Anyone get the NCSDK loaded on a TX2?

[idata]

It failed for me trying to get something. The TX2 runs 16.04 Unbuntu

So I was wondering if anyone succeeded at this.

[idata]

@chicagobob123 What error are you seeing? Can you try installing API only mode and see if that works? You won't have access to the toolkit (checking, profiling, compiling) but you may be able to run your apps using API only mode. Please visit https://movidius.github.io/blog/ncs-apps-on-rpi/ and follow steps 2 - 4.

API only mode would mean that you will have to compile your graph files on another machine and move them over.

[idata]

I will try to get to this tonight. I am reinstalling the OS from scratch to the put the latest realsense stuff for the D435 on to the TX2.

[idata]

TX1 is all right.

[idata]

what version of jetpack are you using?

[idata]

L4T 28.2

[idata]

Be curious of the steps you took. Did you do a standard install? That did not work for me.

I must admit I have not gotten back to this. I was trying to get the Intel D423 camera to work which is pretty involved as well. I want to get both of these up and running on the same box.

[idata]

It's creating problem for me to install required tensorflow version, I have tensorflow 1.8.0 GPU version anyway. Please help to get rid of it.

[idata]

There is no approved way of doing this. There's a youtube video on making it work which includes rebuilding the kernel. You would be better off buying a wintel board like the one from lattepanda.

[idata]

Thanks @chicagobob123 , However I have managed to install to the ncsdk on tx2. But while compiling the examples of example/tensorflow/inception_v3, I am getting error OSError: ./libmvnc.so: wrong ELF class: ELFCLASS32. I have shared the error in more detail in link below.

https://pastebin.com/fZcky73q

Any suggestion will be appreciated.

[idata]

As mentioned in the prerequisites for basic installation, ncsdk supports x86-64 with Ubuntu (64 bit) 16.04 Desktop and Raspberry Pi 3. By this, I want to point out that both the architectures are 64 bit. But above I mention error OSError: ./libmvnc.so: wrong ELF class: ELFCLASS32, which is because of library is compiled for 32 bit systems. How is this possible to get this error? If I am getting this error anyway, how I can fix this.

I am missing something or making bad logic, please let me know.