The purpose of SvelteML is to offer simple Components that can make ML more accessible. It leverages TensorflowJS to offer Svelte apps ML features out-of-the-box. It relies heavily on Svelte's reactivity feature and event hooks can be used to extend out the ML flow. e.g. on:poses in the Pose Estimator will give you the raw poses directfrom TensorflowJS.
npm install svelteml --save
- Image Classification
- Body Segmentation
- Basic Multi-Pose Estimation
- Object Detection
- Sentence Encoding
- Text Toxicity
- Question and Answers
- Blur Body Parts
- Bokeh Effect
- Face Mesh
- Hand Pose Detection
- Switching to Lerna for multiple repos so the lib can expand in the different areas. Also helpful for tfjs3 when it will have code-splitting ๐
- @svelteml/ui
- @svelteml/classification
- @svelteml/segmentation
- @svelteml/automl
- @svelteml/text
- @svelteml/audio
- Unlock slots with Facial recognition, maybe use faceapi.js
- Demo site for more details
- Audio and speech recognition features
- Additional models using the lower level tfjs apis
- .... and a few other top secret ideas ๐คญ
All Components try to be reactive so although it feels very declarative, it is also reacting to your input. Add an issue in Github if you need a specific behaviour or if there is a bug or would like to recommend something. You know the drill.