A successful day at Leonardi

We recently had the pleasure of spending the day at the caterer Leonardi in Munich. The purpose of the visit was to verify whether NUA EYE was a good fit for a self checkout station at one of their locations, serving primarily salads, bowls and a diverse number of toppings. Anu, Sean and I set off to work early.

As a first step, we decided to learn the different shapes and sizes of the used dishes. This would allow us to use the “auto annotate” feature of NUA EYE later on, when teaching the system the actual food dishes. As the customer places a lot of value on a lean process before service, i.e. spend as little time as possible maintaining the system and learning new dishes, the decision made sense.

The next step was to move on to learning the actual products… salads, pasta, pizza and a number of possible toppings to choose from in small and large glass containers. The added difficulty for us was always the presence of a lid on the glass containers. In addition, many of the toppings looked very similar to the naked eye. We proceeded by taking between 5-10 images per product, capturing essentially different camera angles. Then it was time to train the model and a well deserved coffee break.

After a couple of minutes, the training process was completed. Time to test the accuracy!

As expected, NUA EYE, successfully recognized all the products (never doubted it :-)), even with the added difficulty of lids on the toppings. We recorded a video together with the customer of a few highlights. Hope you enjoy it!

Many thanks to Tobias and the crew at Leonardi for taking such great care of us on the day! More information on Leonardi can be found here: https://leonardi-kg.de/

Best Michael