October 27
Can a game see you, and does it see everyone the same way?
You'll leave with: a game you play hands-free, plus real numbers on how reliable its AI is.
- Go to scratch.mit.edu and click Create. Click Add Extension (bottom left) and choose Face Sensing. Allow the camera.
- Put a go to nose block inside a forever loop and click it. Your sprite follows your nose.
- Now try to break it: turn your head, cover part of your face, back away, put two faces on screen.
- Check the box next to face tilt to show its number. How far can you tilt before it loses you?
Name three ways you broke it. Why do you think each one failed?
- Pick a mechanic: steer by tilting your head (when face tilts), or catch things with your nose (when this sprite touches a nose).
- Add a goal, a challenge, and a score.
- Tune it until a brand new player can win within a minute.
- Add a Makey Makey pad as a second input: your face for one action, your hand for another.
- Make the game harder the longer you survive.
A classmate plays your game hands-free. Can they win?
- Pick one face move. Try it 10 times. Count how many work.
- Repeat in three conditions: dim light or a bright window behind you, far from the camera, and a different player.
- Which condition failed most? Change your game to handle it (bigger targets, slower speed, or a keyboard fallback) and test again.
- Take Scratch's Fool the AI challenge from the Face Sensing launch page: find the strangest way to trick it. What does that tell you about how it learned faces?
Our face controls worked ___ out of 10 in normal light and ___ out of 10 in ___. After our fix: ___.
- Share your best and worst condition.
- What did you change, and did the number go up?
- Who might this game not work for, and how would you fix that?
You built a game that sees you, tested it like an engineer, and changed your design so it works for more people. That last step is what separates careful AI from careless AI.
The finale: you design a game and controller for a real player whose body works differently from yours.
Computer vision is AI that understands camera images. It unlocks phones, helps cars find lanes, and lets games react to your body. The engineers who build it spend much of their time finding where it fails.
- Your UHAA mentors are your first stop.
- LA Harbor College in Wilmington has tech and computer pathways, close to home.
- Free local job help: the Harbor Gateway WorkSource Center in San Pedro.
- Put your numbers in your project description: face controls worked 9 out of 10 times after my fix.
- List AI testing and data collection as skills.
- Explain one way the AI failed and what you did about it.
- Show that you think about who technology leaves out. Employers notice.
UHAA is here to help you take the next step toward these careers, from building a resume to making real local connections. Ask your mentors, that is what they are here for.
