NCSDK 2.05 not working in Virtualbox 5.2 with Ubuntu 16.04.04

[idata]

The NCSDK 1.12 works fine in Ubuntu 16.04 with virtual box however when I upgraded it to NCSDK 2.05 I cannot even get the hello_ncs example to run. It continues to give me an NCS not found error and disconnects the device which requires me to unplug and plug back for it to be detected. I tried various settings for the USB in virtual box and under linux but nothing works. As soon as I downgrade back to 1.12 everything works, even for the multistick examples (I have 4 of them now).

Ubuntu has been updated and upgraded (just in case this was the issue). All of the examples in the NCSDK 2.05 will compile but not run.

Is there anything I can do to try and get the NCSDK 2.05 to work?

[idata]

Yeah it seems to be completely broken - see my comments here: https://ncsforum.movidius.com/discussion/886/neural-compute-sdk-version-2-05-has-been-released#latest

[idata]

Dear colleagues,

I have the same issue mentioned above, I tried to make run the video-objects app from ncappzoo v2:

--2018-07-01 21:41:28-- https://raw.githubusercontent.com/nealvis/media/master/traffic_vid/police_car_6095_shortened_960x540.mp4

Resolving raw.githubusercontent.com (raw.githubusercontent.com)… 151.101.92.133Connecting to raw.githubusercontent.com (raw.githubusercontent.com)|151.101.92.133|:443… connected.HTTP request sent, awaiting response… 416 Range Not SatisfiableThe file is already fully retrieved; nothing to do.

--2018-07-01 21:41:28-- https://raw.githubusercontent.com/nealvis/media/master/traffic_vid/scooters_5638_shortened_960x540.mp4

Resolving raw.githubusercontent.com (raw.githubusercontent.com)… 151.101.92.133Connecting to raw.githubusercontent.com (raw.githubusercontent.com)|151.101.92.133|:443… connected.HTTP request sent, awaiting response… 416 Range Not SatisfiableThe file is already fully retrieved; nothing to do.

making run_py

python3 ./video_objects.py

E: [ 0] dispatcherEventSend:738 Write failed -1

E: [ 0] dispatcherEventReceive:200 dispatcherEventReceive() Read failed -1

please see below the result of hello_ncs_py:

making run

python3 hello_ncs.py;Hello NCS! Device opened normally.E: [ 0] dispatcherEventReceive:200 dispatcherEventReceive() Read failed -1

Goodbye NCS! Device closed normally.

NCS device working.

and lsusb:

Bus 002 Device 001: ID 1d6b:0003 Linux Foundation 3.0 root hub

Bus 001 Device 022: ID 03e7:2150Bus 001 Device 019: ID 80ee:0021 VirtualBox USB TabletBus 001 Device 004: ID 0bda:5756 Realtek Semiconductor Corp. Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub

Thanks in advance for the support!

[idata]

