Solve complex vision inspections with deep learning for manufacturing and logistics

Deep learning simulates the human brain – replicating the ‘human’ learning process, but inside a machine. Thanks to recent advances in vision intelligence, it is now possible to teach vision systems to carry out complex inspections and monitoring in situations where you would previously have needed the experience and knowledge of a human. Once integrated, it can be operated without needing extensive AI or programming knowledge.

Two big recent technical developments have made this possible:

• The use of Graphics Processing Units (GPU) means complex deep learning inspections now takes seconds, rather than hours
• More accessible software means anyone can now use deep learning inspection cameras and programming tools – regardless of your computing or coding experience.

Deep learning is ideal for complex and highly customised industrial applications and inspections in processes where there is significant variation, or where the tolerance levels for acceptable anomalies is very low. It uses hundreds or even thousands of images to train the system, to automate sophisticated tasks. For more routine industrial automation applications, edge vision may be more suited.

Automating defect detection has long been a challenge. Deep learning can now identify complex product defects quickly and reliably.

Check for surface damage, errors or abnormalities – such as inspection of paintwork, textiles, printed text or car body parts

For production lines handling multiple product types, variations or tasks, deep learning can easily handle complex inspections.

Detect damage on car body panels, spot upside-down products or catch misshaped foods on fast-moving conveyors

Deep learning inspection & application is best set up & trained by vision intelligence experts, ready for factory engineers to work with.

The future is ‘lights out’ automation – adopt deep learning to find efficiencies and remain competitive long-term.