After letting food rot in the barren, cold wasteland that is the back of our fridge, we could not take it anymore.
What it does
Helps you track your fridge foods.
Having a database of your fridge contents allows for many applications. It leads to the answers of questions such as:
- How many carrots do I have?
- When will my milk expire? (Will notify via text once expiration nears, for example)
- Do I have all the necessary ingredients to make apple pie?
- What are some of my eating habits?
- Using what is in my fridge, how can I mix things up? (Recipe recommender)
How we built it
Python, a cloud (AWS), tape and a box glued together with our tears, caffeine and apples.
We used Tensorflow to retrain an image neural network to work with our dataset, which we needed to scrape for. We decided to run the training on AWS to reduce training time with GPUs. Python's OpenCV helped a lot with image processing, which we needed to use with our webcam to track things moving in and out of the "fridge". Then, using the data to produce something of value, a report of sorts and a question/answering mechanism, was the last step.
Challenges we ran into
Image processing is confusing. Math. Different setups produced different results. (Docker would have been nice to use)
Accomplishments that we're proud of
We got a nice working demo in the end :D
What we learned
Never underestimate the seemingly easy things - one missed detail or one rushed action affects many hours
What's next for Bubble Wrapped Ice Box 3000
We think it is a great convenience at little cost to implement. The infrastructure of cameras on fridges already exists. We need to work on adding more features to allow it to provide further convenience.