[ PROJECT 1 ]
Waste sorter: dry or wet?
The problem: Who: school housekeeping staff. What: waste is put in the wrong bins. Where: the canteen. Why: correct sorting helps recycling.
Fill a 4Ws canvas Photograph 60+ items per class with varied backgrounds and light Train an image model and test on 30 new photos Report accuracy per class and improve the weakest class Students hand in: Working model, accuracy table and lifecycle poster
AI Project Lifecycle AI Applications with No-Code Tools
[ PROJECT 2 ]
Plant doctor: healthy or diseased leaf?
The problem: Gardeners and farmers need a quick first check for leaf disease.
Scope the problem and agree on 2–3 classes Collect and label leaf photos; check balance Train, test on unseen leaves, and calculate accuracy Write a “limits” note: when should a person not trust it? Students hand in: Model demo and a responsible-use note
AI Project Lifecycle Data and Fairness
[ PROJECT 3 ]
Sound alarm: recognise the school bell
The problem: Students with hearing difficulty miss audio signals.
Record the bell, announcements and background noise as classes Train an audio model Test in different rooms and at different volumes Discuss consent and privacy for recording voices Students hand in: Model demo and test log
AI Applications with No-Code Tools Ethics and Responsible AI
[ PROJECT 4 ]
Fairness audit of our own model
The problem: Check whether a class-built model works equally well for everyone.
Take any class model (e.g. a gesture model) Test it with students of different skin tones, hand sizes and lighting Compare accuracy for each group in a table Retrain with added data and measure again Students hand in: Fairness report with before/after accuracy
Data and Fairness Ethics and Responsible AI