for target 'weights' failed

make[3]: *** [weights] Error 1

make[3]: Leaving directory '/home/pi/workspace/ncsdk/examples/tensorflow/inception_v3'

Makefile:12: recipe for target 'inception_v3/.' failed

make[2]: *** [inception_v3/.] Error 2

make[2]: Leaving directory '/home/pi/workspace/ncsdk/examples/tensorflow'

Makefile:12: recipe for target 'tensorflow/.' failed

make[1]: *** [tensorflow/.] Error 2

make[1]: Leaving directory '/home/pi/workspace/ncsdk/examples'

Makefile:57: recipe for target 'examples' failed

make: *** [examples] Error 2

pi@rpi3:~/workspace/ncsdk $

So some ram issues (as it is on an RPI3), some warnings, but it seems to fail on the tensorflow build…

I was following the 'canned' process for a first install on a virgin RPI image.

any ideas?

just starting out with the movidius & tensorflow.

Thanks!!

idata · ‎01-19-2018

hmm…

ramana.rachakonda

November 2017 0

@jwmac the issue with TF on RPi is that we don’t (yet) have an elegant why to install TF on RPi. The version that is available now has some incompatibilities with our toolset which we are updating.

Long story short, our tools should work if you install RPi 1.3 using source install. Also, if you install and compile and create graph files on a compatible machine (say Ubuntu 16.04 on Virtual Box on your PC), you can copy the graph file over (we have examples for inception with a precompiled graph file on our github) and run your application on RPi.

We are working on making this more homogeneous by working with TF. Thanks for your patience.

idata · ‎06-25-2018

Did you fix this problem? I got the same error message