base64-to-tensor

pure js convert a base64 image to tensor

Installation

npm i base64-to-tensor --save

Getting Started

Make sure to have @tensorflow/tfjs-core installed and a valid tensorflow backend set. You also need to pick between sync package jpeg-js or async package sharp.

1# pure js full sync blocking installation
2npm i @tensorflow/tfjs-core jpeg-js
3# if going to use async non blocking
4npm i @tensorflow/tfjs-core sharp

View the convert.test.ts file for an example setup.

1import { convert, convertAsync } from "base64-to-tensor";
2import { setBackend } from "@tensorflow/tfjs-core";
3import "@tensorflow/tfjs-backend-wasm";
4
5await setBackend("wasm");
6
7const tensor = convert(mybase64); // The base64 must be a valid jpeg image.
8// or use native sharp for increased performance 2x [Expiremental]
9const tensor = await convertAsync(mybase64);
10// output example
11// {
12//       kept: false,
13//       isDisposedInternal: false,
14//       shape: [189, 300, 3],
15//       dtype: "int32",
16//       size: 170100,
17//       strides: [900, 3],
18//       dataId: { id: 1 },
19//       id: 1,
20//       rankType: "3",
21//     }

Why

The benefits of using pure js to calc the image is in a couple areas:

1. size and portablity required is drastically less since you do not need cairo or any of the native img dev converters.
2. speed is also faster since the calcs are done at hand without needing to bridge any calls.
3. can use tensors in worker threads - allows for properly using Tensorflow wasm backends in an API service 🥳.

Benchmarks

Examples of some test ran on a mac m1(64gb):

Name	chars	size	sync	async
jpeg	26791	26.16 KB	100ms	50ms