Aquatic Motion Dataset for 3D Pose Measurement via Polarimetric Imaging

An aquatic motion dataset contains approximately 450,000 frames of video across four sport categories. Shiwei Chu published the dataset on figshare in May 2026. It supports a multi-view polarimetric measurement framework for 3D kinematic analysis.

Use Cases

• Evaluate 3D pose estimation algorithms based on polarimetric imaging and biomechanical constraints.
• Benchmark highlight suppression techniques for dynamic water surfaces based on Fresnel-Mueller decoupling.
• Train dual-constraint regression networks for joint coordinate estimation based on embedded physical models.
• Analyze inter-frame trajectory stability in aquatic sports based on cross-view Stokes consistency.
• Develop real-time motion capture systems for water sports based on GPU-accelerated processing.

Strengths

• Dataset includes approximately 450,000 frames across four sport categories.
• The described method achieved a mean per-joint position error of 64.7 mm and a PCK@100 accuracy of 75.8%.
• Real-time throughput of 108 fps was attained on a single GPU.
• The dataset is accompanied by multiple file formats including code, configuration files, and data.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The dataset size is 15.5 MB, suggesting a relatively small scale of data files.

Provenance

Source

figshare

Collection Method

Likely gathered via a six-camera synchronous acquisition platform capturing aquatic sports.

Freshness

Last updated 2026-05-08 10:37:32; freshness should be verified.