|
Title:
|
DIVE: A DATA-DRIVEN INTELLIGENT VIDEO EVALUATION SYSTEM FOR ENHANCING SWIMMING PERFORMANCE |
|
Author(s):
|
Vanessa Camilleri, Justine Scicluna, Gianluca Aquilina, Dylan Seychell and Emma Swanwick |
|
ISBN:
|
978-989-8704-71-9 |
|
Editors:
|
Paula Miranda and Pedro Isaías |
|
Year:
|
2025 |
|
Edition:
|
Single |
|
Keywords:
|
Swimming, Pose Estimation, Biomechanics, Computer Vision, Sports Performance, Underwater Video Analysis |
|
Type:
|
Full Paper |
|
First Page:
|
106 |
|
Last Page:
|
113 |
|
Language:
|
English |
|
Cover:
|
|
|
Full Contents:
|
if you are a member please login 
|
|
Paper Abstract:
|
This paper presents DIVE (Data-driven Intelligent Video Evaluation), a lightweight system designed to analyse underwater
swimming videos using pose estimation and domain-informed feature extraction. The system aims to support coaches and
athletes by providing interpretable biomechanical metrics without the need for expensive motion-capture infrastructure.
The video capture setup uses off-the-shelf GoPro cameras positioned poolside to ensure stable and consistent footage. Pose
estimation models were evaluated on collected video samples, and features such as stroke rate, symmetry, alignment, and
turn performance were extracted based on joint coordinates.
The system includes a modular architecture with pipelines for video pre-processing, pose estimation, feature computation,
and future feedback presentation. Domain experts contributed to the design of task-specific features to ensure relevance
and interpretability. Preliminary analysis suggests that model choice and camera angle substantially impact pose accuracy.
While still under development, DIVE lays the foundation for scalable, context-aware performance feedback in swimming.
Future work will involve completing quantitative validation using annotated ground truth, improving phase detection
accuracy, and designing user-facing tools for real-time visual feedback. |
|
|
|
|
|
|
