Comments (45)
@3073
First of all, i need to confirm to you the following.
1.Is NCSDK 2.05.xx.xx installed? "$ git clone -b ncsdk2 http://github.com/Movidius/ncsdk"
2.Is protobuf version 3.5.1? "$ protoc --version"
3.Is tensorflow version 1.9.0?
4.Is NCAPPZOO done git clone already? "$ git clone -b ncsdk2 https://github.com/movidius/ncappzoo.git"
NCAPPZOO is a sample program repository of NCS.
Experience, it will not work properly unless upgraded to tensorflow 1.8.0 or 1.9.0.
from mobilenet-ssd.
here is the confirmation for your question.
- I installed NCSDK 1.x " git clone https://github.com/movidius/ncsdk.git " . it is version 1
- when I write "protoc --version" on the terminal of RPI , it gives me 3.0.0
- tensorflow version is 1.7 it is installed by default when version 1 NCS SDK is installed
Thank you
from mobilenet-ssd.
@3073
OK.
The sample programs of NCSDK v1.0 and v2.0 are not completely compatible.
For your environment, please download the program for v1.0 first.
v1.0
https://github.com/movidius/ncappzoo.git
https://github.com/PINTO0309/MobileNet-SSD/tree/v1.0
https://github.com/PINTO0309/MobileNet-SSD-RealSense/tree/v1.0
v2.0
https://github.com/movidius/ncappzoo/tree/ncsdk2
https://github.com/PINTO0309/MobileNet-SSD.git
https://github.com/PINTO0309/MobileNet-SSD-RealSense
I recommend you read the README firmly.
All you have to do is insert the stick into the USB port and execute it.
A sample of the command to generate a graph file from the Caffe model is shown below.
$ mvNCCompile aaa.prototxt -w bbb.caffemodel -s 12
from mobilenet-ssd.
Thank you. as I understand the procedure is:
- I first install NCS SDK on my pi
- then Install NCS SDK ( the same version like what is installed on my pi) on my laptop
- generate graph file using the trained model weight and prototxt files (for caffe model) on my laptop
- move the graph file and other pyhton file to the RPI
- insert the Movidius on the USB port of the RPI and execute the code
- then have fun with the accelerated computing of Movidius
please correct me
thank you again
from mobilenet-ssd.
@3073
That's right!
All flows look correctly.
from mobilenet-ssd.
from mobilenet-ssd.
git clone https://github.com/movidius/ncsdk.git
does not work . I think the repository is closed temporarly.
Do you know any other link for cloning ncsdk ?
from mobilenet-ssd.
@3073
First of all, I recommend that you check the NCS Forum.
https://ncsforum.movidius.com/discussion/1053/ncsdk-on-github-is-going-opensource-current-software-releases-moved-temporarily
from mobilenet-ssd.
from mobilenet-ssd.
from mobilenet-ssd.
from mobilenet-ssd.
https://ncsforum.movidius.com/search?Search=make+examples
https://ncsforum.movidius.com/search?Search=virtualbox
https://ncsforum.movidius.com/discussion/948/procedure-for-make-examples-ncsdk-v2-05-00-02-raspberrypi3-raspbian-stretch-ubuntu16-04#latest
Is there one stick?
Are you using "USB Hub"?
Are you using Virtualbox?
from mobilenet-ssd.
from mobilenet-ssd.
@3073
OK.
Intel admits that using Virtualbox has problems.
Long ago, I succeeded in running Virtualbox with NCSDK 1.12.xx, but there are only articles in Japanese.
If you really want to try it with Virtualbox, please carefully read the English articles I presented earlier.
but when I try to connect the movidius to the RPI the detection speed is like the previous...
Are you using the program of my Github repository?
from mobilenet-ssd.
from mobilenet-ssd.
from mobilenet-ssd.
@3073
I am taking a nap and I got up now.
Japan is now 17: 00.
AttributeError: 'module' object has no attribute 'Enumerate_devices'
You Installed → NCSDK v2.05.xx
Probably, Your smaple program version → NCSDK v1.12.xx
Since NCSDK v2.05 has greatly changed API specifications, errors are displayed.
do you think I should install ubuntu along side windows instead of using
virtualbox for ubuntu installation will fix the problem?
It is not recommended to use Virtualbox so that it is unstable and slow in operation.
what about installing version two SDK ?
That is probably the correct answer.
from mobilenet-ssd.
@3073
Sorry.
You Installed → NCSDK v2.05.xx
Probably, Your smaple program version → NCSDK v1.12.xx
from mobilenet-ssd.
from mobilenet-ssd.
by the way how do you know whether a program is ncsdk version 1 or 2?
You need to see a part of the program.
It distinguis between uppercase and lowercase letters.
For example,
v2.xx.xx program
https://github.com/PINTO0309/MobileNet-SSD/blob/v2.0/MultiStickSSD.py
mvnc.global_set_option(mvnc.GlobalOption.RW_LOG_LEVEL, 2)
devices = mvnc.enumerate_devices()
v1.xx.xx program
https://github.com/PINTO0309/MobileNet-SSD/blob/v1.0/MultiStickSSD.py
mvnc.SetGlobalOption(mvnc.GlobalOption.LOG_LEVEL, 2)
devices = mvnc.EnumerateDevices()
do you mean mean if I installed version 1 NCSDK on my laptop to generate
graph file, that graph file is of the version 1. Therefore to use the graph
in Pi for detection with movidius, I should have install NCSDK version 1 on
my Pi because the graph is generated in a version 1 NCSDK installed pc?
Now, the program you are trying to run with RaspberryPi is compatible with NCSDK v1.
