Mixing Vodka and AI: Brand Management and Visual Inspection

Visual inspection is the oldest method of quality control. Humans excel at detecting cracks, warping, subtle flaws and missing parts. Depending on the product, you can rely on taste and smell to spot the differences. We adapt to unpredictability, are easily trained and can learn quickly by example.

However, when we are tired, distracted, or bored, we make mistakes. Studies have shown that manual inspection tasks can have error rates of up to 30%. Often these “mistakes” are actually false positives where an inspector has begun to question their decision making.

Given these error rates, resulting product quality issues, and higher costs due to waste, additional screening, and manufacturing downtime, there is a growing demand for technologies based on the AI to provide decision support for manual tasks. In particular, these technologies are well suited for low-volume, higher-value, and custom products where fully automated inspection is neither cost-effective nor practical.

AI Brand Management for Food and Beverage

Dairy Distillery, a Canadian spirits manufacturer that pioneered a unique process to produce vodka from a dairy by-product, is using AI visual inspection to add decision support for the manual labeling and quality control (QC).

Brand appearance plays an important role in consumer choice, and the manufacturer competes against larger players with much larger marketing budgets. Additionally, operating in the food and beverage market carries another set of risks. About 60% of businesses in the market experience a recall. While food quality-related recalls grab headlines and can significantly damage a brand’s reputation, typically one-third of US Food and Safety Inspection recalls are related to manufacturing errors. packaging and labeling. These mislabeled or mislabeled products may not impact consumer safety, but they can lead to costly shipping delays and rework for a manufacturer.

For the distillery, their primary concern is maintaining a cohesive and eye-catching appearance so they can earn a prominent place on a store shelf. The distillery uses a bottle fashioned after an old-fashioned milk bottle, with distinctive, eye-catching labeling. The main label and a cap label are applied by machine. A human must accurately place an emblem logo that visually aligns with brand elements on other labels to ensure consistent and attractive shelf display. With multiple products and short production runs, it is not cost effective for the distillery to fully automate their labeling process.

During a long shift, the placement of the emblem would begin to change as the operator grew tired. Errors often went unnoticed until the final packaging stage, when staff were then tasked with manually removing and replacing labels. This resulted in downtime, production delays and additional costs. Worse still, there was always the risk that a mislabeled product could reach store shelves.

AI-powered visual inspection provides operator decision support to help eliminate labeling errors. The system includes camera, edge processing, display panel and pre-packaged AI plug-ins from common inspection requirements. Pre-Package Inspection Skills are easily trained to inspect labeling and verify products through various manufacturing stages, and can also be customized for specific requirements.

Pleoraimage2

Small-footprint visual inspection systems integrate into existing manufacturing operations and processes, with embedded AI algorithms that are easily trained to a customer’s unique requirements.

As a fully integrated tabletop system, the operator uses the camera and edge processing to acquire images and create an AI model. They don’t need to switch between cameras, PCs, or cloud-based systems; making the system easy to use, very cost effective and all in a small footprint.

Without the need for programming skills, the distillery’s quality manager trained the image comparison plugin to add decision support for his labeling process. With a single image of a known good product – a “golden reference” – the system automatically identifies key brand elements on the bottle. In this example, the plugin identifies the position of the main tag and the cap tag, and adds a graphic overlay on the visual display to guide the correct placement of the inset tag for the operator.

AI-powered visual inspection ensures brand consistency and accuracy for the distillery, as well as cost savings because labeling doesn’t have to be removed and replaced due to human error. The technology is also being used by the manufacturer as a training tool for new operators, so they can quickly understand the proper placement of brand elements on the bottle and the difference between “good and bad” products.

As production expands, the quality manager or operator can easily update the visual inspection system with additional “golden references” to provide labeling advice for new bottles, labeling and packaging. The operator simply chooses the correct plug-in for the product to be inspected. An additional image backup plugin could also be used to capture images of products at different stages of production for batch tracking. This will also provide the manufacturer with key data related to their manual assembly and inspection processes for root cause analysis and productivity management.

As the distillery adds more automation to its production, the visual inspection system can provide valuable quality control for in-process or finished products to ensure all machines and humans are operating smoothly. synchronized. For this application, the distillery is developing a custom plug-in that provides a quick pass/fail assessment on label placement for all bottles. This helps eliminate stressful subjective decision making for operators and will increase production as errors can be identified long before final packaging. Later, the plug-in could be further trained to assess things like fill levels, cap seal integrity, and ensuring the correct labels are used for custom or regional products.

Pleora3

Pleora’s visual inspection system is easily trained to help guide manual placement of bottle labeling, with an additional AI plugin used for in-process and finished product quality checks.

Modification of the visual inspection

Adding decision support and automating visual inspection processes help ensure consistency. The technology helps a tired operator at the end of a shift, a new employee who doesn’t know what makes a product “good” or “bad,” or simply speeds up the inspection process. It enables manufacturers to take advantage of new technologies for processes where automation is prohibitively expensive, including the production of lower volume, higher value goods.

Written by Ed Goffin, Marketing Manager at Pleora Technologies.

Cathy W. Howerton