4.10.0

OpenCV (Release and Debug builds including python bindings and docs) built with VS2022 against CUDA 12.5., Nvidia Video Codec SDK 12.2 and cuDNN 9.2.0.

Suitable for all devices of compute capability >= 5.0 with binary compatible code for devices of compute capability 5.0-9.0.

Nvidia GPU Computing Toolkit v12.5 and cuDNN 9.2.0 are required dependencies.

Note: This wheel relies on cuDNN being installed in the CUDA Toolkit directory. Therefore you can either

1. Download the cuDNN Tarball (not the installer) and extract its contents to your CUDA directory, or

2. Add the path to the bin folder inside the cuDNN installation directory to your PATH_TO_PYTHON_DIST/Lib/site-packages/cv2/config.py file. e.g.

   import os

   BINARIES_PATHS = [
   os.path.join('D:/build/opencv/install', 'x64/vc17/bin'),
   os.path.join(os.getenv('CUDA_PATH', 'C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v12.5'), 'bin')
   os.path.join('C:/Program Files/NVIDIA/CUDNN/v9.1/bin/12.4')
   ] + BINARIES_PATHS

CMAKE_ARGS

-DBUILD_opencv_world=ON
-DWITH_CUDA=ON
-DENABLE_CUDA_FIRST_CLASS_LANGUAGE=ON
-DCUDA_ARCH_BIN=5.0,5.2,6.0,6.1,7.0,7.5,8.0,8.6,8.9,9.0
-DCUDA_ARCH_PTX=9.0
-DINSTALL_TESTS=ON 
-DINSTALL_C_EXAMPLES=ON

OpenCV 4.9.0

OpenCV (Release and Debug builds including python bindings and docs) built with VS2022 against CUDA 12.3., Nvidia Video Codec SDK 12.1 and cuDNN 8.9.7.

Suitable for all devices of compute capability >= 5.0 with binary compatible code for devices of compute capability 5.0-9.0.

Nvidia GPU Computing Toolkit v12.3 and cuDNN 8.9.7 are required dependencies.

CMAKE_ARGS

-DBUILD_opencv_world=ON
-DWITH_CUDA=ON
-DENABLE_CUDA_FIRST_CLASS_LANGUAGE=ON
-DCUDA_ARCH_BIN=5.0,5.2,6.0,6.1,7.0,7.5,8.0,8.6,8.9,9.0
-DCUDA_ARCH_PTX=9.0
-DINSTALL_TESTS=ON 
-DINSTALL_C_EXAMPLES=ON
-DBUILD_DOCS=ON

4.8.0.20230804

OpenCV (Release with symbols build including python bindings and docs) built with VS2022 against CUDA 12.2., Nvidia Video Codec SDK 12.1 and cuDNN 8.9.3.

Suitable for all devices of compute capability >= 5.0 with binary compatible code for devices of compute capability 5.0-9.0.

Nvidia GPU Computing Toolkit v12.2 and cuDNN 8.9.3 are required dependencies.

CMAKE_ARGS

-DBUILD_opencv_world=ON
-DWITH_CUDA=ON
-DCUDA_ARCH_BIN=5.0,5.2,6.0,6.1,7.0,7.5,8.0,8.6,8.9,9.0
-DCUDA_ARCH_PTX=9.0
-DINSTALL_TESTS=ON 
-DINSTALL_C_EXAMPLES=ON
-DBUILD_WITH_DEBUG_INFO=ON
-DBUILD_DOCS=ON

4.7.0

OpenCV (Release/Debug builds with python bindings) built with VS2022 against CUDA 12.0, Nvidia Video Codec SDK 12.0 and cuDNN 8.7.0.

Suitable for all devices of compute capability >= 5.0 with binary compatible code for devices of compute capability 5.0-9.0.

This has a hard dependancy on Nvidia GPU Computing Toolkit v12.0 and also requires cuDNN 8.7.0 for accelerated inference when using the dnn module.

