New CalTennis dataset advances 3D human pose estimation research

Researchers have introduced CalTennis, a large-scale multi-view video dataset designed for evaluating monocular-to-3D human pose estimation. The dataset features over 11 million frames from 40 tennis players, captured with synchronized multi-camera setups. This benchmark aims to provide a label-free evaluation method for pose estimation algorithms, particularly for athletic motions. Initial benchmarking on CalTennis revealed that while 3D joint angles are accurately recovered, depth and foot contact estimation remain challenging, highlighting areas for future research in pose estimation and action analysis. AI

IMPACT This dataset could accelerate progress in 3D human pose estimation, impacting fields like sports analytics and virtual reality.

RANK_REASON The cluster contains a research paper detailing a new dataset and benchmark for a specific AI task. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• arXiv
• CalTennis
• Monocular-to-3D Pose Estimation

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →