2024: Validating of novel neural network-based fabric production process monitoring solution

“Reducing material losses by detecting interruptions in a fast-moving fabric knitting line” aimed to validate a quality control support mechanism that detects errors in mop knitting in real time and helps reduce the loss of waste material in the factory. The demo project proved that an effective quality control solution can be created using the Ultralytics Yolo v8 algorithm. The project started with software and hardware research. In parallel with the camera setup and detection software validation, the hardware was modelled and 3D printed for installation. The solution has been set up in the factory and fitted to the knitting machine, where it detects the defects that occur during knitting. Although it is not possible to accurately assess the material losses, since the solution has not been in use in the company for a long time, it can already be said that the solution is very promising and when detecting a faulty mop, it displays the location and accuracy of the fault on the screen, and gives a sound signal and a light signal when the fault has occurred. The end result is a solution that can effectively improve the quality control process. Longer analysis and further developments of the solution are planned within the company, and it would be possible to adapt similar systems to other production lines.