Increase Supply Chain Traceability with Edge Learning-Based OCR Technology
Alphanumeric codes are used on almost everything we use or eat, whether it be consumer electronics, automobile parts, food and beverages, tobacco, or pharmaceuticals. These codes serve a variety of purposes, from serialization to identifying lot numbers and expiration dates. Serialization of high-value items such as consumer electronics items, automotive parts, tobacco, or pharmaceuticals helps prevent counterfeiting, which devalues customers’ perception of high-value brands. In the food and beverage as well as the pharmaceutical industries, governing bodies, such as the FDA require manufacturers to print lot numbers and expiration dates so any potential recalls can be tracked all the way to the store from which they were sold. Consumer cleaning products with hazardous chemicals must have Unique Formula Identifier (UFI) codes that provide the composition of hazardous components to assist poison control centers with mitigation.
As part of the production process, manufacturers must not only mark codes on their products; they must also verify each code prior to final packaging to ensure both that the specific code matches with the item, and that the code has been captured for record-keeping purposes. In normal high-volume operations, these codes are read electronically by cameras and processed by optical character recognition (OCR) software. In some cases, they are manually read and recorded by an operator.
Challenges to Implementing Automated Machine Vision-Based OCR Solutions
Automating OCR solutions into manufacturing environments can be challenging for many reasons. Characters may be worn, deformed, skewed, poorly etched, or have irregular spacing. Codes are marked using different methods – such as laser etching, dot matrix printing, or stamping, which can make it hard for OCR programs to decipher the text. Another reason that codes are often hard to read is due to the background material on which they are applied. In some cases, the texture, color, or reflectivity makes deciphering the codes difficult. Examples of this are shiny metal backgrounds of machined parts or food cans, as well as sustainable packaging material that is fibrous or presents a mottled (noisy) background. Lastly, most OCR-based solutions are “programmed” by training the system with dozens or hundreds of different examples, which takes time and may require specific machine vision knowledge. In situations, where organizations don’t have the internal expertise or the time to spend training the system, or they have frequent product changeovers, this presents implementation challenges.
Unreadable OCR Codes Translate to Inefficiencies
When these methods don’t work as expected, there are impacts to the production process: items that aren’t read correctly (or at all) must be flagged and removed for manual reconciliation. This costs additional time and money which reduces overall equipment effectiveness (OEE). If for any reason a product or a lot of products slip through the system with incorrect or unreadable codes on them, then potentially recalling that item becomes extremely challenging.
How to Select the Right OCR Solution
When it comes to selecting the appropriate technology to solve these challenges, manufacturers should look for solutions that are quick to set up and easy to train, while delivering robust readability. OCR tools must be able to read and decipher alphanumeric codes that are marked on reflective, low-contrast, and/or non-flat surfaces. To ensure fast and simple deployment, especially during frequent product changeovers, it’s important to consider an OCR solution that can quickly be retrained with only a few images.
Keeping in mind that a major reason that manufactures must put these codes on their products is due to government regulatory compliance, it’s important for every manufacturer, regardless of their size and internal resources, to be able to deploy solutions that help them comply with traceability regulations. Mid-size and smaller manufacturers often don’t have the internal resources to set up and deploy complex OCR systems. For this reason, choosing a system that sets up in minutes via an intuitive interface is key to successful deployment.
Cognex ViDi EL Read Tool Simplifies OCR Reading
Cognex 2D vision systems with edge learning-based OCR accurately read alphanumeric codes on fast moving production lines. The embedded ViDi EL Read tool deciphers a variety of text and font types using advanced optical character recognition (OCR). It reads multi-line text, as well as characters against challenging backgrounds, including reflective, low-contrast, and non-flat surfaces. The tool can also be quickly retrained to handle new text and accommodate process variations. With minimal training required, the tool simplifies job setup and delivers fast, accurate character reading. Here are several application examples where Cognex vision systems with ViDi EL Read capabilities solve the common challenges of reading alphanumeric codes for traceability:
Automotive/EV – Part Identification
Machined automotive parts often have embossed, human-readable date and lot information for easy identification. These codes can be hard to read due to low contrast, distortion, or high reflectivity. All these issues lead to inaccuracies when this text is read by conventional machine vision OCR technology.
Consumer Packaged Goods-Diaper Wipes Pouch Lot Code Verification
Personal care products, such as diaper wipe pouches, are marked with an alphanumeric lot code for traceability and verification to facilitate potential product recalls. These codes can be hard to read due to low contrast print, reflective backgrounds or because characters can become deformed or skewed by inadvertent folds in the packaging.
Electronics-USB Serial Number Verification
Consumer electronics components, such as USB connectors are serialized to provide supply chain traceability and deter counterfeiting. These codes are often hard to read due to reflectivity or the presence of crooked or deformed characters.
Food and Beverage-Soup Can Code Reading
Food and beverage containers, such as soup cans, use alphanumeric codes to display lot codes and expiration dates. These codes, which are often printed using a dot-matrix ink jet method, can become degraded because of the manufacturing process, or are challenging to read due to uneven surfaces.
Pharmaceutical-Vaccine Label Inspection
Pharmaceuticals, including vaccines, must display the lot code and expiry date on both primary and secondary packages to provide supply chain traceability and easy management of potential recalls. These codes can be distorted, faded, or deformed due to poor printing, or become hard to read due to external markings occluding part of the code.
Improve OEE and Traceability: Incorporate Edge Learning OCR Into Your Production Environment
Cognex machine vision systems equipped with edge learning-based OCR tools improve supply chain traceability to facilitate management of potential recalls, help prevent counterfeiting, and ensure customer safety. This highly accurate solution is easy to use, sets up quickly, and easily integrates into production lines. Because only a few images are needed to train the system, they increase the efficiency of product changeovers, which helps OEE.