The library ml5.js(https://ml5js.org/) is the higher level API built on the top of Tensorflow.js. The library TensorFlow.js(https://js.tensorflow.org/) is for training and deploying ML models in the browser or on Node.js.
Mljs is a part of Sophia project focusing on the coding tutorial to machine learning and deep learning. More details please refer to Sophia, https://github.com/jiankaiwang/sophia.
Reference:
- Ml5-example : https://github.com/ml5js/ml5-examples
ML5.js is heavily inspired by Processing and p5.js. In this section, we try to provide you with two different kinds of javascript scripts based on either native ml5.js or p5.js included.
- Basis (error-first callbacks and promises) : scripts
- frame classification on a video streaming : scripts, ml5, ml5+p5
- Image classification using feature extraction with MobileNet : scripts, ml5, ml5+p5
- Regression using feature extraction : scripts, ml5+p5
- Text generation with LSTM : scripts, ml5+p5
- Interactive text generation with LSTM : scripts, ml5+p5
- Style transfer : scripts, ml5+p5
- Style transfer with a video streaming : scripts, ml5+p5
- Pixel 2 Pixel : scripts, ml5+p5
- PoseNet detection with a video streaming : scripts, ml5+p5
- Object detection in a video streaming with YOLO : scripts, ml5+p5
- Word2Vec : scripts, ml5+p5
- Image classification based on D-CNN network : Here we demo how to train an image classification on a deep CNN network, and dump out the variable weights in binary format. Next, we also show how to view the weights.
In this section, we provide you with more details about lower level API to machine learning or deep learning tasks.
- Setup Tensorflow.js: Tutorial
- Script Tag: jsFiddle demo, browser_setup.html
- Setup via NPM and Packaging via Bundle Tools : npm_browser.js,
- Implement on the Server-Side : nodejs_setup.js
- A Quickstart
- Convert from a pertained models. MD
- Tensorflow.js Visualization : tfjs-vis
- Using TensorflowBoard