Hi Team,
We've built a vision application which does a bunch of preprocessing of the images before passing the processed image to the AI model for inferencing. We've explored RZV2L for deploying this application. However, we're stuck because there doesn't seem to be a way to utilize just the DRP part of the DRP-AI for acceleration of our image processing chain. We don't really care much about the inferencing part (we don't really need to have the AI part).
Can we run traditional image processing such as applying a "Median Filter" or doing "Canny Edge Detection" ONLY (and NOT AI) on RZV2L?From the available literature, it doesn't appear that there is any way that we can access and use just the "DRP" part of "DRP-AI". At the same time, none of the docs clearly indicate that we cannot access just the DRP part of the DRP-AI either.I'd like to use everything that is available on RZV2L, except the AI part. Is there any way to do that?
Your valuable insights and inputs / clarifications are valuable and much appreciated. Thank you!Kind Regards,Sameer
Hi Sameer,please have a look at the ISP support package that is available at the RZ/V2L website.It contains dedicated DRP libraries for image processing.As mentioned before, there's an alternative "OpenCV on DRP" package under construction.Please ask your local sales representative for details.Kind regards,Stefan