Pharmaceutical packaging
Catch missing tablets, blister-foil defects, label errors, and serialisation faults before packs leave the line.
Automated quality inspection for pharmaceutical packaging, running on a refurbished iPhone alongside the blister machine, the cartoner, and the serialisation station.
What is automated quality inspection for pharmaceutical packaging?
AI defect detection for pharmaceutical packaging uses a camera and an AI model to watch every pack as it leaves the blister machine, the cartoner, and the serialisation station, and to flag non-conforming units before they reach the cartoning queue. Instead of relying on a line operator at the inspection station or on rigid rule-based vision, the AI learns the specific blister format, tablet shape, carton artwork, and serialisation layout of your SKU portfolio, and applies a consistent visual checkpoint across shifts, line speeds, and SKU changeovers.
Pharmaceutical packaging is particularly hard to inspect at line speed because every batch must comply with EU GMP Annex 1, EU FMD, DSCSA, and the ICH guidelines for batch records. The pack itself is variable: a tablet blister sits at a different angle from a vial pack, a printed laminate looks different against the conveyor than against the cartoner plastic, and the DataMatrix code on the carton can shift by a millimetre and still pass the verifier but fail the pharmacy scanner. Rule-based vision built around a single SKU breaks the moment you swap to a different blister format or a different tablet shape. AI-led inspection handles those variations because the model learns from real production frames rather than from a fixed threshold.
The result is an automated visual checkpoint that complements your existing GMP inspection station and your serialisation reader, and gives you a pack-by-pack image record that ties into your batch record. When a regulator query comes back six months later, you can pull the frames from the exact production window in the batch and either confirm the defect or push back with evidence.
Defects we catch on pharmaceutical packaging lines
Missing or broken tablets
Missing tablets are blister cavities that arrive empty at the lidding station, caused by feeder-bowl misalignment, cavity-clearance jams, or upstream tablet press deviations. Broken tablets are the partial-fill version, where a chipped, cracked, or split tablet sits in the cavity. GMP inspection cameras catch the gross deviations but miss the cases where a fragment fills the cavity correctly enough to pass the count-by-cavity sensor. The AI model learns the in-spec tablet shape and colour for each SKU and flags both missing tablets and broken tablets at the lidding exit, with the frames available for the batch record.
Blister-foil pinholes and seal defects
Foil pinholes are micro-holes in the lidding aluminium, caused by aluminium-stock pinhole carryover, sealing-jaw debris, or heat-seal temperature drift. Seal defects are the complementary failure: a wrinkled or weakened seal that lets moisture into the cavity over the storage life. Both are nearly invisible at the line and ruin the product weeks later. The AI model learns the in-spec foil texture and seal pattern for each blister format and flags pinholes and seal anomalies as soon as the local pattern deviates from the reference.
Cavity-fill and orientation errors
Cavity-fill errors include cavities with the wrong tablet (mix-up across product changeovers), tablets sitting at the wrong orientation (split tablet visible vs. score side), or tablets pressed into adjacent cavities. Orientation errors look correct from the count-by-cavity sensor but fail the pharmacy at dispensing. The AI model holds the in-spec tablet appearance and orientation for each SKU and flags both mix-up and orientation errors at the lidding exit, before the pack continues to the cartoner.
Carton label and artwork defects
Carton defects include print smears, missing colours, off-centre artwork, missing braille embossing, and incorrect SKU artwork loaded onto the cartoner. Hospital pharmacies reject cartons for missing braille, and the regulator audits print fidelity on every batch. Manual operators catch the worst cases but miss the borderline placements and the gradual print-quality drift after a cartoner ink-roll change. The AI model holds the in-spec carton artwork for each SKU and flags artwork defects as soon as the local pattern deviates.
Serialisation and DataMatrix legibility
Serialisation defects include faded DataMatrix codes from a worn inkjet head, low-contrast prints on textured carton stock, codes that shift outside the spec area, and missing tamper-evident features. EU FMD and DSCSA both require legible, scannable codes at every supply-chain step. The verification scanner on the line catches some defects but misses the borderline cases that pass the line and fail at the wholesaler. The AI model OCR-checks every code at the inspection station and flags both legibility and contrast issues before the pack enters the aggregation tunnel.