CMAKE_ARGS

-DBUILD_opencv_world=ON
-DWITH_CUDA=ON
-DCUDA_ARCH_BIN=5.0,5.2,6.0,6.1,7.0,7.5,8.0,8.6,8.9,9.0
-DCUDA_ARCH_PTX=9.0
-DCUDA_FAST_MATH=ON
-DWITH_CUBLAS=ON    
-DWITH_OPENGL=ON
-DINSTALL_TESTS=ON 
-DINSTALL_C_EXAMPLES=ON
-DBUILD_WITH_DEBUG_INFO=ON

4.6.0.20221106

OpenCV (Release/Debug builds) built with VS2022 against CUDA 11.8, Nvidia Video Codec SDK 11.1 and cuDNN 8.6.0.

Suitable for all devices of compute capability >= 3.5 with binary compatible code for devices of compute capability 3.5-8.6.

Nvidia GPU Computing Toolkit v11.8 is required for import cv2 to work and cuDNN 8.6.0 for accelerated inference when using the dnn module.

CMAKE_ARGS

-DBUILD_opencv_world=ON
-DWITH_CUDA=ON
-DCUDA_FAST_MATH=ON
-DWITH_CUBLAS=ON    
-DWITH_NVCUVID=ON
-DWITH_NVCUVENC=ON
-DWITH_OPENGL=ON
-DINSTALL_TESTS=ON 
-DINSTALL_C_EXAMPLES=ON
-DBUILD_WITH_DEBUG_INFO=ON

4.6.0.20221004

OpenCV (Release build) built against CUDA 11.7, Nvidia Video Codec SDK 11.1 and cuDNN 8.4.1.

Suitable for devices of compute capability 3.5-8.6.

Nvidia GPU Computing Toolkit v11.7 is required for import cv2 to work and cuDNN 8.4.1 for accelerated inference when using the dnn module.

CMAKE_ARGS

-DBUILD_opencv_world=ON
-DWITH_CUDA=ON
-DCUDA_FAST_MATH=ON
-DWITH_CUBLAS=ON    
-DWITH_NVCUVID=ON
-DWITH_OPENGL=ON
-DINSTALL_TESTS=ON 
-DINSTALL_C_EXAMPLES=ON
-DBUILD_WITH_DEBUG_INFO=ON

4.6.0.20220920

OpenCV static lib built against CUDA 11.7, Nvidia Video Codec SDK 11.1 and cuDNN 8.4.1.

Suitable for all devices of compute capability >= 3.5 with binary compatible code for devices of compute capability 3.5-8.6.

Nvidia GPU Computing Toolkit v11.7 is required for import cv2 to work and cuDNN 8.4.1 for accelerated inference when using the dnn module.

CMAKE_ARGS

-DBUILD_opencv_world=ON
-DBUILD_SHARED_LIBS=OFF
-DWITH_CUDA=ON
-DCUDA_FAST_MATH=ON
-DWITH_CUBLAS=ON
-DCUDA_ARCH_PTX=8.6
-DWITH_NVCUVID=ON

OpenCV 4.6.0+

OpenCV (including python bindings) built against CUDA 11.7, Nvidia Video Codec SDK 11.1 and cuDNN 8.4.1.

Suitable for all devices of compute capability >= 3.5 with binary compatible code for devices of compute capability 3.5-8.6.

Nvidia GPU Computing Toolkit v11.7 is required for import cv2 to work and cuDNN 8.4.1 for accelerated inference when using the dnn module.

CMAKE_ARGS

-DWITH_CUDA=ON
-DCUDA_FAST_MATH=ON
-DWITH_CUBLAS=ON
-DCUDA_ARCH_PTX=8.6
-DWITH_NVCUVID=ON

Windows

set CMAKE_ARGS="-DWITH_CUDA=ON -DCUDA_FAST_MATH=ON -DWITH_CUBLAS=ON -DCUDA_ARCH_PTX=8.6 -DWITH_NVCUVID=ON"