I understand that, in general, OpenCV functions are not GPU-accelerated when called from Python. However, according to T-API python support implemented by PolarNick239 · Pull Request #6078 · opencv/opencv · GitHub the T-API should allow for GPU acceleration. Does Nvidia have any stated position on Python, OpenCV, and acceleration?
I am primarily interested in the Jetson TK1 and TX2 platforms and their use in robotics.