Patient information leaflet defects
Leaflet defects include missing leaflets, wrong-language leaflets in a multi-market run, misfolded leaflets that fail the cartoner insertion, and leaflets with print defects. Hospital pharmacies reject cartons with wrong-language leaflets, and the regulator audits leaflet fidelity on every batch. Manual operators catch missing leaflets but miss the borderline language errors and the misfolded inserts that survive the cartoner. The AI model learns the in-spec leaflet appearance for each SKU-market combination and flags leaflet defects at the cartoner infeed.
The lighting setup that makes this work on a pharmaceutical packaging line is a diffuse overhead light over the blister exit to read tablets, foil, and seals, plus a low-angle ring light at the cartoner to read artwork and braille and a side-mount camera at the serialisation station for code legibility. An iPhone Pro with macro and wide-angle lenses handles the seven defect families from a single inspection station per critical control point. We synchronise the rig with the conveyor encoder so flagged packs drive a downstream divert decision before the aggregation tunnel. We spec the optics with you during onboarding and design the rig to sit alongside, not replace, your existing regulated inspection systems.
How Enao runs on a pharmaceutical packaging line
The full hardware rig costs less than €1,000 and consists of a refurbished iPhone Pro, a diffuse overhead light with an optional low-angle ring light for foil and artwork inspection, a USB-C cable, and a mount that clamps over the blister exit, the cartoner outfeed, or the serialisation station. PLC integration is not required for the first deployment, the rig fits in a flight case, and the line keeps running while you set it up. The rig sits alongside, not in place of, your existing GMP inspection station and your serialisation reader.
Onboarding is self-serve. Your line team mounts the rig, opens the Enao app, and starts collecting reference frames at the next changeover. Day one returns 80% accuracy without any prior labelling, and by day fourteen the model is operating above the manual inspector on the defect families it has seen, improving with every flagged pack that the line confirms or rejects. All flagged frames feed into your batch record for traceability.
Each line teaches its own model what its blister format, tablet shape, and carton artwork look like. When you swap to a new SKU on the same machine, the model adapts in a single shift. When you bring a sister line online with a similar product family, the second model starts from the first model's experience and the marginal effort drops sharply.
Out-of-spec packs stop reaching the aggregation tunnel, deviations are logged at the inspection point with image evidence rather than at the QA office, and your operators get back the hours of attention they need for the parts of the job that still need a human, including line setup, GMP changeover validation, and regulator audits.
How Enao compares to manual inspection and traditional machine vision
For pharmaceutical packagers the comparison sharpens around five dimensions, with the explicit caveat that AI inspection complements rather than replaces regulated GMP systems.
Setup time on a pharmaceutical packaging line. — Manual checks at the cartoner miss subtle blister and leaflet defects. Traditional machine vision (Cognex, Overview.ai, Globalvision, packaging-gateway) requires three to nine months of integration and a six-figure budget. Enao is deployed in a week by your own team on a refurbished iPhone, day one at 80% accuracy.
Hardware cost per line. — Manual visual inspection: none upfront, ongoing labour cost. Traditional pharma vision: €150,000 to €500,000 per line for industrial cameras, multiple inspection heads, and integration. Enao: under €1,000 per line with a refurbished iPhone Pro, lamp, and mount.
Handling new SKUs, blister formats, and tablets. — Manual visual inspection: re-train operators for every new SKU. Traditional pharma vision: rewrite the recipe per SKU plus full validation, often outsourced to the integrator. Enao: re-teach the model on new SKUs in a single shift, no code to touch.
Detection accuracy on subtle foil and artwork drift. — Manual visual inspection: high at shift start, drops measurably after three hours. Traditional pharma vision: strong on count-by-cavity and serialisation reading, weak on subtle foil-pinhole detection and artwork drift. Enao: learns foil and artwork signatures from reference frames and holds accuracy across shifts and runs.
Who runs it. — Manual visual inspection: trained line operator. Traditional pharma vision: system integrator or a specialised vision engineer. Enao: your line team, with the rig sitting alongside your existing regulated systems.
Hospital pharmacies and wholesalers escalate over the cost of a missing-tablet recall, and the cost of a regulator finding sits well above the cost of an iPhone-based inspection rig that sits alongside your existing systems. Enao is built for that gap.
