PadelTracker100: A Dataset for Intelligent Player and Ball Tracking in Padel Sports
Creators
-
Bada-Nerín, Roberto
(Researcher)1
-
Paula Rodríguez
(Researcher)1
-
Kreibel, Denis
(Supervisor)1
-
Teodoro Rodríguez, Joao Víctor
(Researcher)1
-
Pedrayes, Oscar D.
(Supervisor)1
-
Usamentiaga, Ruben
(Supervisor)2
-
Fontenla-Seco, Yago
(Supervisor)1
-
Ben Lestón, Pablo
(Annotator)1
-
Rodríguez Trillo, Sebastián
(Annotator)1
-
Lozano García, Nicolás
(Annotator)1
-
Mosquera Bonasorte, Franco
(Researcher)
-
1.
NTT (Japan)
-
2.
Universidad de Oviedo
Description
This dataset represents the first large-scale, fully annotated dataset of padel, a rapidly expanding sport, featuring comprehensive tracking of ball trajectories, pose estimation, and recognition of shot events. The dataset comprises approximately 100,000 frames extracted from two matches at the 2022 World Padel Tour (WPT) Finals recorded in 1920x1080@30 videos. To enhance data quality, these frames were captured using a single camera angle that minimizes occlusions, accompanied by detailed annotations at the frame level, including ball positions, player poses, and shot event information.
The included Jupyter notebook demonstrates the practical application of the dataset.
Videos:
---
For any communication regarding this dataset, we provide an open contact email where users can send questions, feedback, or suggestions for improvement: PadelTracker100@gmail.com