Product image
Systems by Application

Machine Learning

The word “Machine learning” is used when a machine imitates "cognitive" functions, often associated with humans such as “learning” or “problem solving” this is the very basics of machine learning.

Machine learning is often implemented via a (convolutional) neural network and can be supervised or unsupervised. Supervised learning can provide solid results and is often considered the best approach, but it requires many annotated training images. Annotation is time consuming and it can be difficult to get images upfront in a machine vision application. 

JLI vision is now introducing "Hybrid Vision" which we consider to be a combination of: 

  • Machine learning 
  • Traditional machine vision 
  • Craftsmanship.

Structural applications such as measuring dimensions and depth can be performed by traditional machine vision. They are easy to test and potentially expand with functionality. Hybrid vision is the perfect solution for "aesthetic" applications. Aesthetic applications consists of ensuring that the finished product fulfills the end-users or consumers expectations. Typical defects could be scratches, holes or color deviations. As opposed to structural applications it can be difficult to define objective requirement specifications. In practice good and bad samples are collected and used as reference in manual inspection. Previously aesthetic applications were often left unsolved because of the shortcomings of traditional machine vision, but with hybrid vision it is in some cases possible.

Case study on Industrial Machine learning

Application example: Wood inspection

  • Objective: Detection of gnarls and resin pockets in glue boards
  • Solution: Glue Boards are scanned on a conveyor and processed using a combination of machine learning (determines if candidate is defective), traditional machine vision (selects candidates) and 3D (checks surface of glue boards). 
  • Benefit: Solves a time consuming manual inspection task inline in production and enables fully automatic repair of glue boards.
  • Result: Achievable accuracy +95% which is better and more consistent than manual inspection

Machine learning for glue boards

Red marking equals gnarl detected.
Green marking equals resin pocket detected.

Application example: Glass inspection

  • Objective: Detection of open- and closed end airlines in glass tubing
  • Solution: Using traditional machine vision all airline defects are detected. Defects are then in real-time processed by machine learning in order to determine whether the defect is open- or closed end. 
  • Benefit: Improves yield by reducing scrap.
  • Result: Achievable accuracy +95% 

Machine learning for tube inspection

Closed end airline (OK) on the left and open end airline (reject) on the right.

Application example: Steel inspection

  • Objective: Detection of freak defect on the surface of rails directly after production
  • Solution: Rails are scanned by passing through an inspection tunnel. Images are processed real time using a combination of machine learning, traditional. Machine learning network is taught unsupervised.
  • Benefit: Solves a time consuming manual inspection task inline in production.
  • Result: Achievable accuracy +90% which is better and more consistent than manual inspection

Machine learning for rail inspection

Freak defect on the surface of a rail in the hot end of production.