We were inspired by the potential of Machine Learning and the internet of things. The smart refrigerator was the perfect use case for this. It is a challenging task that would provide a significant level of convenience and that greatly inspired us to work on this project.
What it does
iFridge is a smartphone application that allows the user to keep track of his/her items in the refrigerator on the go. It also provides recipe recommendations to the user based on the ingredients in the fridge.
How we built it
We developed a image recognition model by collecting image data manually and processing them with TensorFlow. Using a Raspberry Pi connected to a camera module, we were able to set up an image stream that provided us with a constantly updated view of what is in the fridge. We then set up a back-end server that processes this image and updates the smartphone application with new data.
Challenges we ran into
The most significant challenge we had was the difficulty in training the image recognition model such that it could recognize all our objects accurately.
Accomplishments that we're proud of
Training and developing our own image recognition models without using readily available libraries.
What we learned
We learned how to collect and process real world data to recognize image entities in a real world situation. Moreover, we learned how to make use of Raspberry pi
What's next for 003 - iFridge
To improve our image recognition by training it on a larger data set with a larger variety of items. We also plan to support multiple object recognition with more advanced machine learning models.