I am trying to follow: https://github.com/jetsonhacks/installTensorFlowTX2
to install tensorflow on my TX2. After ./setTensorFlowEV.sh I get the following error:
Invalid path to cuDNN toolkit. Neither of the following two files can be found:
/usr/lib/aarch64-linux-gnu/lib64/libcudnn.so.6.0.21
/usr/lib/aarch64-linux-gnu/libcudnn.so.6.0.21
/usr/lib/aarch64-linux-gnu/libcudnn.so.6.0.21
This suggest I do not have cudnn6 installed on my TX2. Since tx2 is aarch64 and not x86 I am a bit stuck as nvidia only provides binary for x86 etc and not for aarch64. I understand I can flash my device with newest jetpack to get the cudnn.
Is there any other simpler way (without flashing my device) to install cudnn6 on tx2?
You can use JetPack to install cuDNN without flashing the device. Just
open JetPack, click next until you reach the screen showing all the available packages and set everything to no action
select cuDNN, set it to install and click next
A screen will show up asking you for the ip, username and password of your Jetson. Fill that out and click next
JetPack will now SSH into your Jetson and install cuDNN for you
Related
I've got a Codesys 3.5 project running on a CCPilot V700 display unit. I've got CodeSys runtime version 3.5.16 installed and it runs my Codesys project just fine.
However, when I want to add a CANbus, I click "Add device" --> Fieldbuses --> CANbus, which shows me a list of available CAN drivers. I can tick the box "Display all versions", which will allow me to see all installed CAN drivers. I'm out of luck: no version 3.5.16. I do have 3.5.15 and 3.5.17 though, but when I try those, I get the error: "a bus error has occurred", and the bus state is "NO DRIVER".
Looking at the Codesys website under the download section, I can't find any driver downloads. How can I install the right version of device drivers?
Those are installed with the codesys installation. So if you want see version 3.5.16 you need to install codesys 3.5.16.
I used the following tutorial to install macports and opencv:
http://www.jeffreythompson.org/blog/2012/09/21/installing-opencv-for-python-on-mac-lion/
It works in that after typing "python" I can call "import cv" without errors.
My first question: how do I get this to work with IDLE. I think the issue is via the terminal I'm using Python 2.7.5. IDLE is using Python 2.7.3. If this is the problem, what is the easiest way to fix this.
My second question: how do I get opencv to work in eclipse with pydev? I can't really find much helpful information. I have installed opencv via macports. But I can't get Eclipse to recognize that opencv has been installed. "Import cv" command says no module named cv exists.
I was struggling with this problem today. Here are the steps I followed using macports
Install macports and xcode
Follow Sam Khans post listed here to install python, relink it in your terminal and then install OpenCV:
http://samkhan13.wordpress.com/2012/06/18/using-opencv-with-python-on-your-mac-os-x/
NOTE: At this point you should be able to enter a terminal session and type "python" and then "import cv2" successfully.
Open eclipse and click on "eclipse" then "preferences" in the bar menu at the top of the program.
In the pop-up that appears click the arrow next to "pydev" to expand the menu then click "Interpreter - Python"
Press the "new" button on the right of the window
In the pop-up type in the name for this interpreter (I did python27 for my python2.7 version)
Next link the python executable for the desired version under the macports install tree "/opt/local/bin/pythonX.X" where X.X is your version (i.e. 2.7) and click ok.
A list of the available python libraries should populate. Select only those under the macport install tree (so we don't confuse our interpreter and cross list with another python build)
Finally click OK (which should automatically apply the changes).
The window will compile the interpreter and VOILA Python, Numpy and OpenCV working together through pydev in eclipse.
Need to install non-PNP kernel device driver under Windows 7. Old reliable INSTDRV.EXE fails on StartService, claiming "2 -- system cannot find the file specified" although fully qualified path to driver SYS file is absolutely positively accurate. I realize Device Manager has an "install legacy driver" Action menu item, but that requires identifying a specific device, which I don't have. Why doesn't INSTDRV.EXE work under Windows 7? How can I install a non-PNP kernel device driver under Windows 7? Thanks.
As far as I know, on modern Windows you have to write .inf file. Read this short MSDN article.
There is a tool in WDK that helps installing drivers. Just copy it to your driver's folder and run it.
Update:
After submitting this answer I've found another way of installing non-pnp drivers. There is another tool, DevCon. It can be found among other WDK tools (Windows Kits\10\Tools\x64). Just copy DevCon.exe to your driver's folder and execute like this:
devcon.exe install pcscgate.inf Root\pcscgate
Root\pcscgate - device identifier that matches the one in [Model] section of .inf file (yes, you still need that file). DevCon creates an empty device node with this identifier and then installs driver.
I had a similar problem just now, finally solved it by setting Linker/Input/Ignore All Default Libraries to Yes (/NODEFAULTLIB).
I've been trying to get BeagleBone work with OpenCV and a capture camera, but I couldn't manage to do it so far. I use Logitech C270 webcam. Here is what I did so far:
I installed OpenCV binaries and tried my camera. Capture didn't work. I downloaded the latest version from Git and compiled it. Then it worked on my desktop.
Since Ångström have pre-installed OpenCV, I directly tried to connect the webcam. But I had "select timeout" errors and when I saved the captured image, it was all black.
I deleted OpenCV from Ångström (opkg remove opencv and opencv-dev). It said it worked, but libraries were still in /usr. So I deleted them manually and copied libraries that I cross-compiled in my host. But now, I cannot even connect to camera. Capture returns null again.
I checked opkg list-installed, still OpenCV 2.4 is listed there. However, libraries in /usr/lib and include are the ones that I copied. How can I fix this problem?
Note: Webcam is recognized by Ångström, and dmesg shows successful connection to the webcam.
It seems something was wrong on installation of Linux itself. I re-compiled the latest Ångström and installed it to BeagleBone. Everything worked fine as it should. It seems sometimes it's better to take the hard way.
I can't install the nvidia installation in cuda toolkit. I don't have a nvidia card. Does that means I can't use emgu.cv? I been on this all yesterday and emgu.cv is still giving me problems.
Here the link that still shows my problem:
Exception initializing Emgu.CV
Actually you can. Run the installer. Then go to the directory where it has unzipped all the files and choose nvcuda.dl_ and using command prompt expand it into nvcuda.dll.
Place this dll in system32 folder.
This should solve the 'nvcuda.dll not found' error.