OpenCV python wheels built against CUDA 13.0, Nvidia Video Codec SDK 13.0 and cuDNN 9.12.0.
Suitable for all devices of compute capability >= 7.5 with binary compatible code for devices of compute capability 7.5-12.1.
Nvidia GPU Computing Toolkit v13.0 is required for import cv2 to work and cuDNN 9.12.0 for accelerated inference when using the dnn module.
Nvidia CMake configuration output
--   NVIDIA CUDA:                   YES (ver 13.0, CUFFT CUBLAS NVCUVID NVCUVENC)
--     NVIDIA GPU arch:             75 80 86 89 90 100 103 120 121
--     NVIDIA PTX archs:            121
--
--   cuDNN:                         YES (ver 9.12.0)
Build Summary
Ubuntu 22.04 Build commands
export "CMAKE_ARGS=-DWITH_CUDA=ON -DCUDA_ARCH_BIN=7.5;8.0;8.6;8.9;9.0;10.0;10.3;12.0;12.1 -DCUDA_ARCH_PTX=12.1"
export ENABLE_CONTRIB=1
export ENABLE_ROLLING=1
python setup.py bdist_wheel --py-limited-api=cp37