ncnn: a high-performance neural network inference framework optimized for mobile, embedded, and desktop deployment

What it solves

ncnn is a high-performance neural network inference framework designed to make it easier and more efficient to deploy deep learning models on resource-constrained devices, such as mobile phones, embedded systems, PCs, and browsers.

How it works

It provides a lightweight runtime that has no third-party dependencies, allowing it to run across CPU and Vulkan GPU backends. To get models into the framework, it includes tools like pnnx for converting models from PyTorch and ONNX formats into the ncnn format.

Who it’s for

Developers who need to deploy AI models to edge devices, mobile applications, or desktop software without relying on heavy external runtimes.

Highlights

• High-performance inference optimized for mobile and embedded deployment.
• Zero third-party runtime dependencies.
• Supports both CPU and Vulkan GPU backends.
• Includes pnnx for seamless conversion from PyTorch and ONNX.
• Broad platform support including Android, iOS, Windows, macOS, Linux, WebAssembly, and HarmonyOS.