GoldGuard sees every movement.
A compact machine-vision sensor above the showcase: matte black, one gold accent, a central lens, hidden cabling. It belongs in a jewelry store and it sees like an instrument.
Pieces, not pixels.
Segmentation outlines each piece. Classification names it. Tracking gives it an identity across frames. The count is what is visible now.
A hand passes. The piece is kept.
Occlusion is expected on a counter. GoldGuard holds the identity through it and confirms the piece when it is visible again.
Removed. Returned. Recognised.
When a piece leaves the showcase, the track closes and the event is reported with its zone and time. When it comes back, GoldGuard recognises the return.
Intelligent machine vision for the showcase.
Purpose-built for jewelry: small, reflective, overlapping objects, handled by staff and customers all day.
A mask per piece, so touching and overlapping items stay separate.
Rings, necklaces, bracelets, earrings and coins, located in every frame.
Counted across frames, so a passing hand does not change the count.
Identity follows the piece as it is lifted, shown and set down.
A piece that leaves the showcase becomes an event with a zone and a time.
A returned piece is recognised and the event is closed.
GoldGuard Vision is not a surveillance feed. It is a sensor that reports pieces, movement and events to the same timeline as the weight — and hands over evidence windows only when an event needs them.