Diets are often hard to follow, due to foods which are too accessible and their contents being not what they seem to be. Therefore, having an app which can put your attention on these dietary traps would be quite useful.
What it does
Giving you a better insight, how big the calorie intake actually is, especially for foods, which seem to be rather unproblematic. This way you can learn to better estimate which foods are actually to avoid which you usually wouldn't. This can lead to weight loss if overweight, and to a better ability to plan for cooking meals, therefore reducing food waste.
How we built it
Tensorflow models for object recognition, swift application for making the phone app, server client based communication, git for version control.
Challenges we ran into
Training sets were really bad, use of new software, volume estimation is actually way harder than we expected.
Accomplishments that we're proud of
We didn't give up, we got a prototype running, which works approximately as we want
What we learned
Having a variety of operating systems may not be the best idea for a Hackathon, but is not impossible.
What's next for Slim down the planet
Collect better datasets, fix bugs, improve accuracy.