Automating Dried Apricot Sorting with Machine Vision

Sieves and simple color sorters classify product against fixed thresholds. That works when the acceptable range is narrow and the defect is obvious. Dried apricots sit at the other extreme: acceptable pieces span a wide range of shades and shapes, and the difference between an acceptable piece and a reject can be a localized surface deviation rather than an overall property.

Vision-based sorting inverts the batch approach. Each piece is imaged and assessed individually, so the accept-or-reject decision follows from that piece's surface and geometry instead of a single global setting. The defect definition itself stays configurable, because what counts as a reject differs between processors and between export requirements.

A large part of the demand for dried-fruit sorting comes from aflatoxin controls in export markets. Sorting does not replace laboratory analysis — the laboratory remains the reference for actual aflatoxin content. What sorting changes is the share of visually suspect product that reaches those controls in the first place, by removing suspect pieces before the lot is presented.

DOKTEK FoodTech has demonstrated robotic sorting on dried apricots and dried figs; footage is shown in the applications section. Sorting performance depends on the defect definition, the product and the line, so it is evaluated on your samples rather than published as a universal number.