FruGyAi

Stale banana
Fresh orange
stale Apple
Stale banana
stale Orange
Fresh banana
Fresh Apple
3D rendering of the sorting system

Inspiration

The below factors inspired us to create our product, FruGyAI. Coined from Fruit, Vegetable and Artificial Intelligence. The fact that a fruit or vegetable has some form of deformity or discoloration in its appearance doesn't mean it is not edible. Globally, food quality is mostly associated with aesthetics which shouldn’t be so. Example, the fact that I have a big ear doesn’t mean I don’t belong to the human race.

SDG goal 2 talks about 0 hunger and comes up with some solutions; Buy Funny Fruit—many fruits and vegetables are thrown out because their size, shape, or color are not “right”. Buying these perfectly good funny fruits, utilizes food that might otherwise go to waste.

A startup created a social business out of food wastage by collecting discarded fruits/ vegetables and creating a shelter where needy people could collect them.

A search shows that 30%-40% of food is wasted because of their appearance. Source:link

The above inspiration tells us of the need to ensure that food is not discarded because of it's aesthetics and if it's discarded, we can glean from thesse discarded foods and create a social business out of it to provide for the needy.

What it does

FruGyAi uses computer vision technology to sort out bad fruits from good ones. This is made possible after training it on 9,000 + datasets. It uses the AI aspects for the detection and a yet to be implemented mechanism programmed in Arduino to do the sorting.

How we built it

After our research, we looked for and downloaded free datasets from Kaggle on fruits and vegetables and trained the model on some selected fruits( fresh and stale) using lobe(an AI software for training image recognition models).

Once we were satisfied with the outcome(99% accuracy), we decided to develop a mechanism that would do the sorting when an image is detected as stale or fresh using solidworks. Since we had no Arduino or hardware programming background, we just had to simulate how we would want FruGyAi to operate while we look out for experienced hardware programmers to fuse the two systems.

Challenges we ran into

The training was quite a challenge because of the GPU consumption cost, bias and wrong image detection.

We also wanted to build a react app out of it but upon installing some packages, the program threw lots of errors so we decided to halt that aspect.

Accomplishments that we're proud of

We are proud of our progress as a team especially, with the achievement of 99% accuracy of FruGyAi.

What we learned

Through our search, we learned about how food wastage contributes to food insecurities and global hunger. We also learned that, aesthetics of food doesn't always equals quality.