相关资料

• https://github.com/tensorflow/tfjs
• https://github.com/infinitered/nsfwjs
• model*299 下载地址: https://github.com/infinitered/nsfwjs/tree/master/example/nsfw_demo/public/model 或者 https://s3.amazonaws.com/nsfwdetector/min_nsfwjs.zip

环境准备

• nodejs v14 +
• centos 8.2

由于windows 上没有编译通过，直接在轻量腾讯云上做的测试

yum -y update gcc
yum -y install gcc-c++
npm i @tensorflow/tfjs-node nsfwjs axios --save

代码

参考 nsfwjs 的 Nodejs.App

/**
 * nsfw.js 测试
 */

const axios = require('axios') //you can use any http client
const tf = require('@tensorflow/tfjs-node')
const nsfw = require('nsfwjs');
let path = require('path');
let fs = require('fs');

async function fn () {
  // const pic = await axios.get(`img-link`, {
  //   responseType: 'arraybuffer',
  // });
  let pic = fs.readFileSync('./test.png');
  // console.log(pic);
  console.log(path.resolve('./model/model.json'));
  const model = await nsfw.load('file://./model/',{size : 299}) // To load a local model, nsfw.load('file://./path/to/model/')
  // Image must be in tf.tensor3d format
  // you can convert image to tf.tensor3d with tf.node.decodeImage(Uint8Array,channels)
  const image = await tf.node.decodeImage(pic, 3)
  const predictions = await model.classify(image)
  image.dispose() // Tensor memory must be managed explicitly (it is not sufficient to let a tf.Tensor go out of scope for its memory to be released).
  console.log(predictions)
}
fn()

做了两个测试，一类网络图片，一类本地图片。

结果

坑

• 坑A ： nsfw.load('') 此处加载本地model ，一直在说要求全路径.. 没想到必须要带 file://
• 坑B ： 由于没有细看 nsfw readme ，找的网上的资料，总是报错

expected input_1 to have shape [null,299,299,3] but got array with shape [1,224,224,3].

在 load 内 增加一个 size .

const model = await nsfw.load('file://./model/',{size : 299})