A visualization framework for team sports captured using multiple static cameras

Abstract

Abstract-We present a novel approach for robust localization of multiple people observed using a set of static cameras. We use this location information to generate a visualization of the virtual offside line in soccer games. To compute the position of the offside line, we need to localize players positions, and identify their team roles. We solve the problem of fusing corresponding players positional information by finding minimum weight K-length cycles in a complete K-partite graph. Each partite of the graph corresponds to one of the K cameras, whereas each node of a partite encodes the position and appearance of a player observed from a particular camera. To find the minimum weight cycles in this graph, we use a dynamic programming based approach that varies over a continuum from maximally to minimally greedy in terms of the number of graph-paths explored at each iteration. We present proofs for the efficiency and performance bounds of our algorithms. Finally, we demonstrate the robustness of our framework by testing it on 82,000 frames of soccer footage captured over eight different illumination conditions, play types, and team attire. Our framework runs in near-real time, and processes video from 3 full HD cameras in about 0.4 seconds for each set of corresponding 3 frames.

AFL Player Detection and Tracking

“…Hamid et al [14], [15] proposed an approach for robust localisation of soccer players using a set of cameras viewing the field from different angles. They set up a complete Kpartite graph, with each partition corresponding to one of the K cameras.…”

High performance moves recognition and sequence segmentation based on key poses filtering

“…In general, moves recognition in sports context [1,10,27,18,17,7] has been tackled with complex setup of cameras. El-Sallam et al [7] presented a 24 opto-sensitive cameras setup capturing 50 frames and a markerless system to be used to optimize the athletes techniques during training sessions.…”

Spotlight: Hot target discovery and localization with crowdsourced photos

“…Combining object images across multiple views is a very important step for object tracking or localization. Most existing works formalize this as a path search problem [16,17] . Those methods work when there are only small numbers of cameras and objects, but they are not feasible as solutions to our problem because they are too aggressive in combining similar object images and will give a high rate of false positive errors.…”

“…Researchers have also considered a fusion of multiple features, including position, speed, shape, and chromatic characteristics [25] . Hamid et al [17] used static cameras at fixed positions for football player localization; they used a complete K-partite graph and solved the combination problem by finding minimum weight K-length cycles. Hot target discovery.…”