CalTennis: Large Multi-View Tennis Video Dataset and Benchmark of Monocular-to-3D Pose Estimation

Two new datasets, CalTennis and Bengal-HP_RU, have been introduced to advance pose estimation research. CalTennis comprises 11 million frames of tennis footage, while Bengal-HP_RU features 12,894 labelled head images of Bengali subjects.

The CalTennis dataset, described in a paper on arXiv^[1], contains 51 hours of tennis practice and match play from 40 players, captured with 2-6 synchronized cameras at 60 Hz. It is 10 times larger than existing in-the-wild human motion video datasets and 3 times larger than existing MOCAP-ground-truthed datasets^[1]. The multi-view setup enables inexpensive, label-free evaluation of monocular-to-3D pose estimation algorithms. Meanwhile, the Bengal-HP_RU dataset, also hosted on arXiv^[2], is the first publicly available head pose dataset centred on Bengali subjects, with 12,894 labelled head images annotated with continuous yaw, pitch, and roll values. Images were collected from Wikimedia Commons under free licences^[2]. The dataset is partitioned by Wikimedia uploader identity to prevent data contamination and exhibits substantial diversity in subject age, gender, occlusion, illumination, and background.