Towards Case-Based Morphological Classification for Fashion Product Development

This paper describes a novel approach for the morphological classification based on body scan data. This is an important topic for the fashion industry because traditional sizing systems do not provide perfectly fitting clothes. That is due to the diversity of human bodies and the limited number of primary dimensions in sizing. The idea is to have experts classifying a relevant number of representative body scans and to use this as a case and knowledge base for further automatic classification. This Case-Based Reasoning (CBR) approach is suitable especially for classification problems where many variables are involved and no clear rules can be identified. A new scan is compared with the already classified scans in the case base. The most similar ones are retrieved and their classification adapted to classify the new scan. A sophisticated similarity model ensures that only useful, relevant cases are retrieved from the case base. The system has been tested, including statistical analysis and field tests. An industry project confirmed the feasibility of the approach and now helps the fashion company to adjust their product development to the real needs of the customers. The approach will be used in body measurement portals to optimize size tables and thus the fitting of the clothes as well as in e-shops to recommend the morphologically best garments to a customer.