.
1) Import the TensorFlow.js
2) Import A helper method to show the tensor array
3) Create a model that is used to predict the MNIST dataset. The module constitutes two conv layers two pooling layers and a dense layer. Compile the model using an adam optimizer.
4) Load a pretrained model called mobilenet.
5) Load the image classes.
6) Predict image class.
6.1) Preprocess the image by converting it to a tensor then normalize,
6.2) Predict based on the given model classes
.
https://observablehq.com/@zaidalyafeai/an-intractive-introduction-to-tensorflow-js
.
Alternative:
https://observablehq.com/d/95a9b4e931cccfb0
,