I've just tested with another example the live-image-classifier, same issue: /usr/local/bin/ncsdk/Controllers/FileIO.py:65: UserWarning: You are using a large type. Consider reducing your data sizes for best performance Blob generated make[1]: Leaving directory '/home/felipe/ncsdk/ncappzoo/caffe/AgeNet' felipe@felipe-VirtualBox:~/ncsdk/ncappzoo/apps/live-image-classifier$ make run making gendernet (cd ../../caffe/GenderNet; make compile;) make[1]: Entering directory '/home/felipe/ncsdk/ncappzoo/caffe/GenderNet' make[2]: Entering directory '/home/felipe/ncsdk/ncappzoo/data/age_gender' make[2]: Leaving directory '/home/felipe/ncsdk/ncappzoo/data/age_gender' test -f deploy.prototxt || (wget -P . https://raw.githubusercontent.com/GilLevi/AgeGenderDeepLearning/master/gender_net_definitions/deploy.prototxt; awk 'NR <2 {print}' < deploy.prototxt > temp; cat input_shape.prototxt >> temp; awk 'NR > 6 {print}' < deploy.prototxt >> temp; mv temp deploy.prototxt) test -f gender_net.caffemodel || wget -P . -N https://github.com/GilLevi/AgeGenderDeepLearning/raw/master/models/gender_net.caffemodel mvNCCompile -w gender_net.caffemodel -s 12 deploy.prototxt /usr/local/bin/ncsdk/Controllers/Parsers/TensorFlowParser/Convolution.py:44: SyntaxWarning: assertion is always true, perhaps remove parentheses? assert(False, "Layer type not supported by Convolution: " + obj.type) mvNCCompile v02.00, Copyright @ Intel Corporation 2017 /usr/local/bin/ncsdk/Controllers/FileIO.py:65: UserWarning: You are using a large type. Consider reducing your data sizes for best performance Blob generated make[1]: Leaving directory '/home/felipe/ncsdk/ncappzoo/caffe/GenderNet' making AgeNet (cd ../../caffe/AgeNet; make compile;) make[1]: Entering directory '/home/felipe/ncsdk/ncappzoo/caffe/AgeNet' make[2]: Entering directory '/home/felipe/ncsdk/ncappzoo/data/age_gender' make[2]: Leaving directory '/home/felipe/ncsdk/ncappzoo/data/age_gender' test -f deploy.prototxt || (wget -P . https://raw.githubusercontent.com/GilLevi/AgeGenderDeepLearning/master/age_net_definitions/deploy.prototxt; awk 'NR <2 {print}' < deploy.prototxt > temp; cat input_shape.prototxt >> temp; awk 'NR > 6 {print}' < deploy.prototxt >> temp; mv temp deploy.prototxt) test -f age_net.caffemodel || wget -P . -N https://github.com/GilLevi/AgeGenderDeepLearning/raw/master/models/age_net.caffemodel mvNCCompile -w age_net.caffemodel -s 12 deploy.prototxt /usr/local/bin/ncsdk/Controllers/Parsers/TensorFlowParser/Convolution.py:44: SyntaxWarning: assertion is always true, perhaps remove parentheses? assert(False, "Layer type not supported by Convolution: " + obj.type) mvNCCompile v02.00, Copyright @ Intel Corporation 2017 /usr/local/bin/ncsdk/Controllers/FileIO.py:65: UserWarning: You are using a large type. Consider reducing your data sizes for best performance Blob generated make[1]: Leaving directory '/home/felipe/ncsdk/ncappzoo/caffe/AgeNet' Running live-image-classifier.py python3 live-image-classifier.py --graph ../../caffe/AgeNet/graph --labels ../../data/age_gender/age_categories.txt E: [ 0] dispatcherEventSend:738 Write failed -1 E: [ 0] dispatcherEventReceive:200 dispatcherEventReceive() Read failed -1 Please note that the program gets frozen at this point. Any support is welcome.

[idata]

I'm having issues too, can't get any of the apps to run-

Here is an example from classifier-GUI

ImportError: /usr/local/lib/python3.5/dist-packages/pygraphviz/_graphviz.cpython-35m-x86_64-linux-gnu.so: undefined symbol: AgundirectedMakefile:80: recipe for target 'compile' failedmake[1]: *** [compile] Error 1make[1]: Leaving directory '/home/rob/ncsdk/ncappzoo/caffe/GoogLeNet'Makefile:17: recipe for target 'googlenet' failedmake: *** [googlenet] Error 2

Here is an example from Gender_age

ImportError: /usr/local/lib/python3.5/dist-packages/pygraphviz/_graphviz.cpython-35m-x86_64-linux-gnu.so: undefined symbol: Agundirected

Makefile:34: recipe for target 'compile' failedmake[1]: *** [compile] Error 1make[1]: Leaving directory '/home/rob/ncsdk/ncappzoo/caffe/GenderNet'cp: cannot stat '../../caffe/GenderNet/graph': No such file or directoryMakefile:9: recipe for target 'gendernet' failedmake: *** [gendernet] Error 1rob@rob-VirtualBox:~/ncsdk/ncappzoo/apps/gender_age_lbp$

Help would be appreciated!

[idata]

@yonush @milosgajdos @fpartika @rfranks263 Thank you for reporting this issue. We are aware of the issue and we are looking into creating a fix at the moment.

