Classification

The classification feature uses deep learning to extract a pre-trained class of an image or a product for quality control. The user can add as many classes as is needed and the classes can be any category which is inferable from an image. A class can relate to quality, size, color and/or shape. Simply gather the examples for the classes and train a model.

  • Use cases:
      • Classify products into different quality classes such as GOOD, OK or BAD
      • Classify production by error types such as SCRATCH, MISSING PART or OK
      • Identify different types of products on a production line
      • Increase performance over time with new training examples
      • Add new classes or products later as needed
  • Requirements:
      • Tens or hundreds of examples for each class
      • The difference between classes should be visible to the human eye