From soccer video to ball possession statistics
作者：
Highlights：
• Ball possession statistics in soccer are estimated automatically by counting the number of valid passes of both the teams.
• The pass detection is modeled as a constrained flow network, where the ball and players represent the nodes of the network.
• Additional special purpose nodes representing split, merge, appear and disappear are incorporated in the network.
• The constraint is incorporated using a concept of a group of players within which the split and merge are allowed.
• The minimum cost path in the flow network provides the solution for valid pass events
摘要
•Ball possession statistics in soccer are estimated automatically by counting the number of valid passes of both the teams.•The pass detection is modeled as a constrained flow network, where the ball and players represent the nodes of the network.•Additional special purpose nodes representing split, merge, appear and disappear are incorporated in the network.•The constraint is incorporated using a concept of a group of players within which the split and merge are allowed.•The minimum cost path in the flow network provides the solution for valid pass events
论文关键词：Ball possession statistics,Cost-flow network,Group similarity,Pass event detection,Soccer
论文评审过程：Received 30 September 2020, Revised 14 September 2021, Accepted 18 September 2021, Available online 20 September 2021, Version of Record 1 October 2021.
论文官网地址：https://doi.org/10.1016/j.patcog.2021.108338
