• Portada - Detecting external quality in orange with IngiGrader

Blog

Grading external quality oranges

Detecting external quality in orange with IngiGrader

In recent months we have strengthened the external quality detection module specifically applied to oranges.

We had a very broad dataset applied to clementines and, based on this basis, we have reinforced the system with a specific dataset of more than 100,000 multiple captures, both in NIR and in visible light.


External quality in oranges
Fig 1. Some examples of oranges classified automatically by the system with different defects and levels of affectation

With this development we have adapted the decision module to the specific problem of the orange, for this new model the following have been taken into consideration: external appearance, shape of the fruit or level of affectation in IR.

The sorting module has been oriented to obtain 5 global quality levels so that the lowest levels include the largest defects and the highest levels include the highest quality fruits. Oranges of the varieties were included in the training: navelina, navel, lanelate and navelate.