However, looking at the error message, RaspberryPi seems to have NCSDK v2 installed.
Recommended steps (1)
Your Ubuntu PC:
1. Uninstall NCSDK v1
2. Install NCSDK v2
3. Make graph file by NCSDK v2
4. Copy graph file to RaspberryPi
Your RaspberryPi:
1. Create a sample program for v2.
2. Sample Program execution.
Or below.
Recommended steps (2)
Your Ubuntu PC:
1. Make graph file by NCSDK v1
2. Copy graph file to RaspberryPi
Your RaspberryPi:
1. Uninstall NCSDK v2
2. Install NCSDK v1
3. Create a sample program for v1.
4. Sample Program execution.
from mobilenet-ssd.
from mobilenet-ssd.
from mobilenet-ssd.
Caffe, tesorflow will be installed by default in NCSDK v2?
Yes.
But there are notes.
Tesorflow 1.7.0 installed by default is very unstable.
I recommend that you uninstall it once and then install tensorflow 1.9.0.
$ sudo pip3 uninstall tensorflow
$ sudo apt install libatlas-base-dev
$ sudo pip3 install tensorflow
how to uninstall NCSDDK?
$ cd ~/ncsdk
$ make uninstall
from mobilenet-ssd.
from mobilenet-ssd.
from mobilenet-ssd.
@3073
I will reply to tomorrow.
Because it is 2 am now.
But please tell me the following.
- Will not the situation change when I insert and remove Stick once?
- Can you post the full text of "make examples"?
Good night...
from mobilenet-ssd.
from mobilenet-ssd.
from mobilenet-ssd.
@3073
It's a typical problem with Virtualbox.
The NCS Forum has reported that the behavior of the USB port is wrong.
Please try the following first.
- USB 1.1 Controller OHCI
- USB 2.0 Controller EHCI
- USB 3.0 Controller XHCI
https://ncsforum.movidius.com/discussion/comment/3023/#Comment_3023
Virtualbox is not really recommended.
Note that the main sequence of "make examples" is almost over, so you can safely ignore it as it is.
If you do not need a sample program.
from mobilenet-ssd.
from mobilenet-ssd.
normally I take the weight files from other repo to generate the graph. then I use my own program to execute based on that graph....I think it is normal.
from mobilenet-ssd.
AttributeError: 'module' object has no attribute 'EnumerateDevices' is still alive
You are running the NCSDK v1 sample program.
I do not know what program, but need to rewrite to v2 sample program.
Please take a closer look at Recommended steps (1).
It is correct that the error is displayed now.
If you absolutely want to run the v1 program, it is Recommended steps (2).
Or, a program that simulates the v1 program on v2 is published in ncappzoo, so please try it.
◆ncapi2_shim
Contains a shim python file that allows NCAPI v1 applications run with an installation of NCAPI v2.
To use the shim do the following to your existing NCAPI v1 program.
https://github.com/movidius/ncappzoo/tree/ncsdk2/ncapi2_shim
from mobilenet-ssd.
@3073
◆Migration
https://movidius.github.io/ncsdk/ncapi/python_api_migration.html
from mobilenet-ssd.
from mobilenet-ssd.
@3073
Because it is troublesome to explain, can you use this sample?
By the way, it works faster than "pyimagesearch" sample.
It is a sample of v2 and It works with usb camera.
RealSense is unnecessary.
https://github.com/PINTO0309/MobileNet-SSD-RealSense
◆USB Camera Mode
$ sudo raspi-config
"7.Advanced Options" - "A7 GL Driver" - "G3 Legacy"
$ cd ~/MobileNet-SSD-RealSense
$ python3 MultiStickSSDwithRealSense.py -mod 1 -wd 640 -ht 480
Please replace the graph file with the one you created with confidence.
from mobilenet-ssd.
from mobilenet-ssd.
@3073
The points are below.
Extension of the SWAP area is absolutely necessary.
It will hang up if not specified.
$ sudo nano /etc/dphys-swapfile
CONF_SWAPSIZE=100
↓
CONF_SWAPSIZE=2048
$ sudo /etc/init.d/dphys-swapfile restart swapon -s
And, "-j1" is necessary.
It will hang up if not specified.
$ make -j1
from mobilenet-ssd.
from mobilenet-ssd.
@3073
As I always say, can you trap errors with Try - Except etc?
What is the error message displayed?
from mobilenet-ssd.
from mobilenet-ssd.
@3073
Please post the full text of the following command.
$ sudo ls /dev/video*
from mobilenet-ssd.
from mobilenet-ssd.
@3073
I do not know what happened, but congratulations.
I think that I will solve earlier if you ask questions to Adrian of pyimagesearch.
I do not want to preach, originally, posts unrelated to the repository are violations of manners.
By the way, a girl 's baby was born on the last day of your last issue.
Please close "issue" if you calm down.
from mobilenet-ssd.
from mobilenet-ssd.
@3073
Good morning.
I am not angry, so do not worry.
Rather, I feel that your "issue" will be of help to other engineers.
Actually, I majored in economics.
Computer science is not specialized.
I just touch the device with my hobbies.
I think that I will solve earlier if you ask questions to Adrian of pyimagesearch.
I think that the sample program you had a problem was a program listed in "pyimage search".
Is not it wrong?
The blog administrator of "pyimage search" is a person named Adrian.
Also, it seems that a bulletin board is prepared for "pyimage search".
Regarding the program that Adrian himself created, I thought that asking Adrian questions would solve the problem quickly.
from mobilenet-ssd.
Related Issues (7)
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from mobilenet-ssd.