[idata]

Whenver i am trying to attach the Movidius from USB menu in virtual Box i am getting below error

 

i have already made filter using below settings

 

[idata]

Update got it resolve just by chaning the Virtualbox version from 5.1.26 to 5.1.34 and updating the extension pack,

So did basicallyhttps://download.virtualbox.org/virtualbox/5.2.2/VirtualBox-5.2.2-119230-Win.exehttps://download.virtualbox.org/virtualbox/5.2.2/Oracle_VM_VirtualBox_Extension_Pack-5.2.2-119230.vbox-extpackand in gues install vbox -additions.

Add a new USB filter with Vendor ID: 03e7

The other filter fields can be blank.Add a new USB filter with Vendor ID: 040eThe other filter fields can be blank.Install the NCSDK

[idata]

Hi all,

I'm using virtualbox version 5.2.12 r122591, since this is above the mentioned by @alokprasad it seems to be a diferente issue.@Tome_at_Intel any News about the fix for our issue?Thanks in advance!

[idata]

meet the same error under VirtualBox 5.2.14 r123301

any help?

[idata]

@fpartika @renhanchi , Please try with 5.1.34 , i faced issue with later version of Virtual box ( 5.2 ) it didnt worked

[idata]

Hello,

I am on VirtualBox version 5.2.12. Linux Ubuntu 16.06 LTS. VM is running on MacOSX.

I experienced the same issue as initially posted, BUT in my case I can attach the USB via the Devices>USB settings menu. I am also able to run once the hello_ncs_cpp and the dongle gets detached after that.

:~/workspace/ncsdk/examples/apps/hello_ncs_cpp/cpp$ ./hello_ncs_cpp

Hello NCS! Device opened normally.Goodbye NCS! Device Closed normally.NCS device working.

~/workspace/ncsdk/examples/apps/hello_ncs_cpp/cpp$ ./hello_ncs_cpp

Error - Could not open NCS device. mvncStatus value: -2

If I want to run the application again, I need to manually attach the dongle again via the settings.

[idata]

@mickaeltoumi try using a usb 2.0 hub

[idata]

Hi renhanchi,

Thank you for the suggestion. I just tried, it is actually worst. The connection is unstable with USB2.0 and USB1.1. At least with the USB3 I can run the example once.

I have a pop-up message in VirtualBox saying: "Failed to attach the USB device Movidius VSC Loopback Device [0100] to the virtual machine Ubuntu 16.04 - Computer Vision".

The hello_ncs_cpp returns: "Error - Could not open NCS device. mvncStatus value: -2"

[idata]

Works under NCSDK V1 but not V2. For V2, running the python3 hello_ncs.py produces the following: Error - Could not open NCS device.

Received similar errors when executing: make examples. The VM >> Devices >> USB shows a connection to Intel Corporation VSC Loopback Device (0100) instead of the Movidius LTD. Movidius MA2x5x [00001]? V1 works ok through this point. Using V1, the USB Device its always set to the Movidius LTD. Movidius MA2x5x [00001]. There is something definitely something incompatible with V2 NCSDK software and the VM.

After removing the NCSDK V2 and reinstalling V1, everything works!

I set up the VM 5.1.34 based on the Intel instructions and their acknowledgment it works under windows 10.

Any suggestions?

[idata]

I am still waiting for a fix for this very same issue. V1 works just fine, going to V2 is just broken. None of the suggestions, settings, etc. make any difference. So currently I have to stick to V1 or have four very useless devices.

[idata]

I'm still awaiting an announcement of a fix as well. At the moment, it is shelfware.

[idata]

Hi @Tome_at_Intel,

Any news about the latest VM version fix?

Regards

[idata]

Hi everyone,

Please refer to my solution for this problem. I can't garantee it it the fix for everybody but it seems to correspond to most people's description of the problem. See this link: https://software.intel.com/en-us/forums/computer-vision/topic/798725

[idata]

@Tome_at_Intel is there any update on this issue….as i'm also stuck here. I had to switch to NCSDK 2 as i needed to work with fifo's. Any update/help is appreciated

[idata]

So awareness of the issue for more than 4 months and not yet even updated https://movidius.github.io/ncsdk/vm_config.html

Because there it clearly says:

• tested with 5.1.34 but other releases should workwhich is not so quite true as it seems to be known that other releases don't really work…

I guess it will be helpful to at least update the docs with the known limitations as most probably not all customers like to figure the limits out the hard way.