This report presents a module for athlete recognition in broadcast videos as part of a video entertainment application. The aim is to identify athletes by means of name or number as printed on their uniform and extract the video clips where the desired athlete is visible. Using an adaptation of a previously published algorithm we extract text from each single frame and we read it by means of an optical character recognizer (OCR). The OCRed text is then compared with an a-priori known list of athletes' names or numbers to provide a score of the presence of each athlete in that frame. Presence plausibility is then checked by a temporal persistence analysis of names/numbers extracted from subsequent frames. Performed experiments show that the method is promising and can provide a useful input for a module for the identification of the athlete presence in a frame.
Segmentation of sport videos by embedded text reading
Modena, Carla Maria
2010-01-01
Abstract
This report presents a module for athlete recognition in broadcast videos as part of a video entertainment application. The aim is to identify athletes by means of name or number as printed on their uniform and extract the video clips where the desired athlete is visible. Using an adaptation of a previously published algorithm we extract text from each single frame and we read it by means of an optical character recognizer (OCR). The OCRed text is then compared with an a-priori known list of athletes' names or numbers to provide a score of the presence of each athlete in that frame. Presence plausibility is then checked by a temporal persistence analysis of names/numbers extracted from subsequent frames. Performed experiments show that the method is promising and can provide a useful input for a module for the identification of the athlete presence in a frame.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
