Linux/TDA2: RVP Linux usecase for OpenCV

Hi Krishnakumar,
Are you facing any issues on running OpenCV on TDA2X RVP board ?

Hello Shiju,

Thank for the information,

We need to run standard VisionSDK OpenCV usecases on RVP board.

What is the best approach?

Regards,

KK

Hello,

While there are no usecases built for OpenCV out of the box, it is possible to add OpenCV to RVP usecases. The biggest difference between RVP and EVM in this regard is all capture on RVP is via FPD Link III. OpenCV usecases must be modified to use the multichannel lvds capture initialization instead of the single camera VIP initialization. Basically, look at a usecase that works with RVP and use the same camera setup initialization in the openCV usecases.

Hopefully that is all you need to change, but I have not tested these exact steps here.

Cheers,

Ben

Hello Ben,

Thanks for your inputs,

Considering that we have link-and-chains, is it sufficient that I set capture mode to be "Camera", so that data comes to my Algorithm module?
Alternatively, I have ethernet based input, which can receive YUV or RGB data. Capture of data is of less conern.

My concern is,
1. Do we need build OpenCV libraries? or the libraries built for EVM A15 core, can be used as it is for RVP A15 core.?
2. We need to change the .cfg, makefile, etc, so that libraries can be linked to algorithm to run the OpenCV example on RVP, in this context tutorial/guidance will be helpful.

Regards,
KK