Proper T-API Tutorials

edit

Is there a complete tutorial on the proper way of using the T-API functions in OpenCV to make code that works without a glitch?

Rule of thumb such as function that syncs the GPU/CPU functions or things to avoid when creating a function. Information like that will be really helpful for beginners to get a hand of the T-API coding style.

Right now i am getting mixed result with the openCL T-API implementations. It works fairly well when series of simple Opencv function is implemented. However, when creating a more complex that combines various openCV functions, the T-API implementation doesn,t seem to help improve the speed. In fact, some functions even runs slower than CPU implementations.

Proper documentation on T-API usage should clear the air on whether I am using the method wrongly or there is something wrong with the T-API implementation.

Please do share blog or article that can help me with this.

Thank you.