Detection of defects (splashes or/and gluing defects) on leather footbeds.
Classical imaging failed due to the highly varying surfaces and defect characteristics.
In this application, an automated OK/NG inspection is implemented at the end of a gluing line. In addition to several robot systems that handle and pack the footbeds, the heart of the automation, the vision control, is realised by using kim> t-series.
Defects on footbeds are detected and the footbeds are declared as OK/NG. The difficulty of this task is not only the varying surfaces and defect characteristics, but also the small size of the defects in comparison to the size of the footbed. The image is captured using a lighting setup adapted to the task and a high-resolution industrial camera. The artificial intelligence in kim> returns information about the defects (see picture). Subsequently, an evaluation of the defect image is performed using the integrated algorithm in kim> and the product is declared as OK/NG.
This task was solved with the help of a segmentation algorithm. With time sensitive applications we use kim> s-series.
Kim> t-series can be used if the custumor wants to use the quality control for additional products in the future. This tool includes the possibility to learn and train new use cases based on the existing application. This way the customer can expand the product portfolio in kim> independently.