AI Enhanced Barcode Scanners

 In Barcode Advice

Previously, we wrote about how AI and barcodes work together. To review:

• Using computer vision, AI-enhanced scanners can read partially damaged barcodes, or barcodes in poor lighting or partial visibility
• AI can use barcode scan data to optimize retail product placement, predict product demand and enhance inventory supply chain efficiency
• AI analysis of barcode data can improve the efficiency of warehouse operations regarding product storage and picking.

The AI behind super-aggressive scanners demands careful consideration, and not just for the other listed applications. Recently we received a marketing email from a scanner manufacturer promising “…no more barcode problems.” Let’s think about that.

AI – scanners: No More Barcode Problems

To be clear, the ability to accurately scan a poorly printed, or partially damaged, or somewhat obscured barcode is a good thing. The key, however, is the adjective “accurately.” Where is the threshold of acceptability? At what point does an AI-enhanced scanner fail to decode an impaired barcode? It’s a question not likely to get an answer from the scanner manufacturer—but it’s an important question.

Which is Worse?

Which is worse, scanner that doesn’t beep after the trigger is pulled, or one that does? Conclusion: the scanner that beeps but incorrectly decodes the barcode is far worse. See this article.
How is an AI-enhanced scanner able to read bad barcodes? It pushes the ISO print quality parameters, moving the threshold between the high and low reflectance values of bars and spaces, interpolating bar and space widths from excessively gained images, and who knows what else.

Killing the Feedback Loop

The performance standards of non-enhanced scanners create a subtle but important feedback loop. When barcodes fail to scan, information quickly flows back up the supply chain to the brand owner and package or label printer. Fines and reparations follow quickly. AI-enhanced scanners complicate this feedback. Barcodes that conventional scanners can’t read are decoded effortlessly by AI-enhanced scanners. Which ISO tolerances were tolerances eclipsed? By how much? You don’t know.

Who is taking the risk?

What about barcodes that don’t scan well with conventional scanners and get rejected, but those same barcodes that are mis-decoded by AI-enhanced scanners? The damage done by the former pales compared with the damage done by the latter.
• When a barcode fails to scan, the feedback loop responds instantly. Product sales stall, anticipated inventory is halted—or piles up. The problem is quickly remedied and the damage is contained.
• When a bad barcode is mis-decoded, the AI-enhanced scanner beeps. Everything seems fine. If the mis-decoded data represents a different product, inventory is incorrectly debited and replenished. Sales data is corrupted. The misidentified product is depleted but not replaced. Customers are unhappy. The feedback loop doesn’t know there is a problem for days or weeks. The damage can be considerable. Who is taking the risk? You are.

 

Risk Management

Barcodes that do not scan well—or at all—are a significant liability. And not just at Walmart. Failing UDI barcodes can trigger medication recalls. Incorrectly decoded UDI barcodes are at least as bad and arguably worse than barcodes that fail. People can be sickened—or killed.

Who is Responsible for AI Hallucinations?

Your barcodes are a promise. Broken promises are a liability. At the end of the day, barcodes must comply with print quality and data presentation standards. AI enhanced scanner performance is a good thing—until it isn’t. It is reasonable to speculate that no AI-enhanced scanner manufacturer will accept responsibility for mis-decoded barcodes that are not ISO compliant. They cannot really promise “no more barcode problems.” More accurately, they are promising “no more small barcode problems.”

Does AI Negate Verification?

It might seem that AI-enhanced scanners eliminate the need for verification, but the opposite is true. AI-enhanced scanners will mask barcode problems until they are so bad and have inflicted so much damage that they can no longer be ignored. Who is responsible? Here are the usual suspects, in serial order:

• The brand owner
• The package or label printer or printer-subcontractor
• The graphics designer
AI doesn’t change that.

Barcode verification is like insurance. It seems expensive until you need it, and then it pays for itself. A barcode testing service can provide instant expertise for a fraction of the cost of a barcode verifier,https://barcode-test.com/wp-content/uploads/2019/06/bct-000_icon-scaled-1900.jpg and if you eventually decide to purchase one, they can tell you exactly what you need.

Comments or questions? Contact us here.

Recommended Posts
Contact Us

We're not around right now. But you can send us an email and we'll get back to you, asap.