How the course works.
A course to understand what Edge AI is and to build real models using only a smartphone. Images, sounds, motion and text, from the first piece of data you collect to a model that answers in airplane mode.
Lesson zero lets you try your phone in one minute and shows you how to use the course, the map and what unlocks, Listen and Scroll & read, the quiz and what you need.
Start here →- 30lessons in 7 courses
- 5phone sensors used in the lab
- 0apps to install and accounts to create
- 1smartphone, the one you already have
Every kind of model, box by box.
On one side is the data the phone collects, on the other what the model has to do. Each box is a kind of model that runs on a device today, with a real example. The method to build them is always the same.
Seven courses. The first one is for everyone.
The Basics course explains what Edge AI is and how a model learns. After that you go where you like, one sensor at a time. The last course takes you from model to app, and the final project puts it all together.
-
(01)
The Basics
For everyone, start here
5 lessons
What Edge AI is, what's inside a phone and how a model learns. Five lessons with no formulas that help you understand everything else. 5 ready
-
(02)
Seeing
Images and video
5 lessons
Classify a photo, find objects, cut them out pixel by pixel and recognize the points of hands and body. With your phone's camera. 5 ready
-
(03)
Listening
Sound and voice
4 lessons
To a model, sound is an image made of frequencies. From there it can recognize noises, keywords and whole sentences. 4 ready
-
(04)
Sensing motion
Sensors and time series
4 lessons
The accelerometer and the gyroscope measure motion a hundred times per second. With those numbers you recognize gestures and activities, spot anomalies and estimate values. 4 ready
-
(05)
Understanding language
Text and language models
4 lessons
Words become numbers, and numbers become meaning. All the way to language models that answer even in airplane mode. 4 ready
-
(06)
From model to app
Optimize and deploy
6 lessons
A model that works on a computer often doesn't fit in a phone. Here you learn to measure it, shrink it and run it without a network. 6 ready
-
(07)
Final project
From question to result
2 lessons
A real case followed from start to finish, then yours, with a card to fill in and a checklist. 2 ready
One lab for all the sensors.
Choose a sensor, collect the examples, then train and measure the model. The phone does it all on its own, inside the browser, and in the end the model answers even in airplane mode.
- ChooseThe sensor and the task. Classify, estimate a number or spot anomalies.
- CollectHold to record the examples, or import them from your gallery.
- TrainThe phone trains a small network and draws the two curves while it learns.
- EvaluateConfusion matrix, the examples it got wrong and milliseconds measured on your phone.
- ShrinkYou quantize the model to 8 bits and compare size and accuracy.
- UseYou try it live, even in airplane mode, and export it if you want to take it somewhere else.
- PhotoCameraA 256 × 256 pixel square
- SoundMicrophoneA one-second mel spectrogram, 40 × 98
- MotionAccelerometer and gyroscopeA two-second window on six axes
- HandCamera21 points with three coordinates
- TextKeyboardA whole sentence, 384 numbers
What a lesson looks like.
- The storyAs you scroll, a 3D scene on graph paper builds the idea one piece at a time. Five to seven chapters, each with a practical tip.
- Go deeperThe numbers, the model names and sometimes the code, if you want to go further.
- Try it on your phoneAn exercise you do with the phone in your hand, in the lab or right on the page.
- QuizA few questions, with the explanation right after you answer. No grades.
- New wordsThe terms of the lesson, linked to the glossary.
I'm Mattia Falco, a software engineer in Bari. This course gathers what I learned building models that run on phones, mistakes included.
I made OliveSense, an app that recognizes olive tree diseases from a photo in less than half a second, with no connection, using a chain of four models of four or five megabytes each. I'm building Edgy Studio, an app to collect and label image datasets right from your phone. The course grew out of a page I wrote in Italian, How to teach a phone to see ↗.
Questions people often ask me.
Do I need to know how to code?
No. You read the lessons like a story and you use the lab with buttons. If you code, the Go deeper sections give you the code and the names of the libraries, so you can rebuild everything in your own projects.
Where do my photos and recordings go?
Nowhere. The lab works inside the browser and saves everything in the phone's memory. The site has no server that receives data, uses no cookies and keeps no visitor stats.
How much does it cost?
Nothing. The course is free and doesn't ask you for an account.
Is there a certificate at the end?
Yes. Your browser draws it with the courses you finished, the name you choose and the date, and you download it as an image or a PDF. It shows what you did in this browser, and it isn't an official qualification.
What about Edgy Studio?
Edgy Studio is the app I'm building to collect and label image datasets on your phone. When it's out in the app stores, the lessons about data will also have a version you can do in the app.