Three of us live on the same floor (Stever 2A) and no matter what the time is, you can always find food in the lounge, waiting to be eaten. The amount of food that goes straight into the garbage is enormous. There are a large number of people who could use that food, such as upperclassmen without meal plans, and (more seriously) the homeless.
What it does
It takes pictures of food through a webcam, displays the pictures on a website, identifies the food in the picture and displays it as text, and is capable of sending text alerts specifying what kind of food is present whenever food is added or removed.
How we built it
We agreed to use Python because everyone at least kind of knew how to use it. We used OpenCV to access our webcam and captured frames for analysis. We analyzed the images by using Clarifai's Food Model API, which sent us back a list of food items and the probability that they were present in the image. We used the Flask Python web server along with Socket.IO to serve the dynamic content (the images and food names) and we used Twilio to send text messages whenever food was added or removed.
Challenges we ran into
The video capture and clarifai api went quite smoothly, but none of us are experienced in web development, so that took a long time to figure out. For a long time, we had trouble loading images dynamically through socket.io on the website.
What's next for MealingFull
Polishing is definitely necessary, and we could definitely add more features, such as logins and points, to reward people for donating food.