Are you sure? You might find result SUCCESS for those .dll’s lower down in process monitor. It is more likely you don’t have the Nvidia CUDA dependencies.
I would check the contents of your E:\Users\usr1\anaconda3\Lib\site-packages\cv2\config.py which should add the paths to the folder containing opencv_world.dll and opencv_img_hash490.dll and the path to CUDA, see below for an example
import os BINARIES_PATHS = [ os.path.join('<PATH_TO_BUILD_DIR>/install', 'x64/vc17/bin') os.path.join(os.getenv('CUDA_PATH', 'C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v12.3'), 'bin') ] + BINARIES_PATHS
If your version has both entries then I would confirm that CUDA_PATH
is set to the version of CUDA you built against and not something else as can be the case when using anaconda , i.e.
echo %CUDA_PATH%