eco food

How our project works
Accuracy and real testing
The dataset that was used for training
All of the 25 different classes that are within our model's capability of recognizing

Inspiration

Given the track of sustainability as well as the growing issues of climate change, one of the biggest problem is that individual action often seems daunting. How can someone truly live a carbon neutral and climate friendly life? Eco food aims to help the climate change issue on this personal level by simplifying and changing the way you think about food.

What it does

By taking a picture on your phone, the picture is then analyzed through a convolutional neural network to classify the picture into multiple categories. Based on these categories, a range of values for the carbon emission impact of the food is displayed.

How we built it

For the convolutional neural network, a dataset of images classified into 25 different food categories was used. The CNN machine learning model was implemented in PyTorch.

Challenges we ran into

Inspiration for an idea was a challenge, especially finding a problem that could be coded in 24 hours (preferably less so that we could sleep) and had enough relevant data to train a model on. Then, the training phase of the model was also challenging in varying parameters to get the highest accuracy. Finally, the ML demo app was a challenge as it was our first time using Xcode.