I downloaded the eIQ object recognition CPP sample application source from the following link,
Compiled the source using the imx8mqevk toolchain and tested the binary on the imx8mqevk target board it's running on the latest FSL image 5.4 kernel.
The app was running on the CPU memory only not on the GPU memory so the application response was very slow.
Is it possible to run the same app on the imx8 GPU side for better response time and speed?
Also, share some references for CPP based image processing sources on the GPU side.
The only instruction that manages the gpu are from OpenCV, and unfortunately the examples working as is. You can add to the GPU adding more instruction for the GPU.