Can't use cmake to build my cuda program

I am building a CUDA program which has some use cases of Opencv in it. I decided to use cmake for building my program.

The following is my cmake file :-

cmake_minimum_required(VERSION 3.8)

project(rgb_to_gray LANGUAGES CXX CUDA)

find_package(OpenCV 4 REQUIRED)

find_package(CUDA)

include_directories(${OpenCV_INCLUDE_DIRS})

add_library(cpu.cpp utils.cpp cpu.h utils.h)

add_executable(convert.o rgb_to_gray.cu)

target_link_libraries(convert.o ${OpenCV_LIBS})

But when I try to run it using this command :-
cmake -DCMAKE_CUDA_FLAGS="-arch=sm_60"

It throws this error :-

ERROR: No supported gcc/g++ host compiler found, but clang-8 is available.
           Use 'nvcc -ccbin clang-8' to use that instead.

My CUDA version 11.2
My GPU architecture is Pascal
Compute capability 6.1

As the error complains , do you have any gcc/g++ on your current PATH ?

Yes I do.