Compile OpenCV 4.x with CUDA
Budget: ₹12,500 – ₹37,500 INR
I need a clean, reproducible build of OpenCV 4.x on my Linux machine, compiled from source with CUDA 11.x acceleration switched on. The goal is to accelerate only the core, everyday image-processing functions in cv::cuda—nothing extra from high-level, optional modules—so the build stays lean, fast, and easy to maintain.
You will
• configure CMake to link against the installed CUDA 11.x toolkit and the system’s standard GCC/glibc stack,
• enable CUDA in the core modules while explicitly turning off everything non-essential,
• compile and install the library to a custom prefix, and
• provide a concise build script or set of commands plus a short README so I can reproduce the process on another Linux box or inside a fresh Docker container.
A quick test program showing that a few basic cv::cuda operations run without falling back to the CPU will be the acceptance check.
You will
• configure CMake to link against the installed CUDA 11.x toolkit and the system’s standard GCC/glibc stack,
• enable CUDA in the core modules while explicitly turning off everything non-essential,
• compile and install the library to a custom prefix, and
• provide a concise build script or set of commands plus a short README so I can reproduce the process on another Linux box or inside a fresh Docker container.
A quick test program showing that a few basic cv::cuda operations run without falling back to the CPU will be the acceptance check.
Related categories:
Python
Linux
CUDA
C++ Programming
Software Development
Image Processing
OpenCV
Computer Vision