1、下载ncnn

地址：

https://github.com/Tencent/ncnn/tree/20230223

下载的版本是

ncnn-20230223-android-vulkan

---

2、下载opencv-mobile

下载的版本是：opencv-mobile-4.6.0-android
官网地址：

https://github.com/nihui/opencv-mobile/releases?page=2

3、文件拷贝

将上述下载的ncnn-20230223-android-vulkan与opencv-mobile-4.6.0-android分别解压，拷贝到
目录D:\ncnn-android-yolov8-seg\app\src\main\jni下

---

4、andorid_studio相关配置

我的这个版本是：Android Studio Flamingo | 2022.2.1 Patch 1

---

配置SDK Platforms

配置SDK Platforms

5、文件内参数设置

文件：build.gradle (Project)

// Top-level build file where you can add configuration options common to all sub-projects/modules.
buildscript {repositories {jcenter()google()}dependencies {classpath 'com.android.tools.build:gradle:3.5.0'}
}allprojects {repositories {jcenter()google()}
}

文件：build.gradle (Module: app)

apply plugin: 'com.android.application'android {compileSdkVersion 24buildToolsVersion "29.0.2"defaultConfig {applicationId "com.tencent.yolov8ncnn"archivesBaseName = "$applicationId"minSdkVersion 24}externalNativeBuild {cmake {version "3.10.2"path file('src/main/jni/CMakeLists.txt')}}dependencies {implementation 'com.android.support:support-v4:24.0.0'}ndkVersion '24.0.8215888'
}

文件：gradle-wrapper.properties

#Sun Sep 08 23:09:42 CST 2019
distributionBase=GRADLE_USER_HOME
distributionPath=wrapper/dists
zipStoreBase=GRADLE_USER_HOME
zipStorePath=wrapper/dists
distributionUrl=https://mirrors.cloud.tencent.com/gradle/gradle-5.4.1-all.zip

文件：local.properties

## This file must *NOT* be checked into Version Control Systems,
# as it contains information specific to your local configuration.
#
# Location of the SDK. This is only used by Gradle.
# For customization when using a Version Control System, please read the
# header note.
#Fri Sep 12 17:47:07 CST 2025
cmake.dir=D\:\\Android_SDK\\Sdk\\cmake\\3.10.2.4988404
ndk.dir=D\:\\Android_SDK\\Sdk\\ndk\\24.0.8215888
sdk.dir=D\:\\Android_SDK\\Sdk

文件：settings.gradle

文件：CMakeLists.txt

5、重构项目：

6、打包apk

但是打包过程中会出现报错，提示信息如下：

Execution failed for task ':app:stripDebugDebugSymbols'.
> No toolchains found in the NDK toolchains folder for ABI with prefix: arm-linux-androideabi* Try:
Run with --stacktrace option to get the stack trace. Run with --info or --debug option to get more log output. Run with --scan to get full insights.

解决方案：

同时安装低版本的ndk（如version=21.0.6113669），将低版
本ndk中toolchains 文件夹下的arm-linux-androideabi等文件复制到高版
本ndk的toolchains 文件夹中

复制后如下所示：：复制到24.0.8215888对应的目录下

---

再次重构，打包，效果如下所示：

7、部署自己训练的实例分割模型

#测试  ## 模型推理
yolo segment predict model=yolov8s-seg.pt   source=0a2cf558-d1bd-4ad8-9724-ce5760192bb9.jpg

模型转换

yolo export model=yolov8s-seg.pt format=onnx simplify=True opset=12

# 使用导出的.onnx模型推理
## 默认使用gpu
yolo segment predict model=yolov8s-seg.onnx source=0a2cf558-d1bd-4ad8-9724-ce5760192bb9.jpg 
## 默认使用cpu
yolo segment predict model=yolov8s-seg.onnx source=0a2cf558-d1bd-4ad8-9724-ce5760192bb9.jpg device=cpu

对.onnx模型进行去冗余操作：

安装相关依赖包：

pip install onnx-simplifier

# 去冗余指令：
python -m onnxsim   yolov8s-seg.onnx   yolov8s-seg-sim.onnx