微信小程序开发记录——Error:No backend found in registry

在学习开发图片识别的微信小程序时，要实现从网络下载模型并在小程序上使用模型预测，在微信开发工具上能够成功运行并识别，可在手机预览或是真机调试时始终在模型预测时报如下错误：

No backend found in registry.
Error: No backend found in registry.
    at t.getSortedBackends (weapp:///miniprogram_npm/@tensorflow/tfjs-core/index.js:1522:65)
    at t.initializeBackendsAndReturnBest (weapp:///miniprogram_npm/@tensorflow/tfjs-core/index.js:1527:25)
    at t.get [as backend] (weapp:///miniprogram_npm/@tensorflow/tfjs-core/index.js:1426:24)
    at t.makeTensor (weapp:///miniprogram_npm/@tensorflow/tfjs-core/index.js:1711:41)
    at Object.ee [as zeros] (weapp:///miniprogram_npm/@tensorflow/tfjs-core/index.js:2398:15)
    at n.apply (weapp:///miniprogram_npm/@tensorflow/tfjs-layers/index.js:873:16)
    at e.n.addWeight (weapp:///miniprogram_npm/@tensorflow/tfjs-layers/index.js:1665:17)
    at e.n.build (weapp:///miniprogram_npm/@tensorflow/tfjs-layers/index.js:5613:26)
    at eval (weapp:///miniprogram_npm/@tensorflow/tfjs-layers/index.js:1580:13)
    at X (weapp:///miniprogram_npm/@tensorflow/tfjs-layers/index.js:556:15)

未在注册表中找到后端？可在package.json中以及安装了相关依赖

"dependencies": {
    "@tensorflow/tfjs-backend-webgl": "2.0.1",
    "@tensorflow/tfjs-backend-cpu": "2.0.1",
    ...
  },

进行各种尝试后，在模型进行预测的js文件中加上

require("@tensorflow/tfjs-backend-webgl")
require("@tensorflow/tfjs-backend-cpu")

解决报错且成功能在预览以及真机调试上实现模型预测！