A Methodology for Real-Time Data Verification exploiting Deep Learning and Model Checking

摘要: This paper presents a methodology for real-time data extraction and verification. In particular, considering the lacking of in sport analytics context, we propose method to generate data-set player positions from soccer game videos, deep learning techniques, order extract position, terms x-axis y-axis, related accuracy detection. The experiment is performed on several local games, captured using stationary camera situated tribune area close center field. positioned cover entire Clearly, this can allow us other types information, such as distance between players ball, covered by all specific situation game. All informations are stored CSV (comma separated value) file that be used verify behavioural properties exploiting formal methods.