Skip to content

Install

The install script in Tutorial 1 works on Linux, macOS, and Windows, detects the accelerator, and selects the matching feature flags. For manual installs, specific driver versions, or source builds, use the guides below.

SituationGuide
Linux with an NVIDIA GPULinux with CUDA
Apple Silicon MacmacOS with Metal
Windows (native or WSL)Windows
Build from sourceBuild from source

The cargo features reference maps GPU generations to feature flags. For containerised or production deployment, see the Deploy guides.

Video input uses FFmpeg at runtime. The install commands and runtime checks are in Set up video input.