[O] Anti-Pineapple Pizza Detection

Inspiration

Historically speaking, pineapple on pizza is not something that people can accept. Italy, the hometown of pizza, even set it illegal. The State of Italy has unilaterally announced that putting pineapple on pizza under any circumstances is tantamount to an Act of War under International Law. Therefore, we decided to use AI-based technology to detect whether there are pineapples or not. With these methods, the market can detect what they are going to do with those pizzas and decide how seriously they will treat them (throw them in the trash can or feed it to animals.) On the other hand, we can help those special populations (blind, anosmia, etc.) who are against pineapple on pizza to notice it better and build the defense.

What it does

Users can paste a snippet of base64/encoded image. Our program can output how likely (how dangerous) it is to have pineapple on it.

How we built it

We used Google Colab and TensorFlow to run our model ~and used google cloud to host that as a website~. Since ML is data-hungry, we wrote some scripts to collect data from the Internet.

Challenges we ran into

The internet is not working well at my house, so our work has been disrupted several times. We finished this project after we were running and worked outside for a whole day.
It is Ziqian’s first time doing a Hackathon.
Additionally, there are a group of people in this world that support pineapple on pizza, so our project is a little bit too controversial in their opinion, and they might do something to stop us from doing this.
Pineapple and pizza are both yellow, so it is harder for AI to detect the difference.
We do not have enough data resources.

Accomplishments that we're proud of

We chose the hardest field in CS - machine learning. And during this hackathon, we learned a lot about image processing, training AI models, and using python libraries. We used an open-source data set.

What we learned

How does CNN (convolutional neural network) work
How to use google colab more fluently
How to use TensorFlow more fluently
What is a general Hackathon process ( for our new teammate)

What's next for Anti-Pineapple

Since 24 hours is not enough to train a model and collect enough data, we will work on it further by training for a longer time and collecting more data. We might also need to twist our neural network in the meantime. And after this network is working well, we could make it a website or app to make it more convenient for people to use. We could also enlarge the field to detect other things like whether there are strawberries in Chill or Noodles in a Pie. Because we solved the problems of similarity in colors between pineapple and cheese, we could also use it to detect other similar color foods.