-
(Optional) Watch the beginner's guide video (see references above) for an overview.
-
Some other important things to know about ML...
Machine learning libraries:
- Tensorflow is a well-known open source ML library from Google, for python language.
- tf.js is Tensorflow for javascript developers. But a better beginner starting point is...
- ML5.js, a javascript library that aims to make ML approachable, from a learning perspective as well as a cost perspective.
- Teachable Machine, from Google, building off all of the above and making a user-friendly UI for training models.
Machine learning as a field:
- ML can be very powerful, handling tasks with more ambiguity such as interpreting handwriting, recognizing faces, etc.
- ML is generally quite costly since it has to do an incredible number of calculations, sometimes requiring expensive GPUs.
- ML is being increasingly used in many aspects of life, from captcha to facial recognition, and there are many ethical considerations to this.
- Get started watching the tutorial video Teachable Machine 1: Image Classification. Remix current project for starter code, it's basically the "Code Template".
- Find 3-4 objects around your house that are easy to hold and that you will map to emojis.
- Go to Teachable Machine website and choose image project, standard image model.
- Make a class for every object you have and use webcam option to upload about 50-100 images of your object in different positions. Move it around, rotate it, get different angels!
- Train the model (DO NOT LEAVE TAB) and test to see if it's working well. If not, recollect data and re-train.
- Export Model > Tensorflow.js > Upload (shareable link). Hit "Upload My Model" and get the shareable link!
- Come back to glitch, remix this project.
- Follow coding steps in video! There's also a final code link in the description if you are stuck.
- Feel free to play around and build off this or try some other ML5.js tutorials from The Coding Train!