Product label identification with OCR for model‐specific reuse, repair, and recycling: A case study for washing machines

Abstract

AbstractIn this study, a procedure to automatically identify the product model by evaluating an image of the label is presented. First, object character recognition (OCR) extracts text from the product label. Second, to identify the product model, the extracted text is compared with unique model references in a database, giving access to other model‐specific information. For this comparison, a novel variation of the partial ratio matching algorithm was developed. The product‐model identification procedure is integrated into an interactive web application, which allows for model identification to be performed during preparation for reuse, repair, and recycling: The SmartRe application. Three datasets consisting of 466, 422, and 771 images of washing machine product labels were gathered in collaboration with a (1) professional repair company for consumer devices, (2) a nonprofit repair and reuse center that resells devices in second‐hand stores, and (3) a large recycling company. Results demonstrate that 96%, 91%, and 40% of the product models were correctly read from the product label with OCR, respectively. Of these recognized models, 51%, 88%, and 76% were successfully identified with the SmartRe application by comparing the extracted text with the model database. Further analysis also demonstrated that 72% of the washing machine models identified at the nonprofit repair and reuse center were also found at the recycling facility and that 12% of these models are predicted to be less than 10‐years‐old. This highlights the potential of the SmartRe application to assist in product triage for reuse at recycling centers.