📱 → 🥭🥭🥭 ✅ | 🍌🍌🍌 ✅

A phone screen with two rows of photos: a row of mangoes, a row of bananas, each row labelled.

🎒 What you need

one shared phone with a browser and internet, even for 20 minutes
two kinds of object with about 15 of each — mangoes and bananas, or stones and leaves
good daylight, not a dark room

✋ Do this

On the phone browser, open a free 'teach the computer to see' training page (Google's Teachable Machine works and needs no login).
Make Class 1. Point the camera at your mangoes and hold the record button while you turn them around. Get about 30 pictures.
Make Class 2. Do the same with the bananas. Same light, same background — this matters more than you think.
Press Train. Wait. It takes less than a minute. Something just changed inside that phone.
Hold up a mango it has never seen. Watch the bar jump to 'mango'. That bar moving is your proof — point at it.
Now hold up your own hand. Watch it guess wrongly and confidently. Laugh, and remember it for module 6.

💡 Why it works

The phone never learned what a mango is. It learned which patterns of colour and shape were in the pictures you labelled 'mango'. That is why it still answers when you show it your hand — it must pick one of the two boxes you gave it, so it picks the closest. Real farm apps that spot sick cassava leaves are built the exact same way, just with far more photos.

🔥 Challenge

Train it on your face versus your friend's face. Then try to fool it with a hat, a hand over one eye, or standing in shadow. How many tricks did it take?

📖 New words

trainto show a machine many labelled examples so it can guess later

labelthe name you attach to an example, like 'mango'

📸

Model Trainer

Tap when you have finished this module.