Summary
CalTennis is a large-scale video benchmark for evaluating monocular-to-3D human pose estimation in the wild. It contains over 11 million frames (51 hours) of tennis practice and match footage from 40 players, captured at 60Hz with 2-6 synchronized cameras. The dataset is 10x larger than existing in-the-wild human motion video datasets and 3x larger than existing MOCAP ground-truth datasets, and is the first large-scale benchmark to provide synchronized multi-view recordings of expert athletic motion. The multi-view setup enables low-cost, label-free evaluation of monocular 3D pose estimation algorithms. The authors describe a simple, standardized data collection protocol requiring no specialized equipment or expertise, along with fully automated video calibration and synchronization. Benchmarking state-of-the-art monocular-to-3D pose methods on CalTennis shows that while 3D joint angle recovery is now reasonably accurate, all models struggle to consistently estimate depth and foot contact. Two new performance metrics—footwork and stability—are proposed, alongside qualitative analysis of body-shape inconsistencies, revealing previously unexplored failure modes and concrete improvement opportunities for pose estimation and motion analysis.
Paper Overview
Field: Computer Vision
Authors: Ilona Demler, Xinran Xie, Blake Werner
Published: 2026-06-20
arXiv: 2506.16207
Summary
CalTennis is a large-scale video benchmark designed for evaluating monocular-to-3D human pose estimation in real-world (in-the-wild) conditions.
Dataset Highlights
- Over 11 million frames (51 hours) of tennis practice and matches from 40 players
- Captured at 60Hz with 2-6 synchronized cameras
- 10x larger than existing in-the-wild human motion video datasets
- 3x larger than existing MOCAP ground-truth datasets
- First large-scale benchmark providing synchronized multi-view recordings of expert athletic motion
Key Contributions
1. Label-free evaluation: The multi-view setup enables low-cost, annotation-free evaluation of monocular-to-3D pose estimation algorithms.
2. Standardized collection protocol: A simple protocol that requires no specialized equipment or domain expertise, with fully automated video calibration and synchronization.
3. Benchmark findings: State-of-the-art monocular-to-3D pose methods achieve reasonably accurate 3D joint angle recovery, but all models struggle to consistently estimate depth and foot contact.
4. New metrics: Two novel performance measures—footwork and stability—plus a qualitative study of body-shape inconsistencies.
These metrics reveal previously unexplored failure modes and point to concrete opportunities for improvement in pose estimation and motion analysis.
---
*Auto-collected on 2026-06-21*
This page is an English static mirror generated for search and AI citation.
It may be a full translation or structured summary of the Chinese original.
Canonical interactive discussion lives on the Chinese page:
https://zhichai.net/topic/177981609