JustDuckIt

logo

Inspiration

As coders, we often don’t have the time (or the will) to cook, which leads to one inevitable outcome—fast food, and not always the healthiest choices! We wanted to help fellow coders in workspaces and universities make better eating decisions. That’s where Goose the Duck comes in, our solution to encourage healthier choices without too much thinking involved. Because who doesn't trust a duck? And let’s be honest, when we hit a coding problem, we already talk to rubber ducks for advice (don’t deny it).

P.S. Yes, we know—it’s ironic that the duck is named Goose, but that just makes it even more fun! 😉

What it does

Goose, the not-so-goosey duck, is your personal food advisor. It is trained to help you make better food choices by identifying what's healthy and what's not! Simply present your food, and our duck will let you know if you're about to make a good or bad decision.

How we built it

We used a Raspberry Pi 3 to power Goose’s brain and trained a machine learning model (a CNN) to distinguish between healthy and unhealthy foods based on image classification. Additionally, with a bit of hardware hacking, we created a fun, feathery friend that learned the difference between a kale salad and a Dunkin’ Donuts gazed donut.

Challenges we ran into

Training a duck to recognize food turned out to be trickier than we anticipated! We struggled with finding large enough healthy and unhealthy food datasets, and to make matters worse, our Raspberry Pi often refused to boot at critical moments, making us question not just the project, but our career choices! On top of that, we didn’t receive all the hardware we ordered, forcing us to improvise with the limited resources we had. But in the end, the duck was quacking, and working just as we’d hoped!

Accomplishments that we're proud of

Despite the setbacks, we built everything ourselves—from the coding to the hardware setup to teaching a duck to judge your lunch! It was a proud moment when Goose finally quacked out its first food decision. He's quite the food critic now!

What we learned

We learned that building a project at Hack The North is a lot like feeding a duck—patience and persistence are key! We gained a deeper understanding of food nutrition and the importance of datasets in training models. We also learned (form experience) the intricacies of Raspberry Pi and hardware troubleshooting. Everything related to coding a system to process the webcam display was as well learned on the spot.

What's next for JustDuckIt

Goose might not stop at food! Who knows, maybe Goose will soon help you pick your outfit, plan your day, or even make life decisions like “Should I binge-watch this series or be productive?” We also didn't have a camera available during development, so our next step will also include attaching a camera directly to Goose to make it more independent. We'd also like to add a cooling system to Goose's brain (the Raspberry Pi) in the future to prevent overheating. Unfortunately, we didn’t have a fan available this time around, but it's definitely on our list for next time! We're also excited to add quacky sound effects to Goose in the future, with cheerful 'good' quacks and grumpy 'bad' quacks to keep things fun and engaging!

Stay tuned—Goose is always evolving!