แนะนำ nsfwjs JavaScript Library ที่จะทำให้คุณแยกรูปภาพที่ไม่เหมาะสมได้

packge install
"npm i nsfwjs @tensorflow/tfjs-node express express-fileupload"`
"`
const express = require("express")
const fileUpload = require("express-fileupload")
const path = require("path")

const filesPayloadExists = require("./middleware/filesPayloadExists")
const fileExtLimiter = require("./middleware/fileExtLimiter")

const PORT = process.env.PORT || 3000

const app = express()

app.post(
  "/upload",
  fileUpload({ createParentPath: true }),
  filesPayloadExists,
  fileExtLimiter([".png", ".jpg", ".jpeg"]),
  (req, res) => {
    const files = req.files

    Object.keys(files).forEach((key) => {
      const filepath = path.join(__dirname, "files", files[key].name)
      files[key].mv(filepath, (err) => {
        if (err) return res.status(500).json({ status: "error", message: err })
      })
    })

    return res.json({
      status: "success",
      message: Object.keys(files).toString(),
    })
  }
)

app.listen(PORT, () => console.log(`Server running on port ${PORT}`))
"`

const express = require("express")
const fileUpload = require("express-fileupload")
const path = require("path")
const nsfw = require("nsfwjs")
const tf = require("@tensorflow/tfjs-node")

const filesPayloadExists = require("./middleware/filesPayloadExists")
const fileExtLimiter = require("./middleware/fileExtLimiter")

const PORT = process.env.PORT || 3000

const app = express()
const modelPromise = nsfw.load() // load the model once when the server starts up

app.post(
  "/upload",
  fileUpload({ createParentPath: true }),
  filesPayloadExists,
  fileExtLimiter([".png", ".jpg", ".jpeg"]),
  async (req, res) => {
    const files = req.files
    const model = await modelPromise

    try {
      await Promise.all(
        Object.keys(files).map(async (key) => {
          const image = await tf.node.decodeImage(files[key].data, 3) // decode the image data using tf.node.decodeImage()
          const predictions = await model.classify(image)
          image.dispose() // dispose of the image tensor to free up memory
          if (
            ["Sexy", "Neutral", "Drawing"].includes(predictions[0].className)
          ) {
            const filepath = path.join(__dirname, "files", files[key].name)
            await files[key].mv(filepath) // use await to handle the file move operation
          } else {
            throw new Error("Invalid file content")
          }
        })
      )

      return res.json({
        status: "success",
        message: Object.keys(files).toString(),
      })
    } catch (err) {
      return res.status(400).json({ status: "error", message: err.message })
    }
  }
)

app.listen(PORT, () => console.log(`Server running on port ${PORT}`))