[Simple Question] When end-user will run the program which includes TensorRT


This is simple question about usability of TensorRT.
I never use TensorRT, It would be silly question.

IF I release the Windows program(c++) for image classification by using TensorRT library.

Does End-user need to install somethings like a TensorRT engine or CUDA platform?
I want end-user will use their own GPU for running the program that inferences the image.

Please give me the advice.

Kevin Kim