Welcome to ncnn’s documentation!

Please select a specific version of the document in the lower left corner of the page.

how to build

• how to build
• mlir2ncnn
• build VS2017
• build minimal library

how to use

• efficient roi resize rotate
• ncnn protobuf problem
• FAQ ncnn produce wrong result
• FAQ ncnn throw error
• FAQ ncnn vulkan
• ncnn load model
• openmp best practice
• openmp 最佳实践
• Post Training Quantization Tools
• use ncnnoptimize to optimize model
• use ncnn with alexnet
• ncnn 执行 AlexNet
• use ncnn with opencv
• use ncnn with own project
• use ncnn with pytorch or onnx
• vulkan notes

benchmark

• the benchmark of caffe android lib, mini caffe, and ncnn
• vulkan conformance test

developer guide

• aarch64 mix assembly and intrinsic
• 如何增加自定义层
• arm a53 a55 dual issue
• armv7 mix assembly and intrinsic
• binaryop broadcasting
• custom allocator
• element packing
• how to implement custom layer step by step
• benchmark
• how to be a contributor
• implement elementwise addition with/without broadcast using BinaryOp operation
• ncnn tips and tricks
• new model load api
• new param load api
• operation param weight table
• operators
• preload practice
• param and model file structure
• tensorflow op combination

home

• home
• faq
• application with ncnn inside