A workshop on “AI in image analysis” was held at the DITF in the “Digitalization” area. AI has already been used to solve image analysis problems in many research projects. For example, in the ongoing “TexScan” project, in which a wide range of data is being collected on unknown fabrics in order to optimize automatic process control in textile finishing. This data also includes extensive image analysis information captured by a textile scanner.
Another project is dedicated to the quality control of textile knitwear on knitting machines. What is new here is that the use of image analysis means that irregularities in the knitted fabric can be detected during the manufacturing process and not, as was previously the case, in a subsequent quality check. This enables the process to be readjusted at an early stage. Other topics at the workshop included the classification of shoe shapes using image analysis, the quantification of shear distortion in glass fiber fabrics and the problems arising from the image analysis of microscopic images in fiber and textile testing. What they all have in common is that in order to improve product and manufacturing processes using AI, large amounts of data must be collected and linked together. There is a consensus among the workshop participants that material development will increasingly take place virtually in the future and that the laboratory component of development will decline.
In order to drive forward the expansion of AI in as many DITF research fields as possible, it is necessary to provide data on the manufacture of textile products and their preliminary stages as well as measurement data from the laboratories. Clearly described processes and measurement parameters in the research fields make it possible to create a broadly usable and correlatable database. Training AI models requires large amounts of data, which are traditionally collected in DITF research projects. The DITF are thus laying the foundation for AI-based material and product development and process optimization in the future.