Offensive Tactics Recognition in Broadcast Basketball Videos Based on 2D Camera View Player Heatmaps.

Nico,Tse-Yu Pan,Herman Prawiro,Jian-Wei Peng,Wen-Cheng Chen,Hung-Kuo Chu,Min-Chun Hu

DOI: https://doi.org/10.1145/3591106.3592285

2023-01-01

Abstract:It is essential for sports teams to review their offensive and defensive tactical execution performance as well as understand their opponents’ tactics in order to identify effective counterattack strategies. This study focuses on basketball offensive tactics recognition based on 2D camera view heatmaps. Most of the current tactics recognition methods learn the spatiotemporal correlation of players based on top-view trajectory information. To obtain correct top-view player trajectories, robust camera calibration and player tracking techniques are indispensable. However, for broadcast videos having large camera movement, serious player occlusions, and similar players’ jerseys, it is quite challenging to obtain accurate camera parameters and player tracking results, resulting in poor tactical analysis performance. Instead of applying camera calibration and player tracking, this study attempts to design a tactics recognition method that directly predicts the tactics class from 2D camera-view player heatmaps in the inference phase. Our proposed method uses a recurrent convolutional neural network with coordinate embedding to directly identify the tactics. Moreover, an auxiliary top-view player trajectory reconstruction module is added in the training phase to acquire better latent codes to represent the tactics. The experimental results show that for both supervised and unsupervised settings, our proposed method achieves comparable accuracy to the current tactics classification methods that rely on perfect top-view trajectory input.