Marathon text bib recognition
Marathon text bib recognition

We implemented an automatic system for Marathon Text Bib Recognition (MTBR) in natural image collections covering running races. It is typically a piece of durable paper or cardboard bearing a number as well as the event/sponsor logo. The MTBR, usually pinned onto the competitor’s shirt, is used to identify the competitor among many others during the race. Today, MTBR identification is often done manually, a process made difficult by the sheer number of available photos. This specific application can be studied in the wider context of detecting and recognizing text in natural images of unstructured scenes. Existing methods that fall into this category fail to reliably recognize MTBR, due to the large variability in their appearance, size, and the deformations they undergo. By using the knowledge that the MTBR is located on a person’s body, our dedicated system overcomes these challenges and can be applied without any adjustments to images of various running races taken by professional as well as amateur. Our system receives a set of natural images taken in running sports events. We use Artificial Intelligence trained models to identify the BIB location and scale. This is then processed and fed to a standard optical character recognition (OCR) engine to recognize the text out of the object. We evaluated the contributions of each component of our system.