Food waste is an unnecessary issue.
Based on the MIT Food Cam, we decided to put our own twist on it with intelligent automation.
What it does
Instead of having to check the live feed frequently for food, Foodi leverages artificial intelligence to notify users on what food has arrived!
You like donuts? Someone brought Donuts? Boom, instant notification. Go and grab your donuts, just like that!
How we built it
The project is composed of two major components:
- Live Camera:
- Live feed is ran on a Raspberry Pi Zero W with Motion
- Snapshots are taken at set intervals
- A Python script sends the image to Google Vision API for object detection
- Data is then parsed to Firebase Firestore
- Mobile App:
- Developed with Android Studio
- Information is pulled from Firestore and notifies the user
Challenges we ran into
- Getting the AI to correctly identify objects
- Android Firebase Project error that no one on the internet knew how to fix
Accomplishments that we're proud of
- Getting the Python script to send and receive data from our camera to Google Vision to Firestore
- Getting the AI to recognize foods correctly (more or less)
- Fixing the Firebase problem above
What we learned
- Training an AI to recognize something is a huge pain
- Android Studio has come a long way
What's next for Foodi
We aim to create a network of cameras so you can help curb food waste locally!