Overview

SvelteML is a library that aims to make machine learning more accessible by offering simple components. It leverages the power of TensorflowJS to provide Svelte apps with out-of-the-box machine learning features. SvelteML heavily relies on Svelte's reactivity feature, and event hooks can be utilized to extend the machine learning flow.

Features

Classification / Segmentation: Includes image classification, body segmentation, and basic multi-pose estimation.
Text-based inference: Supports sentence encoding, text toxicity analysis, and question and answers.
Image Effects: Allows for blur effect on body parts, bokeh effect, face mesh rendering, and hand pose detection.
Experiments: Ongoing work on switching to Lerna for multiple repositories to allow expansion of the library in different areas. Also planning to add @svelteml/ui, @svelteml/classification, @svelteml/segmentation, @svelteml/automl, @svelteml/text, @svelteml/audio. Additional features like unlocking slots with facial recognition using faceapi.js, a demo site for more details, audio/speech recognition, and additional models using lower level tfjs APIs are in the pipeline.

Summary

SvelteML is a library that aims to make machine learning more accessible for Svelte developers. With its range of components, it provides out-of-the-box machine learning features powered by TensorflowJS. The library is highly reactive and provides hooks to extend the machine learning flow. Additionally, there are plans for further expansion and features, making it an exciting tool for Svelte developers interested in machine learning.

John Kane

Svelteml