A no-code guide to Machine Learning
Get started with ML in minutes
It’s no secret machine learning (ML) is one of the coolest things out there. Well, at least I seem to think so. The possibilities are endless, and we just seem to have scratched the surface of making useful things with it.
It’s also no secret that it’s not a walk in the park to get started with machine learning if you don’t understand code.
TensorFlow is the place to be if you understand a bit of coding. If you don’t, however, things don’t seem that easy. Well, at least they didn’t until recently. As time passes, ML seems to become more accessible for no-coders.
I thought I’d compile a few places to get started here if you’re not into code and want to get into ML. Places on the web that will help you experiment, play around with, and let you build something cool.
1. RunwayML
RunwayML is probably all you will need for most of your ML needs. Think of it as an App Store for ML models. You can browse a large collection of models, try them out live, and mix and match them to suit any need you might have. Their collection of models is up to date, and honestly incredible.
That’s not all. You can also train your own models, and it supports integrations with Processing, Photoshop, Ableton, Arduino, Unity to name a few.
It has a great community, learning resources, and best of all it is pretty intuitive to use. Runway can run on your browser now, in addition to a desktop app, which I think is amazing.
Here’s a great Introduction to Runway video by The Coding Train:
2. Teachable Machine
Google’s Teachable Machine lets you train your computer to recognize sounds, poses, and images. You can classify these however you like, based on whatever project you are building, and export the model when you’re done to use with your website or app.
Teachable Machine also supports integrations with TensorFlow, ml5js, NodeJS, Framer, and more. It works REALLY well for what it does, all with a few clicks and no coding.
Here’s a short video which can help you get the gist of what it does:
3. ml5js
Okay, I lied when I said all the tools here would be no-code, but hear me out. Remember when I said TensorFlow is the place to be? Ml5js builds is an easy and accessible way to use some of TensorFlow’s models including pose estimation, object detection, image classification, and pitch detection. All this directly on your browser.
The reason why I mentioned ml5js in a no-code guide here is that you don’t need to know a LOT of code to make something. The documentation and examples are so good that they are more than enough to get you started with experimenting with different models. It also happens to be open-source, and the community has some great resources for learning.
It also goes hand in hand with p5js, so if you are a creator, artist, designer, or an enthusiast, you should definitely go check ml5js out. Check out their reference to see what it can do.
The Coding Train has a brilliant YouTube playlist dedicated to ml5js and p5js :
4. Wekinator
Wekinator is another free, open-source software that lets you use machine learning to build new musical instruments, gestural game controllers, computer vision or computer listening systems, and more.
It works similarly to Teachable Machine in the sense you can train your own models with this. This is more flexible but also has a tougher learning curve compared to Teachable Machine.
Its Open Sound Control integrations are what take it to the next level. You can combine this with any software that supports OSC like Processing, Arduino, and Ableton to create magical things like gesture-controlled musical instruments.
For example, I created a school project which detected what kind of trash was placed on the counter and opened the appropriate bin to throw it in using Wekinator with Arduino.
Their Github repo with Wekinator examples is a great place to get started.
Here’s a quick walkthrough of how Wekinator works:
5. ml4a
ml4a, or Machine Learning for Artists is a collection of free educational resources devoted to machine learning for artists, designers, and enthusiasts.
It is a community, not a tool, but it has great tutorials to get you started with ML for art. It also has guides and examples to help you get started with ml5js and OpenFrameworks.
There’s more out there
This barely scratches the surface of what is available out there, but these are tools and communities that helped me get started. Once you start experimenting you will learn what you like and can go deeper into those topics.
It is also incredibly fun to connect some of the above tools with other OSC integrated tools like Ableton, Processing, Runway, Arduino, etc. to create experiences across physical, digital, and sound.
You can always email me at ahujasid@gmail.com or on Twitter @sidahuj for any help, or if you end up making something cool with these tools I’d love to know!
I usually post about things at the intersection of design and tech, so do follow for more content like this.
