Manufacturers across a range of sectors are facing challenges around hiring and retaining a qualified workforce, keeping pace with technology innovations, and delivering on heightened expectations around speed and accuracy.
For those in the automotive manufacturing industry, sustainability, digitalisation of operations and the supply chain, increased expectations around safety, and demand for personalisation are also front-of-mind priorities.
Zebra’s recent Automotive Ecosystem Vision Study says 73% of industry decision-makers think their business will be at a competitive disadvantage if they don’t embrace more digital technologies, with ‘developing software expertise’ cited as a top five investment priority. A recent McKinsey report agrees, saying recent developments in robotics, AI, and machine learning have put us on the cusp of a new automation age.
One key technology innovation rising in importance is deep learning Optical Character Recognition (OCR) software. According to Zebra’s automotive study, industrial machine vision is expected to see an 83% increase in usage between now and 2027 (24% to 44%), according to Original Equipment Manufacturer (OEM) decision-makers surveyed.
The combination of deep learning and OCR meets the need for speed, accuracy and reliable solutions for compliance, quality and presence checks across manufacturing industries. Operations leaders within the automotive, pharmaceutical, electronics and food and beverage industries stand to gain the most with deep learning OCR.
Smart manufacturing
Machine vision and deep learning OCR are enabling smart manufacturing which Gartner defines as orchestrating physical and digital processes within factories and across other supply chain functions. These transform the ways people, processes and technology operate to deliver the information needed to impact decision quality, efficiency, cost and agility. In other words, cultivating automation via deep learning and OCR helps to get the best from hardware, software, and people.
But getting OCR inspection right can be challenging. Stylised fonts, blurred, distorted or obscured characters, reflective surfaces and complex, non-uniform backgrounds can make it impossible to achieve stable results using traditional OCR techniques.
However, there are new tools on the market that offer industrial-quality deep learning OCR and come with ready-to-use neural networks pre-trained using thousands of different image samples. This newer range can deliver high levels of accuracy straight out of the box, even when dealing with very difficult cases.
In an automotive manufacturing setting, it means a deep learning OCR solution can accurately read printed, embossed, matte and metallic serial numbers stamped on batteries, tires, parts and accessories to ensure they correspond with the correct car model’s Vehicle Identification Number (VIN). These solutions can also handle a range of font styles and sizes, changing and ‘harsh’ lighting and manufacturing environments.
Deep learning OCR can also be utilised as part of a wider machine vision solution. For example, in an automotive manufacturing setting, a machine vision solution can be deployed for presence/absence, quality and compliance for connector pin inspection, conformal coating on printed circuit boards, adhesive inspection, wire harnessing, electric battery and polarity, and general assembly verification.
In these scenarios, machine vision solutions deployed using the same machine vision cameras and unifying software platform can inspect items at a much faster speed, flagging suspected defects and errors to an engineer to examine and decide whether it’s a fault or not and move on.
Feeding the review decision back into the neural network helps to keep the continual learning input active – to further develop and enhance the model. It optimises efficiency and removes an important yet tedious manual task from engineers.
The growing value of deep learning
Deep learning’s speed and accuracy enables it to greatly assist engineers, ensuring quality in manufacturing, controlling production costs and enhancing customer satisfaction. However, ease of use is equally valuable, and that’s where deep learning OCR software shines. It’s an easy application to implement and use and can be deployed in a few simple steps—all without the need for machine vision expertise.
Together, more accessible machine vision and deep learning OCR solutions are unlocking new possibilities for industrial imaging professionals and engineers to think and act more like data scientists. This development is needed and welcomed, in the face of ever-growing data velocity, volume and variety and higher levels of speed, safety and accuracy expected.
You can reach to me or the wider machine vision team for further discussion.
About the author
Rudolf Schambeck is the Senior Channel and Market Development Manager, Machine Vision, Germany with Zebra Technologies where he has worked since 2021. Prior to joining Zebra, he spent five years at machine vision company, Cognex in sales engineer, business development, and account manager roles with a focus on B2B and automotive. Prior to Cognex, Rudolf worked for Intercontec Produkt GmbH and Irlbacher Blickpunkt Glas GmbH in B2B and OEM sales engineer, account manager, and sales roles.
