OCR on Pork Legs

Abstract

Optical Character Recognition is still a hard task when performed in real scenarios, since images often show characters from non-frontal poses and sometimes prohibitive lighting conditions. In this context, one of the top companies in Europe for the production and sale of fresh chicken, turkey, pork and rabbit, is interested in automatically tracing its ham production line. For this purpose they need the numbers stamped on the hind leg of the swine to be optically recognized by modern image processing techniques. The first picture (the one with the largest view on the pork leg) can be taken as reference for the processing, while the others are helpful to give context to the production line. In case of follow up, more pictures could be provided by the company.

Since THIS PROJECT HAS BEEN PROPOSED BY A COMPANY, in case of encouraging results, the project can evolve as AN EXTERNAL THESIS. Location: Quinto di Valpantena (VR), duration: 3-6 months. In case of interest please contact the author of the proposal.