Image analysis of textile fabrics with anomaly detection applying AI - job 2 of 6

We have purchased and installed a textile fabrics inspection machine which is supposed to detect weaving anomalies. However, it turns out that the detection is not specific enough. The software relies solely on contrast strengths of certain irregularities in the fabric which does not necessarily have to be a defect but just an uneven surface. If the contrast strength is adjusted and made weaker... the real defects (such as pin holes or broken fibres) are not detected any longer. This results in an unreliable equipment which makes it inconvenient to apply. So, we are looking for a custom software solution based on machine learning which can do image analysis. An operator feeds the system with images of the fabrics which have anomalies. He also indicates if it is a defect and what type of defect. Over time the system shall learn and shall become independent