Artificial intelligence in application. Machine Learning Vision

Machine Learning Vision
The Machine Learning Vision package extends the CP Camera Inspection application module with machine learning functions and learning options. It thus offers a practically oriented introduction to the highly topical subject of machine learning (ML). The application scenario here is optical quality inspection, as used in numerous production companies. Be it for component recognition, for distinguishing good parts from bad parts, for identifying anomalies or generally for determining the quality of a manufactured component.

The package contains the didactic machine learning software, a powerful PC and the necessary accessories and training documentation. It complements the CP Camera Inspection application module, which is already perfectly prepared with the industrial camera module and the integrated variable-position illumination. As a result, the solution offers challenging test and experiment situations with regard to camera angle, field of view and illumination, to which corresponding machine learning algorithms must be adapted.

The included software guides the user through the typical processing steps, starting with the acquisition, storage and pre-processing of the relevant data, the selection and training of a suitable ML procedure and ending with its use in the actual application. The training documentation addresses various tasks of increasing complexity from different areas of optical quality assurance on the basis of machine learning.

The software offers a wide range of customization options, in which case all experiments can be carried out even without programming and ML knowledge. The user interface is browser-based and can thus be operated from various devices in the local network. Communication with a Siemens PLC – provided that the CP Camera Inspection application module is integrated in a CP Lab/Factory system application module takes place via OPC UA interface.

Scope of delivery

  • High-performance computer
  • Workpiece carrier including pallet
  • Front and back shells in black, blue, red, grey
  • 4 PCBs
  • 1 set of fuses
  • Training documentation
  • Pre-trained neural networks for all experiments

Additionally required components

  • USB keyboard and mouse*
  • Monitor with HDMI or DisplayPort connection*

* Not necessary when accessing via a remote connection.

Training content

  • Machine learning and artificial intelligence
  • Basics of supervised, unsupervised, reinforcement learning
  • Introduction to neural networks
  • Introduction to Python and Google Tensorflow
  • Conduct, analysis, optimization of machine learning procedures with the help of the algorithms and tools included in the package, using optical quality inspection as an example
  • Transferring the knowledge gained to other ML application areas