classify_image_mobile

Source: https://codelabs.developers.google.com/codelabs/tensorflowjs-teachablemachine-codelab/index.html#0

Introduction:

We build a simple "teachable machine", a custom image classifier that can be trainsed on the fly in the browser using TensorFlow.js, a powerful and flexible machine learning library for Javascript. We first load and run a popular pre-trained model called MobileNet for image classification in the browser. we then use "transfer learning" to bootstrap NN training with the pre-trained MobileNet model and customizes it.

If you need an introduction or refresher, consider watching:

• a video by 3blue1browns: https://www.youtube.com/watch?v=aircAruvnKk
• a video on Deep Learning in Javascript by Ashi Krishnan. https://www.youtube.com/watch?v=SV-cgdobtTA

To know more

• about tensorflow applications based on inference, derived from (pretrained) transfer models ported to mobile platform, check out: https://beta.observablehq.com/@nsthorat/how-to-build-a-teachable-machine-with-tensorflow-js

Objective

• load pretrained MobileNet model and make a prediction on new data
• make predictions through the webcam
• use intermediate activations of MobileNet to do transfer learning on a new set of classes you define on the fly with the webcam

Requirements:

if you want to modify the code, you will need:

• a text editor, either running locally on your machine or terminal
• some knowledge of HTML, CSS, JavaScript, and your browser's devtools
• a high-level conceptual understanding of Neural Networks

Trials:

To run the webpage, simply open 'index.html' in a Web Browser. If you use the cloud console, simply refresh the preview page.

Downloading the model may take a few seconds.