Awesome-NCNN

NCNN is a high-performance neural network inference framework optimized for the mobile platform. This repo lists some awesome ncnn-based projects. Welcome Star & Fork & Pull Requests!

NCNN 是一个为手机端极致优化的高性能神经网络前向计算框架。本仓库收集了基于NCNN的很棒的项目。欢迎 star & fork & pull request 一键三连！

Optimization

ncnn-with-cuda: Tencent NCNN with added CUDA support

https://github.com/atanmarko/ncnn-with-cuda/

Applications 应用类

针对具体任务、场景的基于ncnn的项目

NanoDet NCNN Android Demo

NanoDet: Super fast and lightweight anchor-free object detection model. 🔥Only 1.8mb and run 97FPS on cellphone.

Providing NanoDet image inference, webcam inference and benchmark using Tencent's NCNN framework.

https://github.com/RangiLyu/nanodet

ncnn-android-deeplabv3plus

The deeplabv3+ person segmentation android example.

https://github.com/runrunrun1994/ncnn-android-deeplabv3plus

ncnn-webassembly-yolov5

Run NCNN based YoloV5 detector in your browser!

https://github.com/nihui/ncnn-webassembly-yolov5

ncnn-picture-enhancement ncnn照片去雾和水下增强

A simple demo to run dehaze / underwater model in Android

https://github.com/JuZiSYJ/ncnn-picture-enhancement

ncnnRay++

A CMake / WASM integration of rayib UI and the Tencent ncnn C++ AI platform

https://github.com/QuantScientist/ncnnRay

Android/iOS camera preview with YOLOv5

移动端目标检测，当前项目使用的是YOLOv5的5s模型，摄像头实时捕获视频流进行检测

https://github.com/WZTENG/YOLOv5_NCNN

face mask detection on Rasberry Pi 树莓派上的口罩检测

https://github.com/Qengineering/Face-Mask-Detection-Raspberry-Pi-64-bits

YOLOv5 NCNN Implementation

YOLOv5 on Android based on NCNN

https://github.com/sunnyden/YOLOV5_NCNN_Android

DAIN ncnn Vulkan

DAIN, Depth-Aware Video Frame Interpolation implemented with ncnn library

https://github.com/nihui/dain-ncnn-vulkan

PFLD pytorch Implementation ，自带 ncnn 推理实现

Practical Facial Landmark Detector with PyTorch and NCNN implementation.

https://github.com/polarisZhao/PFLD-pytorch

C++ implemententation of LFFD with ncnn

LFFD ( A Light and Fast Face Detector for Edge Devices )'s implementation in NCNN.

https://github.com/SyGoing/LFFD-with-ncnn

ncnn在Android的一个测试,包含了人脸检测(face detection),人脸属性(face attributes),人脸识别(face recognition);车辆检测(Vehicle detection),车牌检测(plate detection),车牌识别(plate recognition);人头检测(head detection)的流程

https://github.com/791136190/ncnn_android_face_vehicle

Objects as Points ncnn 实现

https://github.com/wanglaotou/centernet_ncnn

centerface android 工程

https://github.com/JuZiSYJ/centerface-ncnn

PCN-ncnn for rotated face detection

https://github.com/HandsomeHans/PCN-ncnn

Progressive Calibration Networks (PCN) is an accurate rotation-invariant face detector running at real-time speed on CPU, published in CVPR 2018.

deepsort tracking demo

https://github.com/ProLing1994/demo_deepsort

超轻量级中文ocr，支持竖排文字识别, 支持ncnn推理

https://github.com/ouyanghuiyu/chineseocr_lite

waifu2x converter ncnn version, runs fast on intel / amd / nvidia GPU with vulkan

https://github.com/nihui/waifu2x-ncnn-vulkan

ncnn style transfer android example

https://github.com/nihui/ncnn-android-styletransfer

1MB lightweight face detection model (1MB轻量级人脸检测模型)

https://github.com/Linzaer/Ultra-Light-Fast-Generic-Face-Detector-1MB

A collection of ncnn examples: face/mask detection, tracking, recognition...

https://github.com/MirrorYuChen/ncnn_example

Tools 工具类

模型转换工具、对ncnn封装等相关项目

keras2ncnn: A keras h5df to ncnn model converter

https://github.com/MarsTechHAN/keras2ncnn

darknet ncnn android 工程

https://github.com/paleomoon/darknet-ncnn-android

ROS wrapper for NCNN neural inference framework

https://github.com/nilseuropa/ros_ncnn

pyncnn: python wrapper of ncnn with pybind11

https://github.com/caishanli/pyncnn

Caffe INT8 Quantization convert tool

https://github.com/BUG1989/caffe-int8-convert-tools

NCNN lite without C++ support.

https://github.com/nullptr-leo/ncnn-lite

Waifu2x filter for VapourSynth

https://github.com/Nlzy/vapoursynth-waifu2x-ncnn-vulkan

SRMD super resolution for VapourSynth

https://github.com/Kiyamou/VapourSynth-SRMD-ncnn-Vulkan

ncnn .NET wrapper written in C++ and C# for Windows, MacOS and Linux

https://github.com/takuya-takeuchi/NcnnDotNet

matt45m/awesome-ncnn