WESORT Introduces OmniPearl 360 AI Pearl Sorting Machine: Every Pearl, Every Angle

The hardest part of pearl sorting is not seeing a pearl once. It is seeing enough of its surface to make a consistent decision.

A dark spot may sit on the back. A fine crack may become visible only when the pearl turns. Two pearls may appear similar in color but differ in luster or orient. Round pearls, elongated rice-shaped pearls, short rice pearls, and button-shaped pearls also move differently through a sorting system.

These are the challenges behind WESORT OmniPearl 360, a new AI pearl sorting machine developed for multi-angle surface inspection and fine pearl grading. By combining dual-camera image acquisition with controlled pearl rotation during transport, OmniPearl 360 brings more of each pearl's visible surface into the same inspection process.

Why Pearl Sorting Needs More Than a Single View

Pearls are curved, reflective, and naturally variable. When a vision system captures only one fixed view, visible defects on the side or back may remain outside the inspection area. Operators are then left to turn and recheck pearls manually after sorting.

The OmniPearl 360 AI pearl sorting machine changes the way the pearl is presented to the cameras. Two fixed cameras collect images while the belt-driven transport system allows each pearl to rotate as it moves forward. Different surface angles enter the field of view in sequence, expanding inspection coverage and reducing the blind areas created by a fixed pearl position.

WESORT OmniPearl 360 panoramic AI pearl sorting machine demonstration

AI Vision Brings Small Surface Defects Into the Sorting Process

Black spots, dark marks, and fine surface cracks can be small, irregular, and affected by reflection. During long manual shifts, inspection results may vary with operator experience, fatigue, and individual grading habits.

OmniPearl 360 applies industrial AI vision to the high-resolution images collected during rotation. Its distributed inference design supports image analysis close to the acquisition process, while the sorting model is configured around the defects and grading samples confirmed by the customer.

This distinction matters. The system is designed to inspect visible surface characteristics; it is not presented as a method for detecting hidden internal defects. The practical objective is to make confirmed surface-grading rules repeatable across continuous production.

Beyond Color: Luster, Orient, and Customer-Defined Grades

Pearl value is not determined by color alone. Subtle tone differences, luster intensity, and layers of orient may all influence the final grade. A useful pearl color sorter therefore needs to work with the processor's own product standards rather than impose a generic definition of “good” and “bad.”

Using customer-provided grade samples, OmniPearl 360 can be configured to evaluate color, luster characteristics, orient, and other recognizable surface features. Greenish or reddish tones, slight color differences, and disputed boundary samples can be reviewed during testing before the classification rules are put into production.

Designed for Round Pearls and Multiple Specialty Shapes

A pearl grading machine must also account for shape. Round pearls do not travel in the same way as long rice, short rice, or button-shaped pearls. Their contact points, rotation patterns, and visible angles can differ, which means that one setting should not be assumed to fit every material.

OmniPearl 360 is designed to support round pearls and multiple specialty shapes through sample-based configuration. Pearl type, size range, feeding condition, target defects, and grade boundaries are reviewed together so that the sorting setup reflects the material that will actually enter production.

From Manual Rechecking to Repeatable AI-Assisted Grading

The purpose of an AI pearl sorting machine is not to erase the expertise of pearl graders. That expertise is still needed to define acceptable pearls, rejected pearls, and difficult boundary samples. The opportunity is to convert those confirmed decisions into rules that a machine can apply repeatedly.

For a pearl processor, this can reduce repetitive turning and rechecking, support more consistent grading between operators and batches, and make changeovers easier to manage when the relevant settings have been verified. The outcome still depends on the material, feeding condition, target features, and agreed grade definitions.

How WESORT Validates a Pearl Sorting Application

OmniPearl 360 is entering sample testing and application validation. Instead of publishing a fixed performance claim for every pearl, WESORT evaluates each project with representative material.

1. Provide representative samples. Include the pearl type, shape, size range, and typical material variation.
2. Define the target. Mark accepted pearls, clear defects, and disputed boundary samples for black spots, dark marks, fine cracks, color, luster, orient, or shape.
3. Reproduce the real grading task. Explain current rechecking steps, product changeovers, and the production conditions the machine needs to support.
4. Review the test together. Compare accepted, rejected, and boundary outputs before confirming model settings and the equipment configuration.

Sorting results depend on pearl type, size, visible surface condition, feeding behavior, target features, and the customer's grading standard. Final recommendations should be based on representative sample testing.