Introduction

Player tracking in soccer video has received widespread attention in the field of sports video analysis. We present a high-quality Benchmark for Soccer Play Tracking (BSPT).

The overview of BSPT are as follows:

(1) A high-quality soccer player tracking dataset, which contains a total of 120 sequences and 35565 frames, is constructed from four extremely challenging.

(2) Rich annotations are provided, include 2 types of bounding boxes, 8 different frame-level attributes, and 2 different invisible labels for the target player.

(3) In order to provide extensive baselines for future research, we conducted comparative experi- ments on 11 different algorithms and detailed analysis using several evaluation metrics.

 

Datasets

You can download all of the BSPT at following page.

- Download BSPT dataset

- Download Player Label Tool

 

Evaluation

Comparative experiments have been conducted on 11 different algorithms and evaluated for eight attributes, the experimental results on all frames of the BSPT are reported in terms of precision and success. The programs of trackers are obtained from open-source code provided by authors.

The analysis of the evaluation and the original results of the experiment have been published in:

- Evaluation

 

Laboratory of Intelligent Media Computing and Network Security,HUST. ©  2021. All rights reserved.