Metal Stamping
Catch burrs, splits, springback, and dings before parts leave the press shop.
Automated quality inspection for sheet-metal stamping, deep drawing, and progressive-die work, running on a refurbished iPhone next to your press.
What is automated quality inspection for metal stamping?
AI defect detection for metal stamping uses a camera and an AI model to watch every part as it leaves the press, the secondary station, or the wash, and to flag non-conforming parts before they reach the rack. Instead of relying on an operator at the panel or on rigid rule-based machine vision, the model learns from images of conforming and non-conforming parts on your line, and it adapts as your die, your coil, and your lubricant change.
The shop floor calls this in-line visual quality control, AI-based defect detection, or AI vision for metal stamping. The technology family is the same: a fixed camera, a controlled lighting setup, an AI model trained on examples from your line, and a traceability record that every strike was inspected and either accepted, flagged, or rejected.
What it does not do: replace your die maintenance, your tool engineer, or your customer audit. What it does do: make sure the part counts you ship match the part counts that pass spec, every shift, on every die, with a record you can show the auditor when a customer complaint comes back.
Defects we catch on metal stamping lines
Burrs at cut and pierce edges
Sharp slivers of metal left along a punched or trimmed edge. Caused by punch and die wear, by a punch-to-die clearance that is too tight or too loose, or by a die that has chipped. A camera with raking light along the part edge picks up burr height before the operator can run a fingertip across it, and tracks the wear curve so you re-grind the die before the customer rejects start.
Splits and tears at deep-drawn corners
Hairline cracks at the corner of a deep-drawn cup, a drawn ear on a panel, or a bent flange. Tells you the blank holder pressure is wrong, the lubricant has dropped off, or the coil is at the edge of spec. A camera looking at the same corner on every part catches a split the moment it forms, instead of waiting for the customer to crack-test the assembly.
Wrinkles on flanges and side walls
Folds and ripples on a side wall after a deep draw, or on a flange after a hem. Caused by too little blank holder pressure, an unbalanced die, or a coil thickness drift. A camera spots the wrinkle pattern even when the part still meets the dimensional check, and warns the operator before the next coil makes the problem worse.
Springback and dimensional drift
The part leaves the die the right shape, then springs back to a slightly different angle as the elastic stress relaxes. Tells you the coil yield strength has drifted, the die wear has changed the bend radius, or the lubricant has changed the friction. A camera with a fixture or shape print catches the drift the day it starts, hours before the first customer reject.
Surface scratches, dents, and dings
Lines, divots, or impressions on the visible face. Caused by a metal chip lodged in the die, by a transfer arm rubbing the part, or by a stack mark in the rack. Most miss the operator because the next part hides them on the conveyor. A camera at the exit chute catches them shot by shot, even on high-shine surfaces where a human eye gets glare.
Coating and plating defects
Zinc, paint, or e-coat coverage that is too thin, runny, or pinholed on a finished stamping. Tells you the bath chemistry has drifted, a spray nozzle is partially blocked, or the part hung wrong on the rack. A camera trained on your specific part picks up coating gaps and runs the same shot a human inspector would run, with no fatigue.
That is the starting list. During onboarding we calibrate which of these classes matter most on your specific line and tune the model accordingly.
How automated visual inspection runs on a metal stamping line
A press cell that runs visual inspection on Enao looks like the cell next door, with one extra component. A refurbished iPhone is mounted on a stand with a downward or angled view of the exit chute, the conveyor, or a dedicated inspection fixture between the press and the rack. A simple LED bar gives the camera the same light at every strike.
When the part lands on the conveyor, the camera takes a picture. The model on the iPhone classifies the part as OK or as one of the seven defect families above, and writes the result to your traceability log. If a die gives you twenty flagged parts in a row, the operator gets an alert; if a press shows a slow drift on burr height across the day, the dashboard flags it before the customer does.
The model retrains overnight on the previous day's labels, so a die change, a coil change, or a lubricant change is absorbed in one shift instead of one quarter. New part numbers go through the same flow: the operator labels the first hundred strikes, the model takes over from strike one hundred and one, and the tool engineer reviews the labels at the end of the shift.
AI vision vs manual checks on metal stamping lines
Lines that move from manual operator checks or rule-based machine vision to AI-led inspection see the same step changes regardless of part geometry or coil grade.
Detection rate on subtle defects — Traditional machine vision (Overview.ai, Ciclo Vision, Solomon-3D, iFactory) needs labelled image libraries before it ships, and a six-figure integration. Enao reaches 80% accuracy on day one with no labelled data, then climbs past 95% as your operators tag a few hundred examples on the iPhone.
Time to handle a new die or part number — Manual: Operator briefing, golden samples, paper QC sheet. Two to four weeks until the floor reads the new part fluently. Enao: Hundred labelled strikes and the model is running. Same shift, no paper sheet to update on every press.
Traceability when a customer comes back — Manual: Hand-written log on a clipboard, partial coverage, missing shifts. Reconstruction takes a week. Enao: Every strike logged with image, classification, and confidence. Reconstruction takes ten minutes.
Cost to get running — Manual: Adds an inspector per shift per press, recurring monthly cost on top of training. Enao: Hardware under €1,000 per cell. The cost stays flat while the line scales.
Behaviour when the die wears — Manual: Gradual rejects climb until the customer flags it. Days of root-cause work to find the moment. Enao: Dashboard shows the burr-height drift the day it starts. Tool engineer has the time stamp and the image.
