Inspiration
It is often difficult for people to keep track of food items in their fridge as they begin to expire, resulting in additional food waste. This motivated us to address this issue using a monitoring system.
What it does
Our fridge uses a camera to monitor ingredients present, and employs deep learning to identify food items and generate recipes to recommend using food that is close to expiring.
How we built it
We built our project in two stages; the data-collection stage involving the Raspberry Pi and its associated camera, used to gather images at regular intervals and upload them to a database. Then, we used deep learning to classify food ingredients from the images and display both the image and generate recipes on a website.
Challenges we ran into
Many of the datasets we encountered were too general or too noisy to use effectively. We also ran into issues updating certain variables such as the real-time fridge image.
Accomplishments that we're proud of
We managed to build an actual model for the fridge, and fully automate the process of data-collection and accurate classification of real foods, in addition to having a functional website that allows the user to generate recipes.
What we learned
We learned how to effectively employ transfer learning in a practical project, and gained experience with front-end development and database usage.
What's next for DeepCool
We wish to improve the classification capabilities of our fridge in the future, such as being able to distinguish individual food items from groups.
Log in or sign up for Devpost to join the conversation.