Question

很明显，我不能只为我为TensorFlow编写的自定义操作复制并粘贴代码，然后在Lite版本中使用它

我的代码引用了主要代码的内容：

#include "tensorflow/core/framework/op.h"
#include "tensorflow/core/framework/shape_inference.h"
#include "tensorflow/core/framework/op_kernel.h"

using namespace tensorflow;

和

Status MyAddGrad(const Scope& scope, const Operation& op,
                 const std::vector<Output>& grad_inputs,
                 std::Vector<Output>* grad_outputs){

我在TensorFlow Lite中找不到自定义操作的完整样本（甚至最少）。只有我应该修改哪些文件才能将Lite编译为'.aar'

Answer 1

Custom operators指南页面显示了如何在TFLite中实现package com.example.attendanceapp; import android.content.Context; import android.os.AsyncTask; import java.io.BufferedReader; import java.io.IOException; import java.io.InputStream; import java.io.InputStreamReader; import java.net.HttpURLConnection; import java.net.MalformedURLException; import java.net.URL; import java.sql.SQLOutput; public class BackgroundWorker extends AsyncTask<String,Void,String> { Context context; private String id; BackgroundWorker (Context ctx) { context=ctx; } @Override protected String doInBackground(String... params) { String conn_url="http://identifyme.co.nf/app.php?CDATE="+"d20_5_2019"; try { URL url= new URL(conn_url); HttpURLConnection httpURLConnection= (HttpURLConnection)url.openConnection(); httpURLConnection.setRequestMethod("POST"); httpURLConnection.setDoOutput(true); httpURLConnection.setDoInput(true); InputStream inputStream=httpURLConnection.getInputStream(); BufferedReader bufferedReader= new BufferedReader(new InputStreamReader(inputStream,"iso-8859-1")); String result=""; String line=""; bufferedReader.close(); inputStream.close(); httpURLConnection.disconnect(); return result; } catch (MalformedURLException e) { e.printStackTrace(); } catch (IOException e) { e.printStackTrace(); } return null; } @Override protected void onPreExecute() { super.onPreExecute(); } @Override protected void onPostExecute(String result) { //super.onPostExecute(); } @Override protected void onProgressUpdate(Void... values) { //super.onProgressUpdate(values); } } This is a java classby the name of BackgroundWorker , I am calling this in another class on OnCreate of that class : BackgroundWorker backgroundWorker =new BackgroundWorker(this); backgroundWorker.execute(a,null,heightval);自定义操作的规范示例。

您必须为op定义Sin和Prepare函数，如here所示，然后使用Eval调用让OpResolver知道。

TensorFlow Lite的自定义操作是否有最低样本要求？

1 个答案: