记录一下pytorch成功部署到android的版本

最新推荐文章于 2024-05-01 21:58:29 发布

Junekakui

最新推荐文章于 2024-05-01 21:58:29 发布

阅读量766

点赞数 10

文章标签： pytorch 人工智能 python

本文链接：https://blog.csdn.net/Junekakui/article/details/136098222

版权

本文讲述了作者在将PyTorch1.8.0模型转换为1.10.1并在Android应用中部署时遇到的闪退问题，主要集中在版本不兼容和加载模型推理阶段。通过调整AndroidStudio配置和处理数据类型，解决了该问题。

摘要由CSDN通过智能技术生成

模型训练的pytorch==1.10.1（应该是不重要）

转换模型（生成model.ptl）的pytorch==1.8.0

（之前用官方给的pytorch-android项目改出现了在手机上运行闪退的情况，总的来说就是版本不匹配）

所以，我是直接新建了

在这个基础上改的，每一步都在手机上部署，最后查出是加载模型和推理这两个步骤导致闪退

android studio的配置：

gradle-wrapper.properties

distributionBase=GRADLE_USER_HOME
distributionPath=wrapper/dists
distributionUrl=https\://services.gradle.org/distributions/gradle-8.2-bin.zip
zipStoreBase=GRADLE_USER_HOME
zipStorePath=wrapper/dists

build.gradle.kts（Project）

plugins {
    id("com.android.application") version "8.2.2" apply false
}

build.gradle.kts（app）

plugins {
    id("com.android.application")
}


android {
    namespace = "com.example.myapplication"
    compileSdk = 34

    defaultConfig {
        applicationId = "com.example.myapplication"
        minSdk = 30
        targetSdk = 34
        versionCode = 1
        versionName = "1.0"
        testInstrumentationRunner = "androidx.test.runner.AndroidJUnitRunner"
    }

    buildTypes {
        release {
            isMinifyEnabled = false
            proguardFiles(
                getDefaultProguardFile("proguard-android-optimize.txt"),
                "proguard-rules.pro"
            )
        }
    }
    compileOptions {
        sourceCompatibility = JavaVersion.VERSION_1_8
        targetCompatibility = JavaVersion.VERSION_1_8
    }
}

dependencies {
    implementation("androidx.appcompat:appcompat:1.6.1")
    implementation("com.android.support:multidex:2.0.1")
    implementation("com.google.android.material:material:1.9.0")
    implementation("androidx.constraintlayout:constraintlayout:2.1.4")
    testImplementation("junit:junit:4.13.2")
    androidTestImplementation("androidx.test.ext:junit:1.1.5")
    androidTestImplementation("androidx.test.espresso:espresso-core:3.5.1")
    implementation("org.pytorch:pytorch_android_lite:1.9.0")
    implementation("org.pytorch:pytorch_android_torchvision:1.9.0")
}

gradle.properties

# Project-wide Gradle settings.
# IDE (e.g. Android Studio) users:
# Gradle settings configured through the IDE *will override*
# any settings specified in this file.
# For more details on how to configure your build environment visit
# http://www.gradle.org/docs/current/userguide/build_environment.html
# Specifies the JVM arguments used for the daemon process.
# The setting is particularly useful for tweaking memory settings.
org.gradle.jvmargs=-Xmx4608m -Dfile.encoding=UTF-8
# When configured, Gradle will run in incubating parallel mode.
# This option should only be used with decoupled projects. More details, visit
# http://www.gradle.org/docs/current/userguide/multi_project_builds.html#sec:decoupled_projects
# org.gradle.parallel=true
# AndroidX package structure to make it clearer which packages are bundled with the
# Android operating system, and which are packaged with your app's APK
# https://developer.android.com/topic/libraries/support-library/androidx-rn
android.useAndroidX=true
android.enableJetifier=true
# Enables namespacing of each library's R class so that its R class includes only the
# resources declared in the library itself and none from the library's dependencies,
# thereby reducing the size of the R class for that library
android.nonTransitiveRClass=true

读取模型

module = LiteModuleLoader.load(assetFilePath(this, "model.ptl"));

部分推理的代码

（这也是一个问题，关于数据类型的）

参考：Pytorch mobile app error · Issue #2666 · pytorch/vision · GitHub

final Tensor inputTensor = TensorImageUtils.bitmapToFloat32Tensor(bitmap,
                TensorImageUtils.TORCHVISION_NORM_MEAN_RGB, TensorImageUtils.TORCHVISION_NORM_STD_RGB, MemoryFormat.CHANNELS_LAST);
IValue inputs = IValue.from(inputTensor);
IValue[] outputs = module.forward(inputs).toTuple();

在修改的过程中，出现问题了，先做下：clean project、rebuild、关掉as再打开。

太玄学了，希望以后不要再和安卓打照面，新年快乐。