Paper Cup Machine Vision Inspection FAQ

What defects can a vision system catch on a paper cup line?

The reliable classes are side-wall tears and creases, print misregistration, rim curl faults, bottom and side seam gaps, and surface stains. Rim and bottom defects are the hard cases because they sit on curved, reflective, or occluded surfaces, so they need dedicated lighting geometry and often a second camera angle. Blind spots come from where the light is placed, not from the software — if the light does not reach a defect at a useful angle, no algorithm will find it.

What camera resolution does a cup inspection system need?

Size resolution to the smallest defect you must catch. The sensor needs enough pixels to span that defect across the field of view — a 0.1 mm resolution target on a cup side wall typically needs a multi-megapixel sensor and a lens working distance that fills the frame with the cup surface, not the whole line. Buying resolution below the defect size produces a system that looks installed and rejects nothing, and resolution cannot be recovered later in software.

What lighting works best on a curved, printed cup?

Flat front lighting is the trap. Curved, reflective, and printed cups hide defects under diffuse flat light, so most lines use dark-field lighting for side-wall tears and creases, a dome or angled source for the rim, and a dedicated view for the bottom seam. Control ambient light at the station and lock a separate recipe per printed SKU, because a register mark and a surface stain need very different thresholds to separate reliably.

How do you synchronize reject actuation with line speed?

The reject signal must be a position-locked trigger tied to the encoder, not a timer started when the image is captured. Fix the camera-to-actuator distance, verify the delay across the full line-speed range, and confirm every removal with a sensor at the reject chute. On a line that changes speed for stock reasons, a fixed time delay drifts and starts rejecting good cups beside a defect while letting the defect through — an unverified reject is only a hope.

How do you keep false rejects under control?

False rejects trace to lighting and recipe drift, not to the camera. Tune thresholds against a documented good/reject sample set at production speed, keep one recipe per SKU, and re-verify detection after any stock or supplier change, because fiber formation, moisture, and coating shift how a defect reflects light. A false-reject rate above 2% usually means the threshold was set on a still sample under office lighting rather than a moving cup under working light.

Why does the reject log matter more than the reject count?

A count tells you the line is losing cups; a log sorted by class, camera view, and time tells you why. It shows whether tears cluster after a reel change, whether seam gaps follow one forming station, or whether stains appear only on one printed SKU. Wire the log back to the forming station and to the line's OEE board — a detection system that only counts rejects has not finished its job until the fault stops being made. Sources: ISO quality management, UL Solutions machinery safety, ASTM imaging standards, TAPPI paper test methods.