Robots and Strawberries: A New Way to Measure Weight

Robots are stepping into the strawberry fields to help with grading. This task is usually done by hand, but robots could make it faster and gentler on the fruit. The key to grading strawberries is getting their weight right. So, how do these robots figure out the weight of a strawberry? First, let's talk about the data. A bunch of strawberry pictures were taken using a special camera. This camera captures both regular images and depth information. The weight and size of each strawberry were also measured by hand. This data was used to train a model that can guess the weight of a strawberry just by looking at it. Now, here's where it gets tricky. Sometimes, the depth information in the pictures is missing because of things like light or shadows. To fix this, a new method was created. This method uses the regular images to fill in the missing depth information. It finds the strawberry in the picture, figures out its center, and then uses this information to complete the depth image. Pretty clever, huh? With the depth information fixed, a new model was built. This model, called MMF-Net, uses both the completed depth image and the regular image to guess the weight of the strawberry. It combines the strengths of both types of images to make a more accurate guess. Now, let's talk about how well this model works. It was compared to other models, both old-school and new-school. The old-school models include things like Support Vector Regression and Random Forest Regression. The new-school models include things like PointNet and EfficientNet. The results showed that the MMF-Net model was the most accurate. It beat out both the old-school and new-school models. So, what does this all mean? Well, it means that robots could be the future of strawberry grading. They could make the process faster, cheaper, and gentler on the fruit. But, there's still more work to be done. The models need to be tested in real-world conditions, and the robots need to be trained to pick the strawberries without damaging them. It's a work in progress, but it's an exciting one.